Upload 67 files

config.py

@@ -0,0 +1,38 @@
+############离线VC参数
+inp_root=r"白鹭霜华长条"#对输入目录下所有音频进行转换，别放非音频文件
+opt_root=r"opt"#输出目录
+f0_up_key=0#升降调，整数，男转女12，女转男-12
+person=r"weights\洛天依v3.pt"#目前只有洛天依v3
+############硬件参数
+device = "cuda:0"#填写cuda:x或cpu，x指代第几张卡，只支持N卡加速
+is_half=True#9-10-20-30-40系显卡无脑True，不影响质量，>=20显卡开启有加速
+n_cpu=0#默认0用上所有线程，写数字限制CPU资源使用
+############下头别动
+import torch
+if(torch.cuda.is_available()==False):
+    print("没有发现支持的N卡，使用CPU进行推理")
+    device="cpu"
+    is_half=False
+if(device!="cpu"):
+    gpu_name=torch.cuda.get_device_name(int(device.split(":")[-1]))
+    if("16"in gpu_name):
+        print("16系显卡强制单精度")
+        is_half=False
+from multiprocessing import cpu_count
+if(n_cpu==0):n_cpu=cpu_count()
+if(is_half==True):
+    #6G显存配置
+    x_pad       =   3
+    x_query     =   10
+    x_center    =   60
+    x_max       =   65
+else:
+    #5G显存配置
+    x_pad       =   1
+    # x_query     =   6
+    # x_center    =   30
+    # x_max       =   32
+    #6G显存配置
+    x_query     =   6
+    x_center    =   38
+    x_max       =   41

go-web.bat

@@ -0,0 +1 @@
+runtime\python.exe infer-web.py

go.bat

@@ -0,0 +1 @@
+runtime\python.exe infer.py

hubert_base.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f54b40fd2802423a5643779c4861af1e9ee9c1564dc9d32f54f20b5ffba7db96
+size 189507909

infer-web.py

@@ -0,0 +1,193 @@
+import torch, pdb, os,traceback,sys,warnings,shutil
+now_dir=os.getcwd()
+sys.path.append(now_dir)
+tmp=os.path.join(now_dir,"TEMP")
+shutil.rmtree(tmp,ignore_errors=True)
+os.makedirs(tmp,exist_ok=True)
+os.environ["TEMP"]=tmp
+warnings.filterwarnings("ignore")
+torch.manual_seed(114514)
+from infer_pack.models import SynthesizerTrnMs256NSF as SynthesizerTrn256
+from scipy.io import wavfile
+from fairseq import checkpoint_utils
+import gradio as gr
+import librosa
+import logging
+from vc_infer_pipeline import VC
+import soundfile as sf
+from config import is_half,device,is_half
+from infer_uvr5 import _audio_pre_
+logging.getLogger('numba').setLevel(logging.WARNING)
+
+models, saved_cfg, task = checkpoint_utils.load_model_ensemble_and_task(["hubert_base.pt"],suffix="",)
+hubert_model = models[0]
+hubert_model = hubert_model.to(device)
+if(is_half):hubert_model = hubert_model.half()
+else:hubert_model = hubert_model.float()
+hubert_model.eval()
+
+
+weight_root="weights"
+weight_uvr5_root="uvr5_weights"
+names=[]
+for name in os.listdir(weight_root):names.append(name.replace(".pt",""))
+uvr5_names=[]
+for name in os.listdir(weight_uvr5_root):uvr5_names.append(name.replace(".pth",""))
+
+def get_vc(sid):
+    person = "%s/%s.pt" % (weight_root, sid)
+    cpt = torch.load(person, map_location="cpu")
+    dv = cpt["dv"]
+    tgt_sr = cpt["config"][-1]
+    net_g = SynthesizerTrn256(*cpt["config"], is_half=is_half)
+    net_g.load_state_dict(cpt["weight"], strict=True)
+    net_g.eval().to(device)
+    if (is_half):net_g = net_g.half()
+    else:net_g = net_g.float()
+    vc = VC(tgt_sr, device, is_half)
+    return dv,tgt_sr,net_g,vc
+
+def vc_single(sid,input_audio,f0_up_key,f0_file):
+    if input_audio is None:return "You need to upload an audio", None
+    f0_up_key = int(f0_up_key)
+    try:
+        if(type(input_audio)==str):
+            print("processing %s" % input_audio)
+            audio, sampling_rate = sf.read(input_audio)
+        else:
+            sampling_rate, audio = input_audio
+            audio = audio.astype("float32") / 32768
+        if(type(sid)==str):dv, tgt_sr, net_g, vc=get_vc(sid)
+        else:dv,tgt_sr,net_g,vc=sid
+        if len(audio.shape) > 1:
+            audio = librosa.to_mono(audio.transpose(1, 0))
+        if sampling_rate != 16000:
+            audio = librosa.resample(audio, orig_sr=sampling_rate, target_sr=16000)
+        times = [0, 0, 0]
+        audio_opt=vc.pipeline(hubert_model,net_g,dv,audio,times,f0_up_key,f0_file=f0_file)
+        print(times)
+        return "Success", (tgt_sr, audio_opt)
+    except:
+        info=traceback.format_exc()
+        print(info)
+        return info,(None,None)
+    finally:
+        print("clean_empty_cache")
+        del net_g,dv,vc
+        torch.cuda.empty_cache()
+
+def vc_multi(sid,dir_path,opt_root,paths,f0_up_key):
+    try:
+        dir_path=dir_path.strip(" ")#防止小白拷路径头尾带了空格
+        opt_root=opt_root.strip(" ")
+        os.makedirs(opt_root, exist_ok=True)
+        dv, tgt_sr, net_g, vc = get_vc(sid)
+        try:
+            if(dir_path!=""):paths=[os.path.join(dir_path,name)for name in os.listdir(dir_path)]
+            else:paths=[path.name for path in paths]
+        except:
+            traceback.print_exc()
+            paths = [path.name for path in paths]
+        infos=[]
+        for path in paths:
+            info,opt=vc_single([dv,tgt_sr,net_g,vc],path,f0_up_key,f0_file=None)
+            if(info=="Success"):
+                try:
+                    tgt_sr,audio_opt=opt
+                    wavfile.write("%s/%s" % (opt_root, os.path.basename(path)), tgt_sr, audio_opt)
+                except:
+                    info=traceback.format_exc()
+            infos.append("%s->%s"%(os.path.basename(path),info))
+        return "\n".join(infos)
+    except:
+        return traceback.format_exc()
+    finally:
+        print("clean_empty_cache")
+        del net_g,dv,vc
+        torch.cuda.empty_cache()
+
+def uvr(model_name,inp_root,save_root_vocal,save_root_ins):
+    infos = []
+    try:
+        inp_root = inp_root.strip(" ")# 防止小白拷路径头尾带了空格
+        save_root_vocal = save_root_vocal.strip(" ")
+        save_root_ins = save_root_ins.strip(" ")
+        pre_fun = _audio_pre_(model_path=os.path.join(weight_uvr5_root,model_name+".pth"), device=device, is_half=is_half)
+        for name in os.listdir(inp_root):
+            inp_path=os.path.join(inp_root,name)
+            try:
+                pre_fun._path_audio_(inp_path , save_root_ins,save_root_vocal)
+                infos.append("%s->Success"%(os.path.basename(inp_path)))
+            except:
+                infos.append("%s->%s" % (os.path.basename(inp_path),traceback.format_exc()))
+    except:
+        infos.append(traceback.format_exc())
+    finally:
+        try:
+            del pre_fun.model
+            del pre_fun
+        except:
+            traceback.print_exc()
+        print("clean_empty_cache")
+        torch.cuda.empty_cache()
+    return "\n".join(infos)
+
+with gr.Blocks() as app:
+    with gr.Tabs():
+        with gr.TabItem("推理"):
+            with gr.Group():
+                gr.Markdown(value="""
+                    使用软件者、传播软件导出的声音者自负全责。如不认可该条款，则不能使用/引用软件包内所有代码和文件。<br>
+                    目前仅开放白菜音色，后续将扩展为本地训练推理工具，用户可训练自己的音色进行社区共享。<br>
+                    男转女推荐+12key，女转男推荐-12key，如果音域爆炸导致音色失真也可以自己调整到合适音域
+                    """)
+                with gr.Row():
+                    with gr.Column():
+                        sid0 = gr.Dropdown(label="音色", choices=names)
+                        vc_transform0 = gr.Number(label="变调（整数，半音数量，升八度12降八度-12）", value=12)
+                        f0_file = gr.File(label="F0曲线文件，可选，一行一个音高，代替默认F0及升降调")
+                    input_audio0 = gr.Audio(label="上传音频")
+                    but0=gr.Button("转换", variant="primary")
+                    with gr.Column():
+                        vc_output1 = gr.Textbox(label="输出信息")
+                        vc_output2 = gr.Audio(label="输出音频")
+                    but0.click(vc_single, [sid0, input_audio0, vc_transform0,f0_file], [vc_output1, vc_output2])
+            with gr.Group():
+                gr.Markdown(value="""
+                    批量转换，上传多个音频文件，在指定文件夹（默认opt）下输出转换的音频。<br>
+                    合格的文件夹路径格式举例：E:\codes\py39\\vits_vc_gpu\白鹭霜华测试样例（去文件管理器地址栏拷就行了）
+                    """)
+                with gr.Row():
+                    with gr.Column():
+                        sid1 = gr.Dropdown(label="音色", choices=names)
+                        vc_transform1 = gr.Number(label="变调（整数，半音数量，升八度12降八度-12）", value=12)
+                        opt_input = gr.Textbox(label="指定输出文件夹",value="opt")
+                    with gr.Column():
+                        dir_input = gr.Textbox(label="输入待处理音频文件夹路径")
+                        inputs = gr.File(file_count="multiple", label="也可批量输入音频文件，二选一，优先读文件夹")
+                    but1=gr.Button("转换", variant="primary")
+                    vc_output3 = gr.Textbox(label="输出信息")
+                    but1.click(vc_multi, [sid1, dir_input,opt_input,inputs, vc_transform1], [vc_output3])
+
+        with gr.TabItem("数据处理"):
+            with gr.Group():
+                gr.Markdown(value="""
+                    人声伴奏分离批量处理，使用UVR5模型。<br>
+                    不带和声用HP2，带和声且提取的人声不需要和声用HP5<br>
+                    合格的文件夹路径格式举例：E:\codes\py39\\vits_vc_gpu\白鹭霜华测试样例（去文件管理器地址栏拷就行了）
+                    """)
+                with gr.Row():
+                    with gr.Column():
+                        dir_wav_input = gr.Textbox(label="输入待处理音频文件夹路径")
+                        wav_inputs = gr.File(file_count="multiple", label="也可批量输入音频文件，二选一，优先读文件夹")
+                    with gr.Column():
+                        model_choose = gr.Dropdown(label="模型", choices=uvr5_names)
+                        opt_vocal_root = gr.Textbox(label="指定输出人声文件夹",value="opt")
+                        opt_ins_root = gr.Textbox(label="指定输出乐器文件夹",value="opt")
+                    but2=gr.Button("转换", variant="primary")
+                    vc_output4 = gr.Textbox(label="输出信息")
+                    but2.click(uvr, [model_choose, dir_wav_input,opt_vocal_root,opt_ins_root], [vc_output4])
+        with gr.TabItem("训练-待开放"):pass
+
+    # app.launch(server_name="0.0.0.0",server_port=7860)
+    app.launch(server_name="127.0.0.1",server_port=7860)

infer.py

@@ -0,0 +1,48 @@
+import torch, pdb, os,sys,librosa,warnings,traceback
+warnings.filterwarnings("ignore")
+torch.manual_seed(114514)
+sys.path.append(os.getcwd())
+from config import inp_root,opt_root,f0_up_key,person,is_half,device
+os.makedirs(opt_root,exist_ok=True)
+import soundfile as sf
+from infer_pack.models import SynthesizerTrnMs256NSF as SynthesizerTrn256
+from scipy.io import wavfile
+from fairseq import checkpoint_utils
+import scipy.signal as signal
+from vc_infer_pipeline import VC
+
+models, saved_cfg, task = checkpoint_utils.load_model_ensemble_and_task(["hubert_base.pt"],suffix="",)
+model = models[0]
+model = model.to(device)
+if(is_half):model = model.half()
+else:model = model.float()
+model.eval()
+
+cpt=torch.load(person,map_location="cpu")
+dv=cpt["dv"]
+tgt_sr=cpt["config"][-1]
+net_g = SynthesizerTrn256(*cpt["config"],is_half=is_half)
+net_g.load_state_dict(cpt["weight"],strict=True)
+net_g.eval().to(device)
+if(is_half):net_g = net_g.half()
+else:net_g = net_g.float()
+
+vc=VC(tgt_sr,device,is_half)
+
+for name in os.listdir(inp_root):
+    try:
+        wav_path="%s\%s"%(inp_root,name)
+        print("processing %s"%wav_path)
+        audio, sampling_rate = sf.read(wav_path)
+        if len(audio.shape) > 1:
+            audio = librosa.to_mono(audio.transpose(1, 0))
+        if sampling_rate != vc.sr:
+            audio = librosa.resample(audio, orig_sr=sampling_rate, target_sr=vc.sr)
+
+        times = [0, 0, 0]
+        audio_opt=vc.pipeline(model,net_g,dv,audio,times,f0_up_key,f0_file=None)
+        wavfile.write("%s/%s"%(opt_root,name), tgt_sr, audio_opt)
+    except:
+        traceback.print_exc()
+
+print(times)

infer_pack/attentions.py

@@ -0,0 +1,417 @@
+import copy
+import math
+import numpy as np
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+from infer_pack import commons
+from infer_pack import modules
+from infer_pack.modules import LayerNorm
+
+
+class Encoder(nn.Module):
+    def __init__(
+        self,
+        hidden_channels,
+        filter_channels,
+        n_heads,
+        n_layers,
+        kernel_size=1,
+        p_dropout=0.0,
+        window_size=10,
+        **kwargs
+    ):
+        super().__init__()
+        self.hidden_channels = hidden_channels
+        self.filter_channels = filter_channels
+        self.n_heads = n_heads
+        self.n_layers = n_layers
+        self.kernel_size = kernel_size
+        self.p_dropout = p_dropout
+        self.window_size = window_size
+
+        self.drop = nn.Dropout(p_dropout)
+        self.attn_layers = nn.ModuleList()
+        self.norm_layers_1 = nn.ModuleList()
+        self.ffn_layers = nn.ModuleList()
+        self.norm_layers_2 = nn.ModuleList()
+        for i in range(self.n_layers):
+            self.attn_layers.append(
+                MultiHeadAttention(
+                    hidden_channels,
+                    hidden_channels,
+                    n_heads,
+                    p_dropout=p_dropout,
+                    window_size=window_size,
+                )
+            )
+            self.norm_layers_1.append(LayerNorm(hidden_channels))
+            self.ffn_layers.append(
+                FFN(
+                    hidden_channels,
+                    hidden_channels,
+                    filter_channels,
+                    kernel_size,
+                    p_dropout=p_dropout,
+                )
+            )
+            self.norm_layers_2.append(LayerNorm(hidden_channels))
+
+    def forward(self, x, x_mask):
+        attn_mask = x_mask.unsqueeze(2) * x_mask.unsqueeze(-1)
+        x = x * x_mask
+        for i in range(self.n_layers):
+            y = self.attn_layers[i](x, x, attn_mask)
+            y = self.drop(y)
+            x = self.norm_layers_1[i](x + y)
+
+            y = self.ffn_layers[i](x, x_mask)
+            y = self.drop(y)
+            x = self.norm_layers_2[i](x + y)
+        x = x * x_mask
+        return x
+
+
+class Decoder(nn.Module):
+    def __init__(
+        self,
+        hidden_channels,
+        filter_channels,
+        n_heads,
+        n_layers,
+        kernel_size=1,
+        p_dropout=0.0,
+        proximal_bias=False,
+        proximal_init=True,
+        **kwargs
+    ):
+        super().__init__()
+        self.hidden_channels = hidden_channels
+        self.filter_channels = filter_channels
+        self.n_heads = n_heads
+        self.n_layers = n_layers
+        self.kernel_size = kernel_size
+        self.p_dropout = p_dropout
+        self.proximal_bias = proximal_bias
+        self.proximal_init = proximal_init
+
+        self.drop = nn.Dropout(p_dropout)
+        self.self_attn_layers = nn.ModuleList()
+        self.norm_layers_0 = nn.ModuleList()
+        self.encdec_attn_layers = nn.ModuleList()
+        self.norm_layers_1 = nn.ModuleList()
+        self.ffn_layers = nn.ModuleList()
+        self.norm_layers_2 = nn.ModuleList()
+        for i in range(self.n_layers):
+            self.self_attn_layers.append(
+                MultiHeadAttention(
+                    hidden_channels,
+                    hidden_channels,
+                    n_heads,
+                    p_dropout=p_dropout,
+                    proximal_bias=proximal_bias,
+                    proximal_init=proximal_init,
+                )
+            )
+            self.norm_layers_0.append(LayerNorm(hidden_channels))
+            self.encdec_attn_layers.append(
+                MultiHeadAttention(
+                    hidden_channels, hidden_channels, n_heads, p_dropout=p_dropout
+                )
+            )
+            self.norm_layers_1.append(LayerNorm(hidden_channels))
+            self.ffn_layers.append(
+                FFN(
+                    hidden_channels,
+                    hidden_channels,
+                    filter_channels,
+                    kernel_size,
+                    p_dropout=p_dropout,
+                    causal=True,
+                )
+            )
+            self.norm_layers_2.append(LayerNorm(hidden_channels))
+
+    def forward(self, x, x_mask, h, h_mask):
+        """
+        x: decoder input
+        h: encoder output
+        """
+        self_attn_mask = commons.subsequent_mask(x_mask.size(2)).to(
+            device=x.device, dtype=x.dtype
+        )
+        encdec_attn_mask = h_mask.unsqueeze(2) * x_mask.unsqueeze(-1)
+        x = x * x_mask
+        for i in range(self.n_layers):
+            y = self.self_attn_layers[i](x, x, self_attn_mask)
+            y = self.drop(y)
+            x = self.norm_layers_0[i](x + y)
+
+            y = self.encdec_attn_layers[i](x, h, encdec_attn_mask)
+            y = self.drop(y)
+            x = self.norm_layers_1[i](x + y)
+
+            y = self.ffn_layers[i](x, x_mask)
+            y = self.drop(y)
+            x = self.norm_layers_2[i](x + y)
+        x = x * x_mask
+        return x
+
+
+class MultiHeadAttention(nn.Module):
+    def __init__(
+        self,
+        channels,
+        out_channels,
+        n_heads,
+        p_dropout=0.0,
+        window_size=None,
+        heads_share=True,
+        block_length=None,
+        proximal_bias=False,
+        proximal_init=False,
+    ):
+        super().__init__()
+        assert channels % n_heads == 0
+
+        self.channels = channels
+        self.out_channels = out_channels
+        self.n_heads = n_heads
+        self.p_dropout = p_dropout
+        self.window_size = window_size
+        self.heads_share = heads_share
+        self.block_length = block_length
+        self.proximal_bias = proximal_bias
+        self.proximal_init = proximal_init
+        self.attn = None
+
+        self.k_channels = channels // n_heads
+        self.conv_q = nn.Conv1d(channels, channels, 1)
+        self.conv_k = nn.Conv1d(channels, channels, 1)
+        self.conv_v = nn.Conv1d(channels, channels, 1)
+        self.conv_o = nn.Conv1d(channels, out_channels, 1)
+        self.drop = nn.Dropout(p_dropout)
+
+        if window_size is not None:
+            n_heads_rel = 1 if heads_share else n_heads
+            rel_stddev = self.k_channels**-0.5
+            self.emb_rel_k = nn.Parameter(
+                torch.randn(n_heads_rel, window_size * 2 + 1, self.k_channels)
+                * rel_stddev
+            )
+            self.emb_rel_v = nn.Parameter(
+                torch.randn(n_heads_rel, window_size * 2 + 1, self.k_channels)
+                * rel_stddev
+            )
+
+        nn.init.xavier_uniform_(self.conv_q.weight)
+        nn.init.xavier_uniform_(self.conv_k.weight)
+        nn.init.xavier_uniform_(self.conv_v.weight)
+        if proximal_init:
+            with torch.no_grad():
+                self.conv_k.weight.copy_(self.conv_q.weight)
+                self.conv_k.bias.copy_(self.conv_q.bias)
+
+    def forward(self, x, c, attn_mask=None):
+        q = self.conv_q(x)
+        k = self.conv_k(c)
+        v = self.conv_v(c)
+
+        x, self.attn = self.attention(q, k, v, mask=attn_mask)
+
+        x = self.conv_o(x)
+        return x
+
+    def attention(self, query, key, value, mask=None):
+        # reshape [b, d, t] -> [b, n_h, t, d_k]
+        b, d, t_s, t_t = (*key.size(), query.size(2))
+        query = query.view(b, self.n_heads, self.k_channels, t_t).transpose(2, 3)
+        key = key.view(b, self.n_heads, self.k_channels, t_s).transpose(2, 3)
+        value = value.view(b, self.n_heads, self.k_channels, t_s).transpose(2, 3)
+
+        scores = torch.matmul(query / math.sqrt(self.k_channels), key.transpose(-2, -1))
+        if self.window_size is not None:
+            assert (
+                t_s == t_t
+            ), "Relative attention is only available for self-attention."
+            key_relative_embeddings = self._get_relative_embeddings(self.emb_rel_k, t_s)
+            rel_logits = self._matmul_with_relative_keys(
+                query / math.sqrt(self.k_channels), key_relative_embeddings
+            )
+            scores_local = self._relative_position_to_absolute_position(rel_logits)
+            scores = scores + scores_local
+        if self.proximal_bias:
+            assert t_s == t_t, "Proximal bias is only available for self-attention."
+            scores = scores + self._attention_bias_proximal(t_s).to(
+                device=scores.device, dtype=scores.dtype
+            )
+        if mask is not None:
+            scores = scores.masked_fill(mask == 0, -1e4)
+            if self.block_length is not None:
+                assert (
+                    t_s == t_t
+                ), "Local attention is only available for self-attention."
+                block_mask = (
+                    torch.ones_like(scores)
+                    .triu(-self.block_length)
+                    .tril(self.block_length)
+                )
+                scores = scores.masked_fill(block_mask == 0, -1e4)
+        p_attn = F.softmax(scores, dim=-1)  # [b, n_h, t_t, t_s]
+        p_attn = self.drop(p_attn)
+        output = torch.matmul(p_attn, value)
+        if self.window_size is not None:
+            relative_weights = self._absolute_position_to_relative_position(p_attn)
+            value_relative_embeddings = self._get_relative_embeddings(
+                self.emb_rel_v, t_s
+            )
+            output = output + self._matmul_with_relative_values(
+                relative_weights, value_relative_embeddings
+            )
+        output = (
+            output.transpose(2, 3).contiguous().view(b, d, t_t)
+        )  # [b, n_h, t_t, d_k] -> [b, d, t_t]
+        return output, p_attn
+
+    def _matmul_with_relative_values(self, x, y):
+        """
+        x: [b, h, l, m]
+        y: [h or 1, m, d]
+        ret: [b, h, l, d]
+        """
+        ret = torch.matmul(x, y.unsqueeze(0))
+        return ret
+
+    def _matmul_with_relative_keys(self, x, y):
+        """
+        x: [b, h, l, d]
+        y: [h or 1, m, d]
+        ret: [b, h, l, m]
+        """
+        ret = torch.matmul(x, y.unsqueeze(0).transpose(-2, -1))
+        return ret
+
+    def _get_relative_embeddings(self, relative_embeddings, length):
+        max_relative_position = 2 * self.window_size + 1
+        # Pad first before slice to avoid using cond ops.
+        pad_length = max(length - (self.window_size + 1), 0)
+        slice_start_position = max((self.window_size + 1) - length, 0)
+        slice_end_position = slice_start_position + 2 * length - 1
+        if pad_length > 0:
+            padded_relative_embeddings = F.pad(
+                relative_embeddings,
+                commons.convert_pad_shape([[0, 0], [pad_length, pad_length], [0, 0]]),
+            )
+        else:
+            padded_relative_embeddings = relative_embeddings
+        used_relative_embeddings = padded_relative_embeddings[
+            :, slice_start_position:slice_end_position
+        ]
+        return used_relative_embeddings
+
+    def _relative_position_to_absolute_position(self, x):
+        """
+        x: [b, h, l, 2*l-1]
+        ret: [b, h, l, l]
+        """
+        batch, heads, length, _ = x.size()
+        # Concat columns of pad to shift from relative to absolute indexing.
+        x = F.pad(x, commons.convert_pad_shape([[0, 0], [0, 0], [0, 0], [0, 1]]))
+
+        # Concat extra elements so to add up to shape (len+1, 2*len-1).
+        x_flat = x.view([batch, heads, length * 2 * length])
+        x_flat = F.pad(
+            x_flat, commons.convert_pad_shape([[0, 0], [0, 0], [0, length - 1]])
+        )
+
+        # Reshape and slice out the padded elements.
+        x_final = x_flat.view([batch, heads, length + 1, 2 * length - 1])[
+            :, :, :length, length - 1 :
+        ]
+        return x_final
+
+    def _absolute_position_to_relative_position(self, x):
+        """
+        x: [b, h, l, l]
+        ret: [b, h, l, 2*l-1]
+        """
+        batch, heads, length, _ = x.size()
+        # padd along column
+        x = F.pad(
+            x, commons.convert_pad_shape([[0, 0], [0, 0], [0, 0], [0, length - 1]])
+        )
+        x_flat = x.view([batch, heads, length**2 + length * (length - 1)])
+        # add 0's in the beginning that will skew the elements after reshape
+        x_flat = F.pad(x_flat, commons.convert_pad_shape([[0, 0], [0, 0], [length, 0]]))
+        x_final = x_flat.view([batch, heads, length, 2 * length])[:, :, :, 1:]
+        return x_final
+
+    def _attention_bias_proximal(self, length):
+        """Bias for self-attention to encourage attention to close positions.
+        Args:
+          length: an integer scalar.
+        Returns:
+          a Tensor with shape [1, 1, length, length]
+        """
+        r = torch.arange(length, dtype=torch.float32)
+        diff = torch.unsqueeze(r, 0) - torch.unsqueeze(r, 1)
+        return torch.unsqueeze(torch.unsqueeze(-torch.log1p(torch.abs(diff)), 0), 0)
+
+
+class FFN(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        out_channels,
+        filter_channels,
+        kernel_size,
+        p_dropout=0.0,
+        activation=None,
+        causal=False,
+    ):
+        super().__init__()
+        self.in_channels = in_channels
+        self.out_channels = out_channels
+        self.filter_channels = filter_channels
+        self.kernel_size = kernel_size
+        self.p_dropout = p_dropout
+        self.activation = activation
+        self.causal = causal
+
+        if causal:
+            self.padding = self._causal_padding
+        else:
+            self.padding = self._same_padding
+
+        self.conv_1 = nn.Conv1d(in_channels, filter_channels, kernel_size)
+        self.conv_2 = nn.Conv1d(filter_channels, out_channels, kernel_size)
+        self.drop = nn.Dropout(p_dropout)
+
+    def forward(self, x, x_mask):
+        x = self.conv_1(self.padding(x * x_mask))
+        if self.activation == "gelu":
+            x = x * torch.sigmoid(1.702 * x)
+        else:
+            x = torch.relu(x)
+        x = self.drop(x)
+        x = self.conv_2(self.padding(x * x_mask))
+        return x * x_mask
+
+    def _causal_padding(self, x):
+        if self.kernel_size == 1:
+            return x
+        pad_l = self.kernel_size - 1
+        pad_r = 0
+        padding = [[0, 0], [0, 0], [pad_l, pad_r]]
+        x = F.pad(x, commons.convert_pad_shape(padding))
+        return x
+
+    def _same_padding(self, x):
+        if self.kernel_size == 1:
+            return x
+        pad_l = (self.kernel_size - 1) // 2
+        pad_r = self.kernel_size // 2
+        padding = [[0, 0], [0, 0], [pad_l, pad_r]]
+        x = F.pad(x, commons.convert_pad_shape(padding))
+        return x

infer_pack/commons.py

@@ -0,0 +1,164 @@
+import math
+import numpy as np
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+
+def init_weights(m, mean=0.0, std=0.01):
+    classname = m.__class__.__name__
+    if classname.find("Conv") != -1:
+        m.weight.data.normal_(mean, std)
+
+
+def get_padding(kernel_size, dilation=1):
+    return int((kernel_size * dilation - dilation) / 2)
+
+
+def convert_pad_shape(pad_shape):
+    l = pad_shape[::-1]
+    pad_shape = [item for sublist in l for item in sublist]
+    return pad_shape
+
+
+def kl_divergence(m_p, logs_p, m_q, logs_q):
+    """KL(P||Q)"""
+    kl = (logs_q - logs_p) - 0.5
+    kl += (
+        0.5 * (torch.exp(2.0 * logs_p) + ((m_p - m_q) ** 2)) * torch.exp(-2.0 * logs_q)
+    )
+    return kl
+
+
+def rand_gumbel(shape):
+    """Sample from the Gumbel distribution, protect from overflows."""
+    uniform_samples = torch.rand(shape) * 0.99998 + 0.00001
+    return -torch.log(-torch.log(uniform_samples))
+
+
+def rand_gumbel_like(x):
+    g = rand_gumbel(x.size()).to(dtype=x.dtype, device=x.device)
+    return g
+
+
+def slice_segments(x, ids_str, segment_size=4):
+    ret = torch.zeros_like(x[:, :, :segment_size])
+    for i in range(x.size(0)):
+        idx_str = ids_str[i]
+        idx_end = idx_str + segment_size
+        ret[i] = x[i, :, idx_str:idx_end]
+    return ret
+def slice_segments2(x, ids_str, segment_size=4):
+    ret = torch.zeros_like(x[:,  :segment_size])
+    for i in range(x.size(0)):
+        idx_str = ids_str[i]
+        idx_end = idx_str + segment_size
+        ret[i] = x[i, idx_str:idx_end]
+    return ret
+
+
+def rand_slice_segments(x, x_lengths=None, segment_size=4):
+    b, d, t = x.size()
+    if x_lengths is None:
+        x_lengths = t
+    ids_str_max = x_lengths - segment_size + 1
+    ids_str = (torch.rand([b]).to(device=x.device) * ids_str_max).to(dtype=torch.long)
+    ret = slice_segments(x, ids_str, segment_size)
+    return ret, ids_str
+
+
+def get_timing_signal_1d(length, channels, min_timescale=1.0, max_timescale=1.0e4):
+    position = torch.arange(length, dtype=torch.float)
+    num_timescales = channels // 2
+    log_timescale_increment = math.log(float(max_timescale) / float(min_timescale)) / (
+        num_timescales - 1
+    )
+    inv_timescales = min_timescale * torch.exp(
+        torch.arange(num_timescales, dtype=torch.float) * -log_timescale_increment
+    )
+    scaled_time = position.unsqueeze(0) * inv_timescales.unsqueeze(1)
+    signal = torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], 0)
+    signal = F.pad(signal, [0, 0, 0, channels % 2])
+    signal = signal.view(1, channels, length)
+    return signal
+
+
+def add_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4):
+    b, channels, length = x.size()
+    signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
+    return x + signal.to(dtype=x.dtype, device=x.device)
+
+
+def cat_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4, axis=1):
+    b, channels, length = x.size()
+    signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
+    return torch.cat([x, signal.to(dtype=x.dtype, device=x.device)], axis)
+
+
+def subsequent_mask(length):
+    mask = torch.tril(torch.ones(length, length)).unsqueeze(0).unsqueeze(0)
+    return mask
+
+
+@torch.jit.script
+def fused_add_tanh_sigmoid_multiply(input_a, input_b, n_channels):
+    n_channels_int = n_channels[0]
+    in_act = input_a + input_b
+    t_act = torch.tanh(in_act[:, :n_channels_int, :])
+    s_act = torch.sigmoid(in_act[:, n_channels_int:, :])
+    acts = t_act * s_act
+    return acts
+
+
+def convert_pad_shape(pad_shape):
+    l = pad_shape[::-1]
+    pad_shape = [item for sublist in l for item in sublist]
+    return pad_shape
+
+
+def shift_1d(x):
+    x = F.pad(x, convert_pad_shape([[0, 0], [0, 0], [1, 0]]))[:, :, :-1]
+    return x
+
+
+def sequence_mask(length, max_length=None):
+    if max_length is None:
+        max_length = length.max()
+    x = torch.arange(max_length, dtype=length.dtype, device=length.device)
+    return x.unsqueeze(0) < length.unsqueeze(1)
+
+
+def generate_path(duration, mask):
+    """
+    duration: [b, 1, t_x]
+    mask: [b, 1, t_y, t_x]
+    """
+    device = duration.device
+
+    b, _, t_y, t_x = mask.shape
+    cum_duration = torch.cumsum(duration, -1)
+
+    cum_duration_flat = cum_duration.view(b * t_x)
+    path = sequence_mask(cum_duration_flat, t_y).to(mask.dtype)
+    path = path.view(b, t_x, t_y)
+    path = path - F.pad(path, convert_pad_shape([[0, 0], [1, 0], [0, 0]]))[:, :-1]
+    path = path.unsqueeze(1).transpose(2, 3) * mask
+    return path
+
+
+def clip_grad_value_(parameters, clip_value, norm_type=2):
+    if isinstance(parameters, torch.Tensor):
+        parameters = [parameters]
+    parameters = list(filter(lambda p: p.grad is not None, parameters))
+    norm_type = float(norm_type)
+    if clip_value is not None:
+        clip_value = float(clip_value)
+
+    total_norm = 0
+    for p in parameters:
+        param_norm = p.grad.data.norm(norm_type)
+        total_norm += param_norm.item() ** norm_type
+        if clip_value is not None:
+            p.grad.data.clamp_(min=-clip_value, max=clip_value)
+    total_norm = total_norm ** (1.0 / norm_type)
+    return total_norm

infer_pack/models.py

@@ -0,0 +1,664 @@
+import math,pdb,os
+from time import time as ttime
+import torch
+from torch import nn
+from torch.nn import functional as F
+from infer_pack import modules
+from infer_pack import attentions
+from torch.nn import Conv1d, ConvTranspose1d, AvgPool1d, Conv2d
+from torch.nn.utils import weight_norm, remove_weight_norm, spectral_norm
+from infer_pack.commons import init_weights
+import numpy as np
+from infer_pack import commons
+class TextEncoder256(nn.Module):
+    def __init__(
+        self,        out_channels,        hidden_channels,        filter_channels,        n_heads,        n_layers,        kernel_size,        p_dropout,        f0=True    ):
+        super().__init__()
+        self.out_channels = out_channels
+        self.hidden_channels = hidden_channels
+        self.filter_channels = filter_channels
+        self.n_heads = n_heads
+        self.n_layers = n_layers
+        self.kernel_size = kernel_size
+        self.p_dropout = p_dropout
+        self.emb_phone = nn.Linear(256, hidden_channels)
+        self.lrelu=nn.LeakyReLU(0.1,inplace=True)
+        if(f0==True):
+            self.emb_pitch = nn.Embedding(256, hidden_channels)  # pitch 256
+        self.encoder = attentions.Encoder(
+            hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout
+        )
+        self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
+
+    def forward(self, phone, pitch, lengths):
+        if(pitch==None):
+            x = self.emb_phone(phone)
+        else:
+            x = self.emb_phone(phone) + self.emb_pitch(pitch)
+        x = x * math.sqrt(self.hidden_channels)  # [b, t, h]
+        x=self.lrelu(x)
+        x = torch.transpose(x, 1, -1)  # [b, h, t]
+        x_mask = torch.unsqueeze(commons.sequence_mask(lengths, x.size(2)), 1).to(
+            x.dtype
+        )
+        x = self.encoder(x * x_mask, x_mask)
+        stats = self.proj(x) * x_mask
+
+        m, logs = torch.split(stats, self.out_channels, dim=1)
+        return m, logs, x_mask
+class TextEncoder256km(nn.Module):
+    def __init__(
+        self,        out_channels,        hidden_channels,        filter_channels,        n_heads,        n_layers,        kernel_size,        p_dropout,        f0=True    ):
+        super().__init__()
+        self.out_channels = out_channels
+        self.hidden_channels = hidden_channels
+        self.filter_channels = filter_channels
+        self.n_heads = n_heads
+        self.n_layers = n_layers
+        self.kernel_size = kernel_size
+        self.p_dropout = p_dropout
+        # self.emb_phone = nn.Linear(256, hidden_channels)
+        self.emb_phone = nn.Embedding(500, hidden_channels)
+        self.lrelu=nn.LeakyReLU(0.1,inplace=True)
+        if(f0==True):
+            self.emb_pitch = nn.Embedding(256, hidden_channels)  # pitch 256
+        self.encoder = attentions.Encoder(
+            hidden_channels, filter_channels, n_heads, n_layers, kernel_size, p_dropout
+        )
+        self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
+
+    def forward(self, phone, pitch, lengths):
+        if(pitch==None):
+            x = self.emb_phone(phone)
+        else:
+            x = self.emb_phone(phone) + self.emb_pitch(pitch)
+        x = x * math.sqrt(self.hidden_channels)  # [b, t, h]
+        x=self.lrelu(x)
+        x = torch.transpose(x, 1, -1)  # [b, h, t]
+        x_mask = torch.unsqueeze(commons.sequence_mask(lengths, x.size(2)), 1).to(
+            x.dtype
+        )
+        x = self.encoder(x * x_mask, x_mask)
+        stats = self.proj(x) * x_mask
+
+        m, logs = torch.split(stats, self.out_channels, dim=1)
+        return m, logs, x_mask
+class ResidualCouplingBlock(nn.Module):
+    def __init__(
+        self,
+        channels,
+        hidden_channels,
+        kernel_size,
+        dilation_rate,
+        n_layers,
+        n_flows=4,
+        gin_channels=0,
+    ):
+        super().__init__()
+        self.channels = channels
+        self.hidden_channels = hidden_channels
+        self.kernel_size = kernel_size
+        self.dilation_rate = dilation_rate
+        self.n_layers = n_layers
+        self.n_flows = n_flows
+        self.gin_channels = gin_channels
+
+        self.flows = nn.ModuleList()
+        for i in range(n_flows):
+            self.flows.append(
+                modules.ResidualCouplingLayer(
+                    channels,
+                    hidden_channels,
+                    kernel_size,
+                    dilation_rate,
+                    n_layers,
+                    gin_channels=gin_channels,
+                    mean_only=True,
+                )
+            )
+            self.flows.append(modules.Flip())
+
+    def forward(self, x, x_mask, g=None, reverse=False):
+        if not reverse:
+            for flow in self.flows:
+                x, _ = flow(x, x_mask, g=g, reverse=reverse)
+        else:
+            for flow in reversed(self.flows):
+                x = flow(x, x_mask, g=g, reverse=reverse)
+        return x
+
+    def remove_weight_norm(self):
+        for i in range(self.n_flows):
+            self.flows[i * 2].remove_weight_norm()
+class PosteriorEncoder(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        out_channels,
+        hidden_channels,
+        kernel_size,
+        dilation_rate,
+        n_layers,
+        gin_channels=0,
+    ):
+        super().__init__()
+        self.in_channels = in_channels
+        self.out_channels = out_channels
+        self.hidden_channels = hidden_channels
+        self.kernel_size = kernel_size
+        self.dilation_rate = dilation_rate
+        self.n_layers = n_layers
+        self.gin_channels = gin_channels
+
+        self.pre = nn.Conv1d(in_channels, hidden_channels, 1)
+        self.enc = modules.WN(
+            hidden_channels,
+            kernel_size,
+            dilation_rate,
+            n_layers,
+            gin_channels=gin_channels,
+        )
+        self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
+
+    def forward(self, x, x_lengths, g=None):
+        x_mask = torch.unsqueeze(commons.sequence_mask(x_lengths, x.size(2)), 1).to(
+            x.dtype
+        )
+        x = self.pre(x) * x_mask
+        x = self.enc(x, x_mask, g=g)
+        stats = self.proj(x) * x_mask
+        m, logs = torch.split(stats, self.out_channels, dim=1)
+        z = (m + torch.randn_like(m) * torch.exp(logs)) * x_mask
+        return z, m, logs, x_mask
+
+    def remove_weight_norm(self):
+        self.enc.remove_weight_norm()
+class Generator(torch.nn.Module):
+    def __init__(
+        self,
+        initial_channel,
+        resblock,
+        resblock_kernel_sizes,
+        resblock_dilation_sizes,
+        upsample_rates,
+        upsample_initial_channel,
+        upsample_kernel_sizes,
+        gin_channels=0,
+    ):
+        super(Generator, self).__init__()
+        self.num_kernels = len(resblock_kernel_sizes)
+        self.num_upsamples = len(upsample_rates)
+        self.conv_pre = Conv1d(
+            initial_channel, upsample_initial_channel, 7, 1, padding=3
+        )
+        resblock = modules.ResBlock1 if resblock == "1" else modules.ResBlock2
+
+        self.ups = nn.ModuleList()
+        for i, (u, k) in enumerate(zip(upsample_rates, upsample_kernel_sizes)):
+            self.ups.append(
+                weight_norm(
+                    ConvTranspose1d(
+                        upsample_initial_channel // (2**i),
+                        upsample_initial_channel // (2 ** (i + 1)),
+                        k,
+                        u,
+                        padding=(k - u) // 2,
+                    )
+                )
+            )
+
+        self.resblocks = nn.ModuleList()
+        for i in range(len(self.ups)):
+            ch = upsample_initial_channel // (2 ** (i + 1))
+            for j, (k, d) in enumerate(
+                zip(resblock_kernel_sizes, resblock_dilation_sizes)
+            ):
+                self.resblocks.append(resblock(ch, k, d))
+
+        self.conv_post = Conv1d(ch, 1, 7, 1, padding=3, bias=False)
+        self.ups.apply(init_weights)
+
+        if gin_channels != 0:
+            self.cond = nn.Conv1d(gin_channels, upsample_initial_channel, 1)
+
+    def forward(self, x, g=None):
+        x = self.conv_pre(x)
+        if g is not None:
+            x = x + self.cond(g)
+
+        for i in range(self.num_upsamples):
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            x = self.ups[i](x)
+            xs = None
+            for j in range(self.num_kernels):
+                if xs is None:
+                    xs = self.resblocks[i * self.num_kernels + j](x)
+                else:
+                    xs += self.resblocks[i * self.num_kernels + j](x)
+            x = xs / self.num_kernels
+        x = F.leaky_relu(x)
+        x = self.conv_post(x)
+        x = torch.tanh(x)
+
+        return x
+
+    def remove_weight_norm(self):
+        for l in self.ups:
+            remove_weight_norm(l)
+        for l in self.resblocks:
+            l.remove_weight_norm()
+class SineGen(torch.nn.Module):
+    """ Definition of sine generator
+    SineGen(samp_rate, harmonic_num = 0,
+            sine_amp = 0.1, noise_std = 0.003,
+            voiced_threshold = 0,
+            flag_for_pulse=False)
+    samp_rate: sampling rate in Hz
+    harmonic_num: number of harmonic overtones (default 0)
+    sine_amp: amplitude of sine-wavefrom (default 0.1)
+    noise_std: std of Gaussian noise (default 0.003)
+    voiced_thoreshold: F0 threshold for U/V classification (default 0)
+    flag_for_pulse: this SinGen is used inside PulseGen (default False)
+    Note: when flag_for_pulse is True, the first time step of a voiced
+        segment is always sin(np.pi) or cos(0)
+    """
+
+    def __init__(self, samp_rate, harmonic_num=0,
+                 sine_amp=0.1, noise_std=0.003,
+                 voiced_threshold=0,
+                 flag_for_pulse=False):
+        super(SineGen, self).__init__()
+        self.sine_amp = sine_amp
+        self.noise_std = noise_std
+        self.harmonic_num = harmonic_num
+        self.dim = self.harmonic_num + 1
+        self.sampling_rate = samp_rate
+        self.voiced_threshold = voiced_threshold
+
+    def _f02uv(self, f0):
+        # generate uv signal
+        uv = torch.ones_like(f0)
+        uv = uv * (f0 > self.voiced_threshold)
+        return uv
+
+    def forward(self, f0,upp):
+        """ sine_tensor, uv = forward(f0)
+        input F0: tensor(batchsize=1, length, dim=1)
+                  f0 for unvoiced steps should be 0
+        output sine_tensor: tensor(batchsize=1, length, dim)
+        output uv: tensor(batchsize=1, length, 1)
+        """
+        with torch.no_grad():
+            f0 = f0[:, None].transpose(1, 2)
+            f0_buf = torch.zeros(f0.shape[0], f0.shape[1], self.dim,device=f0.device)
+            # fundamental component
+            f0_buf[:, :, 0] = f0[:, :, 0]
+            for idx in np.arange(self.harmonic_num):f0_buf[:, :, idx + 1] = f0_buf[:, :, 0] * (idx + 2)# idx + 2: the (idx+1)-th overtone, (idx+2)-th harmonic
+            rad_values = (f0_buf / self.sampling_rate) % 1###%1意味着n_har的乘积无法后处理优化
+            rand_ini = torch.rand(f0_buf.shape[0], f0_buf.shape[2], device=f0_buf.device)
+            rand_ini[:, 0] = 0
+            rad_values[:, 0, :] = rad_values[:, 0, :] + rand_ini
+            tmp_over_one = torch.cumsum(rad_values, 1)# % 1  #####%1意味着后面的cumsum无法再优化
+            tmp_over_one*=upp
+            tmp_over_one=F.interpolate(tmp_over_one.transpose(2, 1), scale_factor=upp, mode='linear', align_corners=True).transpose(2, 1)
+            rad_values=F.interpolate(rad_values.transpose(2, 1), scale_factor=upp, mode='nearest').transpose(2, 1)#######
+            tmp_over_one%=1
+            tmp_over_one_idx = (tmp_over_one[:, 1:, :] - tmp_over_one[:, :-1, :]) < 0
+            cumsum_shift = torch.zeros_like(rad_values)
+            cumsum_shift[:, 1:, :] = tmp_over_one_idx * -1.0
+            sine_waves = torch.sin(torch.cumsum(rad_values + cumsum_shift, dim=1) * 2 * np.pi)
+            sine_waves = sine_waves * self.sine_amp
+            uv = self._f02uv(f0)
+            uv = F.interpolate(uv.transpose(2, 1), scale_factor=upp, mode='nearest').transpose(2, 1)
+            noise_amp = uv * self.noise_std + (1 - uv) * self.sine_amp / 3
+            noise = noise_amp * torch.randn_like(sine_waves)
+            sine_waves = sine_waves * uv + noise
+        return sine_waves, uv, noise
+class SourceModuleHnNSF(torch.nn.Module):
+    """ SourceModule for hn-nsf
+    SourceModule(sampling_rate, harmonic_num=0, sine_amp=0.1,
+                 add_noise_std=0.003, voiced_threshod=0)
+    sampling_rate: sampling_rate in Hz
+    harmonic_num: number of harmonic above F0 (default: 0)
+    sine_amp: amplitude of sine source signal (default: 0.1)
+    add_noise_std: std of additive Gaussian noise (default: 0.003)
+        note that amplitude of noise in unvoiced is decided
+        by sine_amp
+    voiced_threshold: threhold to set U/V given F0 (default: 0)
+    Sine_source, noise_source = SourceModuleHnNSF(F0_sampled)
+    F0_sampled (batchsize, length, 1)
+    Sine_source (batchsize, length, 1)
+    noise_source (batchsize, length 1)
+    uv (batchsize, length, 1)
+    """
+
+    def __init__(self, sampling_rate, harmonic_num=0, sine_amp=0.1,
+                 add_noise_std=0.003, voiced_threshod=0,is_half=True):
+        super(SourceModuleHnNSF, self).__init__()
+
+        self.sine_amp = sine_amp
+        self.noise_std = add_noise_std
+        self.is_half=is_half
+        # to produce sine waveforms
+        self.l_sin_gen = SineGen(sampling_rate, harmonic_num,
+                                 sine_amp, add_noise_std, voiced_threshod)
+
+        # to merge source harmonics into a single excitation
+        self.l_linear = torch.nn.Linear(harmonic_num + 1, 1)
+        self.l_tanh = torch.nn.Tanh()
+
+    def forward(self, x,upp=None):
+        sine_wavs, uv, _ = self.l_sin_gen(x,upp)
+        if(self.is_half==True):sine_wavs=sine_wavs.half()
+        sine_merge = self.l_tanh(self.l_linear(sine_wavs))
+        return sine_merge,None,None# noise, uv
+class GeneratorNSF(torch.nn.Module):
+    def __init__(
+        self,
+        initial_channel,
+        resblock,
+        resblock_kernel_sizes,
+        resblock_dilation_sizes,
+        upsample_rates,
+        upsample_initial_channel,
+        upsample_kernel_sizes,
+        gin_channels=0,
+        sr=40000,
+        is_half=False
+    ):
+        super(GeneratorNSF, self).__init__()
+        self.num_kernels = len(resblock_kernel_sizes)
+        self.num_upsamples = len(upsample_rates)
+
+        self.f0_upsamp = torch.nn.Upsample(scale_factor=np.prod(upsample_rates))
+        self.m_source = SourceModuleHnNSF(
+            sampling_rate=sr,
+            harmonic_num=0,
+            is_half=is_half
+        )
+        self.noise_convs = nn.ModuleList()
+        self.conv_pre = Conv1d(
+            initial_channel, upsample_initial_channel, 7, 1, padding=3
+        )
+        resblock = modules.ResBlock1 if resblock == "1" else modules.ResBlock2
+
+        self.ups = nn.ModuleList()
+        for i, (u, k) in enumerate(zip(upsample_rates, upsample_kernel_sizes)):
+            c_cur = upsample_initial_channel // (2 ** (i + 1))
+            self.ups.append(
+                weight_norm(
+                    ConvTranspose1d(
+                        upsample_initial_channel // (2**i),
+                        upsample_initial_channel // (2 ** (i + 1)),
+                        k,
+                        u,
+                        padding=(k - u) // 2,
+                    )
+                )
+            )
+            if i + 1 < len(upsample_rates):
+                stride_f0 = np.prod(upsample_rates[i + 1:])
+                self.noise_convs.append(Conv1d(
+                    1, c_cur, kernel_size=stride_f0 * 2, stride=stride_f0, padding=stride_f0 // 2))
+            else:
+                self.noise_convs.append(Conv1d(1, c_cur, kernel_size=1))
+
+        self.resblocks = nn.ModuleList()
+        for i in range(len(self.ups)):
+            ch = upsample_initial_channel // (2 ** (i + 1))
+            for j, (k, d) in enumerate(
+                zip(resblock_kernel_sizes, resblock_dilation_sizes)
+            ):
+                self.resblocks.append(resblock(ch, k, d))
+
+        self.conv_post = Conv1d(ch, 1, 7, 1, padding=3, bias=False)
+        self.ups.apply(init_weights)
+
+        if gin_channels != 0:
+            self.cond = nn.Conv1d(gin_channels, upsample_initial_channel, 1)
+
+        self.upp=np.prod(upsample_rates)
+
+    def forward(self, x, f0,g=None):
+        har_source, noi_source, uv = self.m_source(f0,self.upp)
+        har_source = har_source.transpose(1, 2)
+        x = self.conv_pre(x)
+        if g is not None:
+            x = x + self.cond(g)
+
+        for i in range(self.num_upsamples):
+            x = F.leaky_relu(x, modules.LRELU_SLOPE)
+            x = self.ups[i](x)
+            x_source = self.noise_convs[i](har_source)
+            x = x + x_source
+            xs = None
+            for j in range(self.num_kernels):
+                if xs is None:
+                    xs = self.resblocks[i * self.num_kernels + j](x)
+                else:
+                    xs += self.resblocks[i * self.num_kernels + j](x)
+            x = xs / self.num_kernels
+        x = F.leaky_relu(x)
+        x = self.conv_post(x)
+        x = torch.tanh(x)
+        return x
+
+    def remove_weight_norm(self):
+        for l in self.ups:
+            remove_weight_norm(l)
+        for l in self.resblocks:
+            l.remove_weight_norm()
+class SynthesizerTrnMs256NSF(nn.Module):
+    """
+    Synthesizer for Training
+    """
+
+    def __init__(
+        self,
+        spec_channels,
+        segment_size,
+        inter_channels,
+        hidden_channels,
+        filter_channels,
+        n_heads,
+        n_layers,
+        kernel_size,
+        p_dropout,
+        resblock,
+        resblock_kernel_sizes,
+        resblock_dilation_sizes,
+        upsample_rates,
+        upsample_initial_channel,
+        upsample_kernel_sizes,
+        spk_embed_dim,
+        gin_channels=0,
+        sr=40000,
+        **kwargs
+    ):
+
+        super().__init__()
+        self.spec_channels = spec_channels
+        self.inter_channels = inter_channels
+        self.hidden_channels = hidden_channels
+        self.filter_channels = filter_channels
+        self.n_heads = n_heads
+        self.n_layers = n_layers
+        self.kernel_size = kernel_size
+        self.p_dropout = p_dropout
+        self.resblock = resblock
+        self.resblock_kernel_sizes = resblock_kernel_sizes
+        self.resblock_dilation_sizes = resblock_dilation_sizes
+        self.upsample_rates = upsample_rates
+        self.upsample_initial_channel = upsample_initial_channel
+        self.upsample_kernel_sizes = upsample_kernel_sizes
+        self.segment_size = segment_size
+        self.gin_channels = gin_channels
+        self.spk_embed_dim=spk_embed_dim
+        self.enc_p = TextEncoder256(
+            inter_channels,
+            hidden_channels,
+            filter_channels,
+            n_heads,
+            n_layers,
+            kernel_size,
+            p_dropout,
+        )
+        self.dec = GeneratorNSF(
+            inter_channels,
+            resblock,
+            resblock_kernel_sizes,
+            resblock_dilation_sizes,
+            upsample_rates,
+            upsample_initial_channel,
+            upsample_kernel_sizes,
+            gin_channels=0,
+            sr=sr,
+            is_half=kwargs["is_half"]
+        )
+        self.enc_q = PosteriorEncoder(
+            spec_channels,
+            inter_channels,
+            hidden_channels,
+            5,
+            1,
+            16,
+            gin_channels=gin_channels,
+        )
+        self.flow = ResidualCouplingBlock(
+            inter_channels, hidden_channels, 5, 1, 3, gin_channels=gin_channels
+        )
+        self.emb_g = nn.Linear(self.spk_embed_dim, gin_channels)
+
+    def remove_weight_norm(self):
+        self.dec.remove_weight_norm()
+        self.flow.remove_weight_norm()
+        self.enc_q.remove_weight_norm()
+
+    def infer(self, phone, phone_lengths, pitch,pitchf, ds,max_len=None):
+        m_p, logs_p, x_mask = self.enc_p(phone, pitch, phone_lengths)
+        if("float16"in str(m_p.dtype)):ds=ds.half()
+        ds=ds.to(m_p.device)
+        g = self.emb_g(ds).unsqueeze(-1)  # [b, h, 1]#
+        z_p = (m_p + torch.exp(logs_p) * torch.randn_like(m_p) * 0.66) * x_mask
+
+        z = self.flow(z_p, x_mask, g=g, reverse=True)
+        o = self.dec((z * x_mask)[:, :, :max_len],pitchf, g=None)
+        return o, x_mask, (z, z_p, m_p, logs_p)
+class SynthesizerTrn256NSFkm(nn.Module):
+    """
+    Synthesizer for Training
+    """
+
+    def __init__(
+        self,
+        spec_channels,
+        segment_size,
+        inter_channels,
+        hidden_channels,
+        filter_channels,
+        n_heads,
+        n_layers,
+        kernel_size,
+        p_dropout,
+        resblock,
+        resblock_kernel_sizes,
+        resblock_dilation_sizes,
+        upsample_rates,
+        upsample_initial_channel,
+        upsample_kernel_sizes,
+        spk_embed_dim,
+        gin_channels=0,
+        sr=40000,
+        **kwargs
+    ):
+
+        super().__init__()
+        self.spec_channels = spec_channels
+        self.inter_channels = inter_channels
+        self.hidden_channels = hidden_channels
+        self.filter_channels = filter_channels
+        self.n_heads = n_heads
+        self.n_layers = n_layers
+        self.kernel_size = kernel_size
+        self.p_dropout = p_dropout
+        self.resblock = resblock
+        self.resblock_kernel_sizes = resblock_kernel_sizes
+        self.resblock_dilation_sizes = resblock_dilation_sizes
+        self.upsample_rates = upsample_rates
+        self.upsample_initial_channel = upsample_initial_channel
+        self.upsample_kernel_sizes = upsample_kernel_sizes
+        self.segment_size = segment_size
+        self.gin_channels = gin_channels
+
+        self.enc_p = TextEncoder256km(
+            inter_channels,
+            hidden_channels,
+            filter_channels,
+            n_heads,
+            n_layers,
+            kernel_size,
+            p_dropout,
+        )
+        self.dec = GeneratorNSF(
+            inter_channels,
+            resblock,
+            resblock_kernel_sizes,
+            resblock_dilation_sizes,
+            upsample_rates,
+            upsample_initial_channel,
+            upsample_kernel_sizes,
+            gin_channels=0,
+            sr=sr,
+            is_half=kwargs["is_half"]
+        )
+        self.enc_q = PosteriorEncoder(
+            spec_channels,
+            inter_channels,
+            hidden_channels,
+            5,
+            1,
+            16,
+            gin_channels=gin_channels,
+        )
+        self.flow = ResidualCouplingBlock(
+            inter_channels, hidden_channels, 5, 1, 3, gin_channels=gin_channels
+        )
+
+    def remove_weight_norm(self):
+        self.dec.remove_weight_norm()
+        self.flow.remove_weight_norm()
+        self.enc_q.remove_weight_norm()
+
+    def forward(self, phone, phone_lengths, pitch, pitchf, y, y_lengths):
+        m_p, logs_p, x_mask = self.enc_p(phone, pitch, phone_lengths)
+
+        z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=None)
+        z_p = self.flow(z, y_mask, g=None)
+
+        z_slice, ids_slice = commons.rand_slice_segments(
+            z, y_lengths, self.segment_size
+        )
+
+        pitchf = commons.slice_segments2(
+            pitchf, ids_slice, self.segment_size
+        )
+        o = self.dec(z_slice, pitchf,g=None)
+        return o, ids_slice, x_mask, y_mask, (z, z_p, m_p, logs_p, m_q, logs_q)
+
+    def infer(self, phone, phone_lengths, pitch, nsff0,max_len=None):
+        # torch.cuda.synchronize()
+        # t0=ttime()
+        m_p, logs_p, x_mask = self.enc_p(phone, pitch, phone_lengths)
+        # torch.cuda.synchronize()
+        # t1=ttime()
+        z_p = (m_p + torch.exp(logs_p) * torch.randn_like(m_p) * 0.66) * x_mask
+        # torch.cuda.synchronize()
+        # t2=ttime()
+        z = self.flow(z_p, x_mask, g=None, reverse=True)
+        # torch.cuda.synchronize()
+        # t3=ttime()
+        o = self.dec((z * x_mask)[:, :, :max_len], nsff0,g=None)
+        # torch.cuda.synchronize()
+        # t4=ttime()
+        # print(1233333333333333333333333,t1-t0,t2-t1,t3-t2,t4-t3)
+        return o, x_mask, (z, z_p, m_p, logs_p)

infer_pack/modules.py

@@ -0,0 +1,522 @@
+import copy
+import math
+import numpy as np
+import scipy
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+from torch.nn import Conv1d, ConvTranspose1d, AvgPool1d, Conv2d
+from torch.nn.utils import weight_norm, remove_weight_norm
+
+from infer_pack import commons
+from infer_pack.commons import init_weights, get_padding
+from infer_pack.transforms import piecewise_rational_quadratic_transform
+
+
+LRELU_SLOPE = 0.1
+
+
+class LayerNorm(nn.Module):
+    def __init__(self, channels, eps=1e-5):
+        super().__init__()
+        self.channels = channels
+        self.eps = eps
+
+        self.gamma = nn.Parameter(torch.ones(channels))
+        self.beta = nn.Parameter(torch.zeros(channels))
+
+    def forward(self, x):
+        x = x.transpose(1, -1)
+        x = F.layer_norm(x, (self.channels,), self.gamma, self.beta, self.eps)
+        return x.transpose(1, -1)
+
+
+class ConvReluNorm(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        hidden_channels,
+        out_channels,
+        kernel_size,
+        n_layers,
+        p_dropout,
+    ):
+        super().__init__()
+        self.in_channels = in_channels
+        self.hidden_channels = hidden_channels
+        self.out_channels = out_channels
+        self.kernel_size = kernel_size
+        self.n_layers = n_layers
+        self.p_dropout = p_dropout
+        assert n_layers > 1, "Number of layers should be larger than 0."
+
+        self.conv_layers = nn.ModuleList()
+        self.norm_layers = nn.ModuleList()
+        self.conv_layers.append(
+            nn.Conv1d(
+                in_channels, hidden_channels, kernel_size, padding=kernel_size // 2
+            )
+        )
+        self.norm_layers.append(LayerNorm(hidden_channels))
+        self.relu_drop = nn.Sequential(nn.ReLU(), nn.Dropout(p_dropout))
+        for _ in range(n_layers - 1):
+            self.conv_layers.append(
+                nn.Conv1d(
+                    hidden_channels,
+                    hidden_channels,
+                    kernel_size,
+                    padding=kernel_size // 2,
+                )
+            )
+            self.norm_layers.append(LayerNorm(hidden_channels))
+        self.proj = nn.Conv1d(hidden_channels, out_channels, 1)
+        self.proj.weight.data.zero_()
+        self.proj.bias.data.zero_()
+
+    def forward(self, x, x_mask):
+        x_org = x
+        for i in range(self.n_layers):
+            x = self.conv_layers[i](x * x_mask)
+            x = self.norm_layers[i](x)
+            x = self.relu_drop(x)
+        x = x_org + self.proj(x)
+        return x * x_mask
+
+
+class DDSConv(nn.Module):
+    """
+    Dialted and Depth-Separable Convolution
+    """
+
+    def __init__(self, channels, kernel_size, n_layers, p_dropout=0.0):
+        super().__init__()
+        self.channels = channels
+        self.kernel_size = kernel_size
+        self.n_layers = n_layers
+        self.p_dropout = p_dropout
+
+        self.drop = nn.Dropout(p_dropout)
+        self.convs_sep = nn.ModuleList()
+        self.convs_1x1 = nn.ModuleList()
+        self.norms_1 = nn.ModuleList()
+        self.norms_2 = nn.ModuleList()
+        for i in range(n_layers):
+            dilation = kernel_size**i
+            padding = (kernel_size * dilation - dilation) // 2
+            self.convs_sep.append(
+                nn.Conv1d(
+                    channels,
+                    channels,
+                    kernel_size,
+                    groups=channels,
+                    dilation=dilation,
+                    padding=padding,
+                )
+            )
+            self.convs_1x1.append(nn.Conv1d(channels, channels, 1))
+            self.norms_1.append(LayerNorm(channels))
+            self.norms_2.append(LayerNorm(channels))
+
+    def forward(self, x, x_mask, g=None):
+        if g is not None:
+            x = x + g
+        for i in range(self.n_layers):
+            y = self.convs_sep[i](x * x_mask)
+            y = self.norms_1[i](y)
+            y = F.gelu(y)
+            y = self.convs_1x1[i](y)
+            y = self.norms_2[i](y)
+            y = F.gelu(y)
+            y = self.drop(y)
+            x = x + y
+        return x * x_mask
+
+
+class WN(torch.nn.Module):
+    def __init__(
+        self,
+        hidden_channels,
+        kernel_size,
+        dilation_rate,
+        n_layers,
+        gin_channels=0,
+        p_dropout=0,
+    ):
+        super(WN, self).__init__()
+        assert kernel_size % 2 == 1
+        self.hidden_channels = hidden_channels
+        self.kernel_size = (kernel_size,)
+        self.dilation_rate = dilation_rate
+        self.n_layers = n_layers
+        self.gin_channels = gin_channels
+        self.p_dropout = p_dropout
+
+        self.in_layers = torch.nn.ModuleList()
+        self.res_skip_layers = torch.nn.ModuleList()
+        self.drop = nn.Dropout(p_dropout)
+
+        if gin_channels != 0:
+            cond_layer = torch.nn.Conv1d(
+                gin_channels, 2 * hidden_channels * n_layers, 1
+            )
+            self.cond_layer = torch.nn.utils.weight_norm(cond_layer, name="weight")
+
+        for i in range(n_layers):
+            dilation = dilation_rate**i
+            padding = int((kernel_size * dilation - dilation) / 2)
+            in_layer = torch.nn.Conv1d(
+                hidden_channels,
+                2 * hidden_channels,
+                kernel_size,
+                dilation=dilation,
+                padding=padding,
+            )
+            in_layer = torch.nn.utils.weight_norm(in_layer, name="weight")
+            self.in_layers.append(in_layer)
+
+            # last one is not necessary
+            if i < n_layers - 1:
+                res_skip_channels = 2 * hidden_channels
+            else:
+                res_skip_channels = hidden_channels
+
+            res_skip_layer = torch.nn.Conv1d(hidden_channels, res_skip_channels, 1)
+            res_skip_layer = torch.nn.utils.weight_norm(res_skip_layer, name="weight")
+            self.res_skip_layers.append(res_skip_layer)
+
+    def forward(self, x, x_mask, g=None, **kwargs):
+        output = torch.zeros_like(x)
+        n_channels_tensor = torch.IntTensor([self.hidden_channels])
+
+        if g is not None:
+            g = self.cond_layer(g)
+
+        for i in range(self.n_layers):
+            x_in = self.in_layers[i](x)
+            if g is not None:
+                cond_offset = i * 2 * self.hidden_channels
+                g_l = g[:, cond_offset : cond_offset + 2 * self.hidden_channels, :]
+            else:
+                g_l = torch.zeros_like(x_in)
+
+            acts = commons.fused_add_tanh_sigmoid_multiply(x_in, g_l, n_channels_tensor)
+            acts = self.drop(acts)
+
+            res_skip_acts = self.res_skip_layers[i](acts)
+            if i < self.n_layers - 1:
+                res_acts = res_skip_acts[:, : self.hidden_channels, :]
+                x = (x + res_acts) * x_mask
+                output = output + res_skip_acts[:, self.hidden_channels :, :]
+            else:
+                output = output + res_skip_acts
+        return output * x_mask
+
+    def remove_weight_norm(self):
+        if self.gin_channels != 0:
+            torch.nn.utils.remove_weight_norm(self.cond_layer)
+        for l in self.in_layers:
+            torch.nn.utils.remove_weight_norm(l)
+        for l in self.res_skip_layers:
+            torch.nn.utils.remove_weight_norm(l)
+
+
+class ResBlock1(torch.nn.Module):
+    def __init__(self, channels, kernel_size=3, dilation=(1, 3, 5)):
+        super(ResBlock1, self).__init__()
+        self.convs1 = nn.ModuleList(
+            [
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=dilation[0],
+                        padding=get_padding(kernel_size, dilation[0]),
+                    )
+                ),
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=dilation[1],
+                        padding=get_padding(kernel_size, dilation[1]),
+                    )
+                ),
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=dilation[2],
+                        padding=get_padding(kernel_size, dilation[2]),
+                    )
+                ),
+            ]
+        )
+        self.convs1.apply(init_weights)
+
+        self.convs2 = nn.ModuleList(
+            [
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=1,
+                        padding=get_padding(kernel_size, 1),
+                    )
+                ),
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=1,
+                        padding=get_padding(kernel_size, 1),
+                    )
+                ),
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=1,
+                        padding=get_padding(kernel_size, 1),
+                    )
+                ),
+            ]
+        )
+        self.convs2.apply(init_weights)
+
+    def forward(self, x, x_mask=None):
+        for c1, c2 in zip(self.convs1, self.convs2):
+            xt = F.leaky_relu(x, LRELU_SLOPE)
+            if x_mask is not None:
+                xt = xt * x_mask
+            xt = c1(xt)
+            xt = F.leaky_relu(xt, LRELU_SLOPE)
+            if x_mask is not None:
+                xt = xt * x_mask
+            xt = c2(xt)
+            x = xt + x
+        if x_mask is not None:
+            x = x * x_mask
+        return x
+
+    def remove_weight_norm(self):
+        for l in self.convs1:
+            remove_weight_norm(l)
+        for l in self.convs2:
+            remove_weight_norm(l)
+
+
+class ResBlock2(torch.nn.Module):
+    def __init__(self, channels, kernel_size=3, dilation=(1, 3)):
+        super(ResBlock2, self).__init__()
+        self.convs = nn.ModuleList(
+            [
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=dilation[0],
+                        padding=get_padding(kernel_size, dilation[0]),
+                    )
+                ),
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=dilation[1],
+                        padding=get_padding(kernel_size, dilation[1]),
+                    )
+                ),
+            ]
+        )
+        self.convs.apply(init_weights)
+
+    def forward(self, x, x_mask=None):
+        for c in self.convs:
+            xt = F.leaky_relu(x, LRELU_SLOPE)
+            if x_mask is not None:
+                xt = xt * x_mask
+            xt = c(xt)
+            x = xt + x
+        if x_mask is not None:
+            x = x * x_mask
+        return x
+
+    def remove_weight_norm(self):
+        for l in self.convs:
+            remove_weight_norm(l)
+
+
+class Log(nn.Module):
+    def forward(self, x, x_mask, reverse=False, **kwargs):
+        if not reverse:
+            y = torch.log(torch.clamp_min(x, 1e-5)) * x_mask
+            logdet = torch.sum(-y, [1, 2])
+            return y, logdet
+        else:
+            x = torch.exp(x) * x_mask
+            return x
+
+
+class Flip(nn.Module):
+    def forward(self, x, *args, reverse=False, **kwargs):
+        x = torch.flip(x, [1])
+        if not reverse:
+            logdet = torch.zeros(x.size(0)).to(dtype=x.dtype, device=x.device)
+            return x, logdet
+        else:
+            return x
+
+
+class ElementwiseAffine(nn.Module):
+    def __init__(self, channels):
+        super().__init__()
+        self.channels = channels
+        self.m = nn.Parameter(torch.zeros(channels, 1))
+        self.logs = nn.Parameter(torch.zeros(channels, 1))
+
+    def forward(self, x, x_mask, reverse=False, **kwargs):
+        if not reverse:
+            y = self.m + torch.exp(self.logs) * x
+            y = y * x_mask
+            logdet = torch.sum(self.logs * x_mask, [1, 2])
+            return y, logdet
+        else:
+            x = (x - self.m) * torch.exp(-self.logs) * x_mask
+            return x
+
+
+class ResidualCouplingLayer(nn.Module):
+    def __init__(
+        self,
+        channels,
+        hidden_channels,
+        kernel_size,
+        dilation_rate,
+        n_layers,
+        p_dropout=0,
+        gin_channels=0,
+        mean_only=False,
+    ):
+        assert channels % 2 == 0, "channels should be divisible by 2"
+        super().__init__()
+        self.channels = channels
+        self.hidden_channels = hidden_channels
+        self.kernel_size = kernel_size
+        self.dilation_rate = dilation_rate
+        self.n_layers = n_layers
+        self.half_channels = channels // 2
+        self.mean_only = mean_only
+
+        self.pre = nn.Conv1d(self.half_channels, hidden_channels, 1)
+        self.enc = WN(
+            hidden_channels,
+            kernel_size,
+            dilation_rate,
+            n_layers,
+            p_dropout=p_dropout,
+            gin_channels=gin_channels,
+        )
+        self.post = nn.Conv1d(hidden_channels, self.half_channels * (2 - mean_only), 1)
+        self.post.weight.data.zero_()
+        self.post.bias.data.zero_()
+
+    def forward(self, x, x_mask, g=None, reverse=False):
+        x0, x1 = torch.split(x, [self.half_channels] * 2, 1)
+        h = self.pre(x0) * x_mask
+        h = self.enc(h, x_mask, g=g)
+        stats = self.post(h) * x_mask
+        if not self.mean_only:
+            m, logs = torch.split(stats, [self.half_channels] * 2, 1)
+        else:
+            m = stats
+            logs = torch.zeros_like(m)
+
+        if not reverse:
+            x1 = m + x1 * torch.exp(logs) * x_mask
+            x = torch.cat([x0, x1], 1)
+            logdet = torch.sum(logs, [1, 2])
+            return x, logdet
+        else:
+            x1 = (x1 - m) * torch.exp(-logs) * x_mask
+            x = torch.cat([x0, x1], 1)
+            return x
+
+    def remove_weight_norm(self):
+        self.enc.remove_weight_norm()
+
+
+class ConvFlow(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        filter_channels,
+        kernel_size,
+        n_layers,
+        num_bins=10,
+        tail_bound=5.0,
+    ):
+        super().__init__()
+        self.in_channels = in_channels
+        self.filter_channels = filter_channels
+        self.kernel_size = kernel_size
+        self.n_layers = n_layers
+        self.num_bins = num_bins
+        self.tail_bound = tail_bound
+        self.half_channels = in_channels // 2
+
+        self.pre = nn.Conv1d(self.half_channels, filter_channels, 1)
+        self.convs = DDSConv(filter_channels, kernel_size, n_layers, p_dropout=0.0)
+        self.proj = nn.Conv1d(
+            filter_channels, self.half_channels * (num_bins * 3 - 1), 1
+        )
+        self.proj.weight.data.zero_()
+        self.proj.bias.data.zero_()
+
+    def forward(self, x, x_mask, g=None, reverse=False):
+        x0, x1 = torch.split(x, [self.half_channels] * 2, 1)
+        h = self.pre(x0)
+        h = self.convs(h, x_mask, g=g)
+        h = self.proj(h) * x_mask
+
+        b, c, t = x0.shape
+        h = h.reshape(b, c, -1, t).permute(0, 1, 3, 2)  # [b, cx?, t] -> [b, c, t, ?]
+
+        unnormalized_widths = h[..., : self.num_bins] / math.sqrt(self.filter_channels)
+        unnormalized_heights = h[..., self.num_bins : 2 * self.num_bins] / math.sqrt(
+            self.filter_channels
+        )
+        unnormalized_derivatives = h[..., 2 * self.num_bins :]
+
+        x1, logabsdet = piecewise_rational_quadratic_transform(
+            x1,
+            unnormalized_widths,
+            unnormalized_heights,
+            unnormalized_derivatives,
+            inverse=reverse,
+            tails="linear",
+            tail_bound=self.tail_bound,
+        )
+
+        x = torch.cat([x0, x1], 1) * x_mask
+        logdet = torch.sum(logabsdet * x_mask, [1, 2])
+        if not reverse:
+            return x, logdet
+        else:
+            return x

infer_pack/transforms.py

@@ -0,0 +1,193 @@
+import torch
+from torch.nn import functional as F
+
+import numpy as np
+
+
+DEFAULT_MIN_BIN_WIDTH = 1e-3
+DEFAULT_MIN_BIN_HEIGHT = 1e-3
+DEFAULT_MIN_DERIVATIVE = 1e-3
+
+
+def piecewise_rational_quadratic_transform(inputs, 
+                                           unnormalized_widths,
+                                           unnormalized_heights,
+                                           unnormalized_derivatives,
+                                           inverse=False,
+                                           tails=None, 
+                                           tail_bound=1.,
+                                           min_bin_width=DEFAULT_MIN_BIN_WIDTH,
+                                           min_bin_height=DEFAULT_MIN_BIN_HEIGHT,
+                                           min_derivative=DEFAULT_MIN_DERIVATIVE):
+
+    if tails is None:
+        spline_fn = rational_quadratic_spline
+        spline_kwargs = {}
+    else:
+        spline_fn = unconstrained_rational_quadratic_spline
+        spline_kwargs = {
+            'tails': tails,
+            'tail_bound': tail_bound
+        }
+
+    outputs, logabsdet = spline_fn(
+            inputs=inputs,
+            unnormalized_widths=unnormalized_widths,
+            unnormalized_heights=unnormalized_heights,
+            unnormalized_derivatives=unnormalized_derivatives,
+            inverse=inverse,
+            min_bin_width=min_bin_width,
+            min_bin_height=min_bin_height,
+            min_derivative=min_derivative,
+            **spline_kwargs
+    )
+    return outputs, logabsdet
+
+
+def searchsorted(bin_locations, inputs, eps=1e-6):
+    bin_locations[..., -1] += eps
+    return torch.sum(
+        inputs[..., None] >= bin_locations,
+        dim=-1
+    ) - 1
+
+
+def unconstrained_rational_quadratic_spline(inputs,
+                                            unnormalized_widths,
+                                            unnormalized_heights,
+                                            unnormalized_derivatives,
+                                            inverse=False,
+                                            tails='linear',
+                                            tail_bound=1.,
+                                            min_bin_width=DEFAULT_MIN_BIN_WIDTH,
+                                            min_bin_height=DEFAULT_MIN_BIN_HEIGHT,
+                                            min_derivative=DEFAULT_MIN_DERIVATIVE):
+    inside_interval_mask = (inputs >= -tail_bound) & (inputs <= tail_bound)
+    outside_interval_mask = ~inside_interval_mask
+
+    outputs = torch.zeros_like(inputs)
+    logabsdet = torch.zeros_like(inputs)
+
+    if tails == 'linear':
+        unnormalized_derivatives = F.pad(unnormalized_derivatives, pad=(1, 1))
+        constant = np.log(np.exp(1 - min_derivative) - 1)
+        unnormalized_derivatives[..., 0] = constant
+        unnormalized_derivatives[..., -1] = constant
+
+        outputs[outside_interval_mask] = inputs[outside_interval_mask]
+        logabsdet[outside_interval_mask] = 0
+    else:
+        raise RuntimeError('{} tails are not implemented.'.format(tails))
+
+    outputs[inside_interval_mask], logabsdet[inside_interval_mask] = rational_quadratic_spline(
+        inputs=inputs[inside_interval_mask],
+        unnormalized_widths=unnormalized_widths[inside_interval_mask, :],
+        unnormalized_heights=unnormalized_heights[inside_interval_mask, :],
+        unnormalized_derivatives=unnormalized_derivatives[inside_interval_mask, :],
+        inverse=inverse,
+        left=-tail_bound, right=tail_bound, bottom=-tail_bound, top=tail_bound,
+        min_bin_width=min_bin_width,
+        min_bin_height=min_bin_height,
+        min_derivative=min_derivative
+    )
+
+    return outputs, logabsdet
+
+def rational_quadratic_spline(inputs,
+                              unnormalized_widths,
+                              unnormalized_heights,
+                              unnormalized_derivatives,
+                              inverse=False,
+                              left=0., right=1., bottom=0., top=1.,
+                              min_bin_width=DEFAULT_MIN_BIN_WIDTH,
+                              min_bin_height=DEFAULT_MIN_BIN_HEIGHT,
+                              min_derivative=DEFAULT_MIN_DERIVATIVE):
+    if torch.min(inputs) < left or torch.max(inputs) > right:
+        raise ValueError('Input to a transform is not within its domain')
+
+    num_bins = unnormalized_widths.shape[-1]
+
+    if min_bin_width * num_bins > 1.0:
+        raise ValueError('Minimal bin width too large for the number of bins')
+    if min_bin_height * num_bins > 1.0:
+        raise ValueError('Minimal bin height too large for the number of bins')
+
+    widths = F.softmax(unnormalized_widths, dim=-1)
+    widths = min_bin_width + (1 - min_bin_width * num_bins) * widths
+    cumwidths = torch.cumsum(widths, dim=-1)
+    cumwidths = F.pad(cumwidths, pad=(1, 0), mode='constant', value=0.0)
+    cumwidths = (right - left) * cumwidths + left
+    cumwidths[..., 0] = left
+    cumwidths[..., -1] = right
+    widths = cumwidths[..., 1:] - cumwidths[..., :-1]
+
+    derivatives = min_derivative + F.softplus(unnormalized_derivatives)
+
+    heights = F.softmax(unnormalized_heights, dim=-1)
+    heights = min_bin_height + (1 - min_bin_height * num_bins) * heights
+    cumheights = torch.cumsum(heights, dim=-1)
+    cumheights = F.pad(cumheights, pad=(1, 0), mode='constant', value=0.0)
+    cumheights = (top - bottom) * cumheights + bottom
+    cumheights[..., 0] = bottom
+    cumheights[..., -1] = top
+    heights = cumheights[..., 1:] - cumheights[..., :-1]
+
+    if inverse:
+        bin_idx = searchsorted(cumheights, inputs)[..., None]
+    else:
+        bin_idx = searchsorted(cumwidths, inputs)[..., None]
+
+    input_cumwidths = cumwidths.gather(-1, bin_idx)[..., 0]
+    input_bin_widths = widths.gather(-1, bin_idx)[..., 0]
+
+    input_cumheights = cumheights.gather(-1, bin_idx)[..., 0]
+    delta = heights / widths
+    input_delta = delta.gather(-1, bin_idx)[..., 0]
+
+    input_derivatives = derivatives.gather(-1, bin_idx)[..., 0]
+    input_derivatives_plus_one = derivatives[..., 1:].gather(-1, bin_idx)[..., 0]
+
+    input_heights = heights.gather(-1, bin_idx)[..., 0]
+
+    if inverse:
+        a = (((inputs - input_cumheights) * (input_derivatives
+                                             + input_derivatives_plus_one
+                                             - 2 * input_delta)
+              + input_heights * (input_delta - input_derivatives)))
+        b = (input_heights * input_derivatives
+             - (inputs - input_cumheights) * (input_derivatives
+                                              + input_derivatives_plus_one
+                                              - 2 * input_delta))
+        c = - input_delta * (inputs - input_cumheights)
+
+        discriminant = b.pow(2) - 4 * a * c
+        assert (discriminant >= 0).all()
+
+        root = (2 * c) / (-b - torch.sqrt(discriminant))
+        outputs = root * input_bin_widths + input_cumwidths
+
+        theta_one_minus_theta = root * (1 - root)
+        denominator = input_delta + ((input_derivatives + input_derivatives_plus_one - 2 * input_delta)
+                                     * theta_one_minus_theta)
+        derivative_numerator = input_delta.pow(2) * (input_derivatives_plus_one * root.pow(2)
+                                                     + 2 * input_delta * theta_one_minus_theta
+                                                     + input_derivatives * (1 - root).pow(2))
+        logabsdet = torch.log(derivative_numerator) - 2 * torch.log(denominator)
+
+        return outputs, -logabsdet
+    else:
+        theta = (inputs - input_cumwidths) / input_bin_widths
+        theta_one_minus_theta = theta * (1 - theta)
+
+        numerator = input_heights * (input_delta * theta.pow(2)
+                                     + input_derivatives * theta_one_minus_theta)
+        denominator = input_delta + ((input_derivatives + input_derivatives_plus_one - 2 * input_delta)
+                                     * theta_one_minus_theta)
+        outputs = input_cumheights + numerator / denominator
+
+        derivative_numerator = input_delta.pow(2) * (input_derivatives_plus_one * theta.pow(2)
+                                                     + 2 * input_delta * theta_one_minus_theta
+                                                     + input_derivatives * (1 - theta).pow(2))
+        logabsdet = torch.log(derivative_numerator) - 2 * torch.log(denominator)
+
+        return outputs, logabsdet

infer_uvr5.py

@@ -0,0 +1,108 @@
+import os,sys,torch,warnings,pdb
+warnings.filterwarnings("ignore")
+import librosa
+import importlib
+import  numpy as np
+import hashlib , math
+from tqdm import tqdm
+from uvr5_pack.lib_v5 import spec_utils
+from uvr5_pack.utils import _get_name_params,inference
+from uvr5_pack.lib_v5.model_param_init import ModelParameters
+from scipy.io import wavfile
+
+class  _audio_pre_():
+    def __init__(self, model_path,device,is_half):
+        self.model_path = model_path
+        self.device = device
+        self.data = {
+            # Processing Options
+            'postprocess': False,
+            'tta': False,
+            # Constants
+            'window_size': 512,
+            'agg': 10,
+            'high_end_process': 'mirroring',
+        }
+        nn_arch_sizes = [
+            31191, # default
+            33966,61968, 123821, 123812, 537238 # custom
+        ]
+        self.nn_architecture = list('{}KB'.format(s) for s in nn_arch_sizes)
+        model_size = math.ceil(os.stat(model_path ).st_size / 1024)
+        nn_architecture = '{}KB'.format(min(nn_arch_sizes, key=lambda x:abs(x-model_size)))
+        nets = importlib.import_module('uvr5_pack.lib_v5.nets' + f'_{nn_architecture}'.replace('_{}KB'.format(nn_arch_sizes[0]), ''), package=None)
+        model_hash = hashlib.md5(open(model_path,'rb').read()).hexdigest()
+        param_name ,model_params_d = _get_name_params(model_path , model_hash)
+
+        mp = ModelParameters(model_params_d)
+        model = nets.CascadedASPPNet(mp.param['bins'] * 2)
+        cpk = torch.load( model_path , map_location='cpu')  
+        model.load_state_dict(cpk)
+        model.eval()
+        if(is_half==True):model = model.half().to(device)
+        else:model = model.to(device)
+
+        self.mp = mp
+        self.model = model
+
+    def _path_audio_(self, music_file ,ins_root=None,vocal_root=None):
+        if(ins_root is None and vocal_root is None):return "No save root."
+        name=os.path.basename(music_file)
+        if(ins_root is not None):os.makedirs(ins_root, exist_ok=True)
+        if(vocal_root is not None):os.makedirs(vocal_root , exist_ok=True)
+        X_wave, y_wave, X_spec_s, y_spec_s = {}, {}, {}, {}
+        bands_n = len(self.mp.param['band'])
+        # print(bands_n)
+        for d in range(bands_n, 0, -1): 
+            bp = self.mp.param['band'][d]
+            if d == bands_n: # high-end band
+                X_wave[d], _ = librosa.core.load(
+                    music_file, bp['sr'], False, dtype=np.float32, res_type=bp['res_type'])
+                if X_wave[d].ndim == 1:
+                    X_wave[d] = np.asfortranarray([X_wave[d], X_wave[d]])
+            else: # lower bands
+                X_wave[d] = librosa.core.resample(X_wave[d+1], self.mp.param['band'][d+1]['sr'], bp['sr'], res_type=bp['res_type'])
+            # Stft of wave source
+            X_spec_s[d] = spec_utils.wave_to_spectrogram_mt(X_wave[d], bp['hl'], bp['n_fft'], self.mp.param['mid_side'], self.mp.param['mid_side_b2'], self.mp.param['reverse'])
+            # pdb.set_trace()
+            if d == bands_n and self.data['high_end_process'] != 'none':
+                input_high_end_h = (bp['n_fft']//2 - bp['crop_stop']) + ( self.mp.param['pre_filter_stop'] - self.mp.param['pre_filter_start'])
+                input_high_end = X_spec_s[d][:, bp['n_fft']//2-input_high_end_h:bp['n_fft']//2, :]
+
+        X_spec_m = spec_utils.combine_spectrograms(X_spec_s, self.mp)
+        aggresive_set = float(self.data['agg']/100)
+        aggressiveness = {'value': aggresive_set, 'split_bin': self.mp.param['band'][1]['crop_stop']}
+        with torch.no_grad():
+            pred, X_mag, X_phase = inference(X_spec_m,self.device,self.model, aggressiveness,self.data)
+        # Postprocess
+        if self.data['postprocess']:
+            pred_inv = np.clip(X_mag - pred, 0, np.inf)
+            pred = spec_utils.mask_silence(pred, pred_inv)
+        y_spec_m = pred * X_phase
+        v_spec_m = X_spec_m - y_spec_m
+
+        if (ins_root is not None):
+            if self.data['high_end_process'].startswith('mirroring'):
+                input_high_end_ = spec_utils.mirroring(self.data['high_end_process'], y_spec_m, input_high_end, self.mp)
+                wav_instrument = spec_utils.cmb_spectrogram_to_wave(y_spec_m, self.mp,input_high_end_h, input_high_end_)
+            else:
+                wav_instrument = spec_utils.cmb_spectrogram_to_wave(y_spec_m, self.mp)
+            print ('%s instruments done'%name)
+            wavfile.write(os.path.join(ins_root, 'instrument_{}.wav'.format(name) ), self.mp.param['sr'], (np.array(wav_instrument)*32768).astype("int16"))  #
+        if (vocal_root is not None):
+            if self.data['high_end_process'].startswith('mirroring'):
+                input_high_end_ = spec_utils.mirroring(self.data['high_end_process'],  v_spec_m, input_high_end, self.mp)
+                wav_vocals = spec_utils.cmb_spectrogram_to_wave(v_spec_m, self.mp, input_high_end_h, input_high_end_)
+            else:
+                wav_vocals = spec_utils.cmb_spectrogram_to_wave(v_spec_m, self.mp)
+            print ('%s vocals done'%name)
+            wavfile.write(os.path.join(vocal_root , 'vocal_{}.wav'.format(name) ), self.mp.param['sr'], (np.array(wav_vocals)*32768).astype("int16"))
+
+if __name__ == '__main__':
+    device = 'cuda'
+    is_half=True
+    model_path='uvr5_weights/2_HP-UVR.pth'
+    pre_fun = _audio_pre_(model_path=model_path,device=device,is_half=True)
+    audio_path = '神女劈观.aac'
+    save_path = 'opt'
+    pre_fun._path_audio_(audio_path , save_path,save_path)

slicer.py

@@ -0,0 +1,151 @@
+import os.path
+from argparse import ArgumentParser
+import time
+
+import librosa
+import numpy as np
+import soundfile
+from scipy.ndimage import maximum_filter1d, uniform_filter1d
+
+
+def timeit(func):
+    def run(*args, **kwargs):
+        t = time.time()
+        res = func(*args, **kwargs)
+        print('executing \'%s\' costed %.3fs' % (func.__name__, time.time() - t))
+        return res
+    return run
+
+
+# @timeit
+def _window_maximum(arr, win_sz):
+    return maximum_filter1d(arr, size=win_sz)[win_sz // 2: win_sz // 2 + arr.shape[0] - win_sz + 1]
+
+
+# @timeit
+def _window_rms(arr, win_sz):
+    filtered = np.sqrt(uniform_filter1d(np.power(arr, 2), win_sz) - np.power(uniform_filter1d(arr, win_sz), 2))
+    return filtered[win_sz // 2: win_sz // 2 + arr.shape[0] - win_sz + 1]
+
+
+def level2db(levels, eps=1e-12):
+    return 20 * np.log10(np.clip(levels, a_min=eps, a_max=1))
+
+
+def _apply_slice(audio, begin, end):
+    if len(audio.shape) > 1:
+        return audio[:, begin: end]
+    else:
+        return audio[begin: end]
+
+
+class Slicer:
+    def __init__(self,
+                 sr: int,
+                 db_threshold: float = -40,
+                 min_length: int = 5000,
+                 win_l: int = 300,
+                 win_s: int = 20,
+                 max_silence_kept: int = 500):
+        self.db_threshold = db_threshold
+        self.min_samples = round(sr * min_length / 1000)
+        self.win_ln = round(sr * win_l / 1000)
+        self.win_sn = round(sr * win_s / 1000)
+        self.max_silence = round(sr * max_silence_kept / 1000)
+        if not self.min_samples >= self.win_ln >= self.win_sn:
+            raise ValueError('The following condition must be satisfied: min_length >= win_l >= win_s')
+        if not self.max_silence >= self.win_sn:
+            raise ValueError('The following condition must be satisfied: max_silence_kept >= win_s')
+
+    @timeit
+    def slice(self, audio):
+        if len(audio.shape) > 1:
+            samples = librosa.to_mono(audio)
+        else:
+            samples = audio
+        if samples.shape[0] <= self.min_samples:
+            return [audio]
+        # get absolute amplitudes
+        abs_amp = np.abs(samples - np.mean(samples))
+        # calculate local maximum with large window
+        win_max_db = level2db(_window_maximum(abs_amp, win_sz=self.win_ln))
+        sil_tags = []
+        left = right = 0
+        while right < win_max_db.shape[0]:
+            if win_max_db[right] < self.db_threshold:
+                right += 1
+            elif left == right:
+                left += 1
+                right += 1
+            else:
+                if left == 0:
+                    split_loc_l = left
+                else:
+                    sil_left_n = min(self.max_silence, (right + self.win_ln - left) // 2)
+                    rms_db_left = level2db(_window_rms(samples[left: left + sil_left_n], win_sz=self.win_sn))
+                    split_win_l = left + np.argmin(rms_db_left)
+                    split_loc_l = split_win_l + np.argmin(abs_amp[split_win_l: split_win_l + self.win_sn])
+                if len(sil_tags) != 0 and split_loc_l - sil_tags[-1][1] < self.min_samples and right < win_max_db.shape[0] - 1:
+                    right += 1
+                    left = right
+                    continue
+                if right == win_max_db.shape[0] - 1:
+                    split_loc_r = right + self.win_ln
+                else:
+                    sil_right_n = min(self.max_silence, (right + self.win_ln - left) // 2)
+                    rms_db_right = level2db(_window_rms(samples[right + self.win_ln - sil_right_n: right + self.win_ln], win_sz=self.win_sn))
+                    split_win_r = right + self.win_ln - sil_right_n + np.argmin(rms_db_right)
+                    split_loc_r = split_win_r + np.argmin(abs_amp[split_win_r: split_win_r + self.win_sn])
+                sil_tags.append((split_loc_l, split_loc_r))
+                right += 1
+                left = right
+        if left != right:
+            sil_left_n = min(self.max_silence, (right + self.win_ln - left) // 2)
+            rms_db_left = level2db(_window_rms(samples[left: left + sil_left_n], win_sz=self.win_sn))
+            split_win_l = left + np.argmin(rms_db_left)
+            split_loc_l = split_win_l + np.argmin(abs_amp[split_win_l: split_win_l + self.win_sn])
+            sil_tags.append((split_loc_l, samples.shape[0]))
+        if len(sil_tags) == 0:
+            return [audio]
+        else:
+            chunks = []
+            if sil_tags[0][0] > 0:
+                chunks.append(_apply_slice(audio, 0, sil_tags[0][0]))
+            for i in range(0, len(sil_tags) - 1):
+                chunks.append(_apply_slice(audio, sil_tags[i][1], sil_tags[i + 1][0]))
+            if sil_tags[-1][1] < samples.shape[0] - 1:
+                chunks.append(_apply_slice(audio, sil_tags[-1][1], samples.shape[0]))
+            return chunks
+
+
+def main():
+    parser = ArgumentParser()
+    parser.add_argument('audio', type=str, help='The audio to be sliced')
+    parser.add_argument('--out', type=str, help='Output directory of the sliced audio clips')
+    parser.add_argument('--db_thresh', type=float, required=False, default=-40, help='The dB threshold for silence detection')
+    parser.add_argument('--min_len', type=int, required=False, default=5000, help='The minimum milliseconds required for each sliced audio clip')
+    parser.add_argument('--win_l', type=int, required=False, default=300, help='Size of the large sliding window, presented in milliseconds')
+    parser.add_argument('--win_s', type=int, required=False, default=20, help='Size of the small sliding window, presented in milliseconds')
+    parser.add_argument('--max_sil_kept', type=int, required=False, default=500, help='The maximum silence length kept around the sliced audio, presented in milliseconds')
+    args = parser.parse_args()
+    out = args.out
+    if out is None:
+        out = os.path.dirname(os.path.abspath(args.audio))
+    audio, sr = librosa.load(args.audio, sr=None)
+    slicer = Slicer(
+        sr=sr,
+        db_threshold=args.db_thresh,
+        min_length=args.min_len,
+        win_l=args.win_l,
+        win_s=args.win_s,
+        max_silence_kept=args.max_sil_kept
+    )
+    chunks = slicer.slice(audio)
+    if not os.path.exists(args.out):
+        os.makedirs(args.out)
+    for i, chunk in enumerate(chunks):
+        soundfile.write(os.path.join(out, f'%s_%d.wav' % (os.path.basename(args.audio).rsplit('.', maxsplit=1)[0], i)), chunk, sr)
+
+
+if __name__ == '__main__':
+    main()

trainset_preprocess_pipeline.py

@@ -0,0 +1,63 @@
+import numpy as np,ffmpeg,os,traceback
+from slicer import Slicer
+slicer = Slicer(
+    sr=40000,
+    db_threshold=-32,
+    min_length=800,
+    win_l=400,
+    win_s=20,
+    max_silence_kept=150
+)
+
+
+
+
+def p0_load_audio(file, sr):#str-ing
+    try:
+        out, _ = (
+            ffmpeg.input(file, threads=0)
+            .output("-", format="s16le", acodec="pcm_s16le", ac=1, ar=sr)
+            .run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True)
+        )
+    except ffmpeg.Error as e:
+        raise RuntimeError(f"Failed to load audio: {e.stderr.decode()}") from e
+    return np.frombuffer(out, np.int16).flatten().astype(np.float32) / 32768.0
+
+def p1_trim_audio(slicer,audio):return slicer.slice(audio)
+
+def p2_avg_cut(audio,sr,per=3.7,overlap=0.3,tail=4):
+    i = 0
+    audios=[]
+    while (1):
+        start = int(sr * (per - overlap) * i)
+        i += 1
+        if (len(audio[start:]) > tail * sr):
+            audios.append(audio[start:start + int(per * sr)])
+        else:
+            audios.append(audio[start:])
+            break
+    return audios
+
+def p2b_get_vol(audio):return np.square(audio).mean()
+
+def p3_norm(audio,alpha=0.8,maxx=0.95):return audio / np.abs(audio).max() * (maxx * alpha) + (1-alpha) * audio
+
+def pipeline(inp_root,sr1=40000,sr2=16000,if_trim=True,if_avg_cut=True,if_norm=True,save_root1=None,save_root2=None):
+    if(save_root1==None and save_root2==None):return "No save root."
+    name2vol={}
+    infos=[]
+    names=[]
+    for name in os.listdir(inp_root):
+        try:
+            inp_path=os.path.join(inp_root,name)
+            audio=p0_load_audio(inp_path)
+        except:
+            infos.append("%s\t%s"%(name,traceback.format_exc()))
+            continue
+        if(if_trim==True):res1s=p1_trim_audio(audio)
+        else:res1s=[audio]
+        for i0,res1 in res1s:
+            if(if_avg_cut==True):res2=p2_avg_cut(res1)
+            else:res2=[res1]
+
+

uvr5_pack/lib_v5/dataset.py

@@ -0,0 +1,170 @@
+import os
+import random
+
+import numpy as np
+import torch
+import torch.utils.data
+from tqdm import tqdm
+
+from uvr5_pack.lib_v5 import spec_utils
+
+
+class VocalRemoverValidationSet(torch.utils.data.Dataset):
+
+    def __init__(self, patch_list):
+        self.patch_list = patch_list
+
+    def __len__(self):
+        return len(self.patch_list)
+
+    def __getitem__(self, idx):
+        path = self.patch_list[idx]
+        data = np.load(path)
+
+        X, y = data['X'], data['y']
+
+        X_mag = np.abs(X)
+        y_mag = np.abs(y)
+
+        return X_mag, y_mag
+
+
+def make_pair(mix_dir, inst_dir):
+    input_exts = ['.wav', '.m4a', '.mp3', '.mp4', '.flac']
+
+    X_list = sorted([
+        os.path.join(mix_dir, fname)
+        for fname in os.listdir(mix_dir)
+        if os.path.splitext(fname)[1] in input_exts])
+    y_list = sorted([
+        os.path.join(inst_dir, fname)
+        for fname in os.listdir(inst_dir)
+        if os.path.splitext(fname)[1] in input_exts])
+
+    filelist = list(zip(X_list, y_list))
+
+    return filelist
+
+
+def train_val_split(dataset_dir, split_mode, val_rate, val_filelist):
+    if split_mode == 'random':
+        filelist = make_pair(
+            os.path.join(dataset_dir, 'mixtures'),
+            os.path.join(dataset_dir, 'instruments'))
+
+        random.shuffle(filelist)
+
+        if len(val_filelist) == 0:
+            val_size = int(len(filelist) * val_rate)
+            train_filelist = filelist[:-val_size]
+            val_filelist = filelist[-val_size:]
+        else:
+            train_filelist = [
+                pair for pair in filelist
+                if list(pair) not in val_filelist]
+    elif split_mode == 'subdirs':
+        if len(val_filelist) != 0:
+            raise ValueError('The `val_filelist` option is not available in `subdirs` mode')
+
+        train_filelist = make_pair(
+            os.path.join(dataset_dir, 'training/mixtures'),
+            os.path.join(dataset_dir, 'training/instruments'))
+
+        val_filelist = make_pair(
+            os.path.join(dataset_dir, 'validation/mixtures'),
+            os.path.join(dataset_dir, 'validation/instruments'))
+
+    return train_filelist, val_filelist
+
+
+def augment(X, y, reduction_rate, reduction_mask, mixup_rate, mixup_alpha):
+    perm = np.random.permutation(len(X))
+    for i, idx in enumerate(tqdm(perm)):
+        if np.random.uniform() < reduction_rate:
+            y[idx] = spec_utils.reduce_vocal_aggressively(X[idx], y[idx], reduction_mask)
+
+        if np.random.uniform() < 0.5:
+            # swap channel
+            X[idx] = X[idx, ::-1]
+            y[idx] = y[idx, ::-1]
+        if np.random.uniform() < 0.02:
+            # mono
+            X[idx] = X[idx].mean(axis=0, keepdims=True)
+            y[idx] = y[idx].mean(axis=0, keepdims=True)
+        if np.random.uniform() < 0.02:
+            # inst
+            X[idx] = y[idx]
+
+        if np.random.uniform() < mixup_rate and i < len(perm) - 1:
+            lam = np.random.beta(mixup_alpha, mixup_alpha)
+            X[idx] = lam * X[idx] + (1 - lam) * X[perm[i + 1]]
+            y[idx] = lam * y[idx] + (1 - lam) * y[perm[i + 1]]
+
+    return X, y
+
+
+def make_padding(width, cropsize, offset):
+    left = offset
+    roi_size = cropsize - left * 2
+    if roi_size == 0:
+        roi_size = cropsize
+    right = roi_size - (width % roi_size) + left
+
+    return left, right, roi_size
+
+
+def make_training_set(filelist, cropsize, patches, sr, hop_length, n_fft, offset):
+    len_dataset = patches * len(filelist)
+
+    X_dataset = np.zeros(
+        (len_dataset, 2, n_fft // 2 + 1, cropsize), dtype=np.complex64)
+    y_dataset = np.zeros(
+        (len_dataset, 2, n_fft // 2 + 1, cropsize), dtype=np.complex64)
+
+    for i, (X_path, y_path) in enumerate(tqdm(filelist)):
+        X, y = spec_utils.cache_or_load(X_path, y_path, sr, hop_length, n_fft)
+        coef = np.max([np.abs(X).max(), np.abs(y).max()])
+        X, y = X / coef, y / coef
+
+        l, r, roi_size = make_padding(X.shape[2], cropsize, offset)
+        X_pad = np.pad(X, ((0, 0), (0, 0), (l, r)), mode='constant')
+        y_pad = np.pad(y, ((0, 0), (0, 0), (l, r)), mode='constant')
+
+        starts = np.random.randint(0, X_pad.shape[2] - cropsize, patches)
+        ends = starts + cropsize
+        for j in range(patches):
+            idx = i * patches + j
+            X_dataset[idx] = X_pad[:, :, starts[j]:ends[j]]
+            y_dataset[idx] = y_pad[:, :, starts[j]:ends[j]]
+
+    return X_dataset, y_dataset
+
+
+def make_validation_set(filelist, cropsize, sr, hop_length, n_fft, offset):
+    patch_list = []
+    patch_dir = 'cs{}_sr{}_hl{}_nf{}_of{}'.format(cropsize, sr, hop_length, n_fft, offset)
+    os.makedirs(patch_dir, exist_ok=True)
+
+    for i, (X_path, y_path) in enumerate(tqdm(filelist)):
+        basename = os.path.splitext(os.path.basename(X_path))[0]
+
+        X, y = spec_utils.cache_or_load(X_path, y_path, sr, hop_length, n_fft)
+        coef = np.max([np.abs(X).max(), np.abs(y).max()])
+        X, y = X / coef, y / coef
+
+        l, r, roi_size = make_padding(X.shape[2], cropsize, offset)
+        X_pad = np.pad(X, ((0, 0), (0, 0), (l, r)), mode='constant')
+        y_pad = np.pad(y, ((0, 0), (0, 0), (l, r)), mode='constant')
+
+        len_dataset = int(np.ceil(X.shape[2] / roi_size))
+        for j in range(len_dataset):
+            outpath = os.path.join(patch_dir, '{}_p{}.npz'.format(basename, j))
+            start = j * roi_size
+            if not os.path.exists(outpath):
+                np.savez(
+                    outpath,
+                    X=X_pad[:, :, start:start + cropsize],
+                    y=y_pad[:, :, start:start + cropsize])
+            patch_list.append(outpath)
+
+    return VocalRemoverValidationSet(patch_list)

uvr5_pack/lib_v5/layers.py

@@ -0,0 +1,116 @@
+import torch
+from torch import nn
+import torch.nn.functional as F
+
+from uvr5_pack.lib_v5 import spec_utils
+
+
+class Conv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(Conv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class SeperableConv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(SeperableConv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nin,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                groups=nin,
+                bias=False),
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=1,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class Encoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.LeakyReLU):
+        super(Encoder, self).__init__()
+        self.conv1 = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.conv2 = Conv2DBNActiv(nout, nout, ksize, stride, pad, activ=activ)
+
+    def __call__(self, x):
+        skip = self.conv1(x)
+        h = self.conv2(skip)
+
+        return h, skip
+
+
+class Decoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.ReLU, dropout=False):
+        super(Decoder, self).__init__()
+        self.conv = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.dropout = nn.Dropout2d(0.1) if dropout else None
+
+    def __call__(self, x, skip=None):
+        x = F.interpolate(x, scale_factor=2, mode='bilinear', align_corners=True)
+        if skip is not None:
+            skip = spec_utils.crop_center(skip, x)
+            x = torch.cat([x, skip], dim=1)
+        h = self.conv(x)
+
+        if self.dropout is not None:
+            h = self.dropout(h)
+
+        return h
+
+
+class ASPPModule(nn.Module):
+
+    def __init__(self, nin, nout, dilations=(4, 8, 16), activ=nn.ReLU):
+        super(ASPPModule, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.AdaptiveAvgPool2d((1, None)),
+            Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        )
+        self.conv2 = Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        self.conv3 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[0], dilations[0], activ=activ)
+        self.conv4 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[1], dilations[1], activ=activ)
+        self.conv5 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.bottleneck = nn.Sequential(
+            Conv2DBNActiv(nin * 5, nout, 1, 1, 0, activ=activ),
+            nn.Dropout2d(0.1)
+        )
+
+    def forward(self, x):
+        _, _, h, w = x.size()
+        feat1 = F.interpolate(self.conv1(x), size=(h, w), mode='bilinear', align_corners=True)
+        feat2 = self.conv2(x)
+        feat3 = self.conv3(x)
+        feat4 = self.conv4(x)
+        feat5 = self.conv5(x)
+        out = torch.cat((feat1, feat2, feat3, feat4, feat5), dim=1)
+        bottle = self.bottleneck(out)
+        return bottle

uvr5_pack/lib_v5/layers_123812KB .py

@@ -0,0 +1,116 @@
+import torch
+from torch import nn
+import torch.nn.functional as F
+
+from uvr5_pack.lib_v5 import spec_utils
+
+
+class Conv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(Conv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class SeperableConv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(SeperableConv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nin,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                groups=nin,
+                bias=False),
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=1,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class Encoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.LeakyReLU):
+        super(Encoder, self).__init__()
+        self.conv1 = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.conv2 = Conv2DBNActiv(nout, nout, ksize, stride, pad, activ=activ)
+
+    def __call__(self, x):
+        skip = self.conv1(x)
+        h = self.conv2(skip)
+
+        return h, skip
+
+
+class Decoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.ReLU, dropout=False):
+        super(Decoder, self).__init__()
+        self.conv = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.dropout = nn.Dropout2d(0.1) if dropout else None
+
+    def __call__(self, x, skip=None):
+        x = F.interpolate(x, scale_factor=2, mode='bilinear', align_corners=True)
+        if skip is not None:
+            skip = spec_utils.crop_center(skip, x)
+            x = torch.cat([x, skip], dim=1)
+        h = self.conv(x)
+
+        if self.dropout is not None:
+            h = self.dropout(h)
+
+        return h
+
+
+class ASPPModule(nn.Module):
+
+    def __init__(self, nin, nout, dilations=(4, 8, 16), activ=nn.ReLU):
+        super(ASPPModule, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.AdaptiveAvgPool2d((1, None)),
+            Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        )
+        self.conv2 = Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        self.conv3 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[0], dilations[0], activ=activ)
+        self.conv4 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[1], dilations[1], activ=activ)
+        self.conv5 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.bottleneck = nn.Sequential(
+            Conv2DBNActiv(nin * 5, nout, 1, 1, 0, activ=activ),
+            nn.Dropout2d(0.1)
+        )
+
+    def forward(self, x):
+        _, _, h, w = x.size()
+        feat1 = F.interpolate(self.conv1(x), size=(h, w), mode='bilinear', align_corners=True)
+        feat2 = self.conv2(x)
+        feat3 = self.conv3(x)
+        feat4 = self.conv4(x)
+        feat5 = self.conv5(x)
+        out = torch.cat((feat1, feat2, feat3, feat4, feat5), dim=1)
+        bottle = self.bottleneck(out)
+        return bottle

uvr5_pack/lib_v5/layers_123821KB.py

@@ -0,0 +1,116 @@
+import torch
+from torch import nn
+import torch.nn.functional as F
+
+from uvr5_pack.lib_v5 import spec_utils
+
+
+class Conv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(Conv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class SeperableConv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(SeperableConv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nin,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                groups=nin,
+                bias=False),
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=1,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class Encoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.LeakyReLU):
+        super(Encoder, self).__init__()
+        self.conv1 = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.conv2 = Conv2DBNActiv(nout, nout, ksize, stride, pad, activ=activ)
+
+    def __call__(self, x):
+        skip = self.conv1(x)
+        h = self.conv2(skip)
+
+        return h, skip
+
+
+class Decoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.ReLU, dropout=False):
+        super(Decoder, self).__init__()
+        self.conv = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.dropout = nn.Dropout2d(0.1) if dropout else None
+
+    def __call__(self, x, skip=None):
+        x = F.interpolate(x, scale_factor=2, mode='bilinear', align_corners=True)
+        if skip is not None:
+            skip = spec_utils.crop_center(skip, x)
+            x = torch.cat([x, skip], dim=1)
+        h = self.conv(x)
+
+        if self.dropout is not None:
+            h = self.dropout(h)
+
+        return h
+
+
+class ASPPModule(nn.Module):
+
+    def __init__(self, nin, nout, dilations=(4, 8, 16), activ=nn.ReLU):
+        super(ASPPModule, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.AdaptiveAvgPool2d((1, None)),
+            Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        )
+        self.conv2 = Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        self.conv3 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[0], dilations[0], activ=activ)
+        self.conv4 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[1], dilations[1], activ=activ)
+        self.conv5 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.bottleneck = nn.Sequential(
+            Conv2DBNActiv(nin * 5, nout, 1, 1, 0, activ=activ),
+            nn.Dropout2d(0.1)
+        )
+
+    def forward(self, x):
+        _, _, h, w = x.size()
+        feat1 = F.interpolate(self.conv1(x), size=(h, w), mode='bilinear', align_corners=True)
+        feat2 = self.conv2(x)
+        feat3 = self.conv3(x)
+        feat4 = self.conv4(x)
+        feat5 = self.conv5(x)
+        out = torch.cat((feat1, feat2, feat3, feat4, feat5), dim=1)
+        bottle = self.bottleneck(out)
+        return bottle

uvr5_pack/lib_v5/layers_33966KB.py

@@ -0,0 +1,122 @@
+import torch
+from torch import nn
+import torch.nn.functional as F
+
+from uvr5_pack.lib_v5 import spec_utils
+
+
+class Conv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(Conv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class SeperableConv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(SeperableConv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nin,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                groups=nin,
+                bias=False),
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=1,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class Encoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.LeakyReLU):
+        super(Encoder, self).__init__()
+        self.conv1 = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.conv2 = Conv2DBNActiv(nout, nout, ksize, stride, pad, activ=activ)
+
+    def __call__(self, x):
+        skip = self.conv1(x)
+        h = self.conv2(skip)
+
+        return h, skip
+
+
+class Decoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.ReLU, dropout=False):
+        super(Decoder, self).__init__()
+        self.conv = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.dropout = nn.Dropout2d(0.1) if dropout else None
+
+    def __call__(self, x, skip=None):
+        x = F.interpolate(x, scale_factor=2, mode='bilinear', align_corners=True)
+        if skip is not None:
+            skip = spec_utils.crop_center(skip, x)
+            x = torch.cat([x, skip], dim=1)
+        h = self.conv(x)
+
+        if self.dropout is not None:
+            h = self.dropout(h)
+
+        return h
+
+
+class ASPPModule(nn.Module):
+
+    def __init__(self, nin, nout, dilations=(4, 8, 16, 32, 64), activ=nn.ReLU):
+        super(ASPPModule, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.AdaptiveAvgPool2d((1, None)),
+            Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        )
+        self.conv2 = Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        self.conv3 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[0], dilations[0], activ=activ)
+        self.conv4 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[1], dilations[1], activ=activ)
+        self.conv5 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.conv6 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.conv7 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.bottleneck = nn.Sequential(
+            Conv2DBNActiv(nin * 7, nout, 1, 1, 0, activ=activ),
+            nn.Dropout2d(0.1)
+        )
+
+    def forward(self, x):
+        _, _, h, w = x.size()
+        feat1 = F.interpolate(self.conv1(x), size=(h, w), mode='bilinear', align_corners=True)
+        feat2 = self.conv2(x)
+        feat3 = self.conv3(x)
+        feat4 = self.conv4(x)
+        feat5 = self.conv5(x)
+        feat6 = self.conv6(x)
+        feat7 = self.conv7(x)
+        out = torch.cat((feat1, feat2, feat3, feat4, feat5, feat6, feat7), dim=1)
+        bottle = self.bottleneck(out)
+        return bottle

uvr5_pack/lib_v5/layers_537227KB.py

@@ -0,0 +1,122 @@
+import torch
+from torch import nn
+import torch.nn.functional as F
+
+from uvr5_pack.lib_v5 import spec_utils
+
+
+class Conv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(Conv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class SeperableConv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(SeperableConv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nin,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                groups=nin,
+                bias=False),
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=1,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class Encoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.LeakyReLU):
+        super(Encoder, self).__init__()
+        self.conv1 = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.conv2 = Conv2DBNActiv(nout, nout, ksize, stride, pad, activ=activ)
+
+    def __call__(self, x):
+        skip = self.conv1(x)
+        h = self.conv2(skip)
+
+        return h, skip
+
+
+class Decoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.ReLU, dropout=False):
+        super(Decoder, self).__init__()
+        self.conv = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.dropout = nn.Dropout2d(0.1) if dropout else None
+
+    def __call__(self, x, skip=None):
+        x = F.interpolate(x, scale_factor=2, mode='bilinear', align_corners=True)
+        if skip is not None:
+            skip = spec_utils.crop_center(skip, x)
+            x = torch.cat([x, skip], dim=1)
+        h = self.conv(x)
+
+        if self.dropout is not None:
+            h = self.dropout(h)
+
+        return h
+
+
+class ASPPModule(nn.Module):
+
+    def __init__(self, nin, nout, dilations=(4, 8, 16, 32, 64), activ=nn.ReLU):
+        super(ASPPModule, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.AdaptiveAvgPool2d((1, None)),
+            Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        )
+        self.conv2 = Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        self.conv3 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[0], dilations[0], activ=activ)
+        self.conv4 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[1], dilations[1], activ=activ)
+        self.conv5 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.conv6 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.conv7 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.bottleneck = nn.Sequential(
+            Conv2DBNActiv(nin * 7, nout, 1, 1, 0, activ=activ),
+            nn.Dropout2d(0.1)
+        )
+
+    def forward(self, x):
+        _, _, h, w = x.size()
+        feat1 = F.interpolate(self.conv1(x), size=(h, w), mode='bilinear', align_corners=True)
+        feat2 = self.conv2(x)
+        feat3 = self.conv3(x)
+        feat4 = self.conv4(x)
+        feat5 = self.conv5(x)
+        feat6 = self.conv6(x)
+        feat7 = self.conv7(x)
+        out = torch.cat((feat1, feat2, feat3, feat4, feat5, feat6, feat7), dim=1)
+        bottle = self.bottleneck(out)
+        return bottle

uvr5_pack/lib_v5/layers_537238KB.py

@@ -0,0 +1,122 @@
+import torch
+from torch import nn
+import torch.nn.functional as F
+
+from uvr5_pack.lib_v5 import spec_utils
+
+
+class Conv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(Conv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class SeperableConv2DBNActiv(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
+        super(SeperableConv2DBNActiv, self).__init__()
+        self.conv = nn.Sequential(
+            nn.Conv2d(
+                nin, nin,
+                kernel_size=ksize,
+                stride=stride,
+                padding=pad,
+                dilation=dilation,
+                groups=nin,
+                bias=False),
+            nn.Conv2d(
+                nin, nout,
+                kernel_size=1,
+                bias=False),
+            nn.BatchNorm2d(nout),
+            activ()
+        )
+
+    def __call__(self, x):
+        return self.conv(x)
+
+
+class Encoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.LeakyReLU):
+        super(Encoder, self).__init__()
+        self.conv1 = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.conv2 = Conv2DBNActiv(nout, nout, ksize, stride, pad, activ=activ)
+
+    def __call__(self, x):
+        skip = self.conv1(x)
+        h = self.conv2(skip)
+
+        return h, skip
+
+
+class Decoder(nn.Module):
+
+    def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.ReLU, dropout=False):
+        super(Decoder, self).__init__()
+        self.conv = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
+        self.dropout = nn.Dropout2d(0.1) if dropout else None
+
+    def __call__(self, x, skip=None):
+        x = F.interpolate(x, scale_factor=2, mode='bilinear', align_corners=True)
+        if skip is not None:
+            skip = spec_utils.crop_center(skip, x)
+            x = torch.cat([x, skip], dim=1)
+        h = self.conv(x)
+
+        if self.dropout is not None:
+            h = self.dropout(h)
+
+        return h
+
+
+class ASPPModule(nn.Module):
+
+    def __init__(self, nin, nout, dilations=(4, 8, 16, 32, 64), activ=nn.ReLU):
+        super(ASPPModule, self).__init__()
+        self.conv1 = nn.Sequential(
+            nn.AdaptiveAvgPool2d((1, None)),
+            Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        )
+        self.conv2 = Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
+        self.conv3 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[0], dilations[0], activ=activ)
+        self.conv4 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[1], dilations[1], activ=activ)
+        self.conv5 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.conv6 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.conv7 = SeperableConv2DBNActiv(
+            nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
+        self.bottleneck = nn.Sequential(
+            Conv2DBNActiv(nin * 7, nout, 1, 1, 0, activ=activ),
+            nn.Dropout2d(0.1)
+        )
+
+    def forward(self, x):
+        _, _, h, w = x.size()
+        feat1 = F.interpolate(self.conv1(x), size=(h, w), mode='bilinear', align_corners=True)
+        feat2 = self.conv2(x)
+        feat3 = self.conv3(x)
+        feat4 = self.conv4(x)
+        feat5 = self.conv5(x)
+        feat6 = self.conv6(x)
+        feat7 = self.conv7(x)
+        out = torch.cat((feat1, feat2, feat3, feat4, feat5, feat6, feat7), dim=1)
+        bottle = self.bottleneck(out)
+        return bottle

uvr5_pack/lib_v5/model_param_init.py

@@ -0,0 +1,60 @@
+import json
+import os
+import pathlib
+
+default_param = {}
+default_param['bins'] = 768
+default_param['unstable_bins'] = 9 # training only
+default_param['reduction_bins'] = 762 # training only
+default_param['sr'] = 44100
+default_param['pre_filter_start'] = 757
+default_param['pre_filter_stop'] = 768
+default_param['band'] = {}
+
+
+default_param['band'][1] = {
+    'sr': 11025,
+    'hl': 128,
+    'n_fft': 960,
+    'crop_start': 0,
+    'crop_stop': 245,
+    'lpf_start': 61, # inference only
+    'res_type': 'polyphase'
+}
+
+default_param['band'][2] = {
+    'sr': 44100,
+    'hl': 512,
+    'n_fft': 1536,
+    'crop_start': 24,
+    'crop_stop': 547,
+    'hpf_start': 81, # inference only
+    'res_type': 'sinc_best'
+}
+
+
+def int_keys(d):
+    r = {}
+    for k, v in d:
+        if k.isdigit():
+            k = int(k)
+        r[k] = v
+    return r
+    
+
+class ModelParameters(object):
+    def __init__(self, config_path=''):
+        if '.pth' == pathlib.Path(config_path).suffix:
+            import zipfile
+            
+            with zipfile.ZipFile(config_path, 'r') as zip:
+                self.param = json.loads(zip.read('param.json'), object_pairs_hook=int_keys)
+        elif '.json' == pathlib.Path(config_path).suffix:
+            with open(config_path, 'r') as f:
+                self.param = json.loads(f.read(), object_pairs_hook=int_keys)
+        else:
+            self.param = default_param
+            
+        for k in ['mid_side', 'mid_side_b', 'mid_side_b2', 'stereo_w', 'stereo_n', 'reverse']:
+            if not k in self.param:
+                self.param[k] = False

uvr5_pack/lib_v5/modelparams/1band_sr16000_hl512.json

@@ -0,0 +1,19 @@
+{
+	"bins": 1024,
+	"unstable_bins": 0,
+	"reduction_bins": 0,
+	"band": {
+		"1": {
+			"sr": 16000,
+			"hl": 512,
+			"n_fft": 2048,
+			"crop_start": 0,
+			"crop_stop": 1024,
+			"hpf_start": -1,
+			"res_type": "sinc_best"
+		}
+	},
+	"sr": 16000,
+	"pre_filter_start": 1023,
+	"pre_filter_stop": 1024
+}

uvr5_pack/lib_v5/modelparams/1band_sr32000_hl512.json

@@ -0,0 +1,19 @@
+{
+	"bins": 1024,
+	"unstable_bins": 0,
+	"reduction_bins": 0,
+	"band": {
+		"1": {
+			"sr": 32000,
+			"hl": 512,
+			"n_fft": 2048,
+			"crop_start": 0,
+			"crop_stop": 1024,
+			"hpf_start": -1,
+			"res_type": "kaiser_fast"
+		}
+	},
+	"sr": 32000,
+	"pre_filter_start": 1000,
+	"pre_filter_stop": 1021
+}

uvr5_pack/lib_v5/modelparams/1band_sr33075_hl384.json

@@ -0,0 +1,19 @@
+{
+	"bins": 1024,
+	"unstable_bins": 0,
+	"reduction_bins": 0,
+	"band": {
+		"1": {
+			"sr": 33075,
+			"hl": 384,
+			"n_fft": 2048,
+			"crop_start": 0,
+			"crop_stop": 1024,
+			"hpf_start": -1,
+			"res_type": "sinc_best"
+		}
+	},
+	"sr": 33075,
+	"pre_filter_start": 1000,
+	"pre_filter_stop": 1021
+}

uvr5_pack/lib_v5/modelparams/1band_sr44100_hl1024.json

@@ -0,0 +1,19 @@
+{
+	"bins": 1024,
+	"unstable_bins": 0,
+	"reduction_bins": 0,
+	"band": {
+		"1": {
+			"sr": 44100,
+			"hl": 1024,
+			"n_fft": 2048,
+			"crop_start": 0,
+			"crop_stop": 1024,
+			"hpf_start": -1,
+			"res_type": "sinc_best"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 1023,
+	"pre_filter_stop": 1024
+}

uvr5_pack/lib_v5/modelparams/1band_sr44100_hl256.json

@@ -0,0 +1,19 @@
+{
+	"bins": 256,
+	"unstable_bins": 0,
+	"reduction_bins": 0,
+	"band": {
+		"1": {
+			"sr": 44100,
+			"hl": 256,
+			"n_fft": 512,
+			"crop_start": 0,
+			"crop_stop": 256,
+			"hpf_start": -1,
+			"res_type": "sinc_best"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 256,
+	"pre_filter_stop": 256
+}

uvr5_pack/lib_v5/modelparams/1band_sr44100_hl512.json

@@ -0,0 +1,19 @@
+{
+	"bins": 1024,
+	"unstable_bins": 0,
+	"reduction_bins": 0,
+	"band": {
+		"1": {
+			"sr": 44100,
+			"hl": 512,
+			"n_fft": 2048,
+			"crop_start": 0,
+			"crop_stop": 1024,
+			"hpf_start": -1,
+			"res_type": "sinc_best"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 1023,
+	"pre_filter_stop": 1024
+}

uvr5_pack/lib_v5/modelparams/1band_sr44100_hl512_cut.json

@@ -0,0 +1,19 @@
+{
+	"bins": 1024,
+	"unstable_bins": 0,
+	"reduction_bins": 0,
+	"band": {
+		"1": {
+			"sr": 44100,
+			"hl": 512,
+			"n_fft": 2048,
+			"crop_start": 0,
+			"crop_stop": 700,
+			"hpf_start": -1,
+			"res_type": "sinc_best"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 1023,
+	"pre_filter_stop": 700
+}

uvr5_pack/lib_v5/modelparams/2band_32000.json

@@ -0,0 +1,30 @@
+{
+	"bins": 768,
+	"unstable_bins": 7,
+	"reduction_bins": 705,
+	"band": {
+		"1": {
+			"sr": 6000,
+			"hl": 66,
+			"n_fft": 512,
+			"crop_start": 0,
+			"crop_stop": 240,
+			"lpf_start": 60,
+			"lpf_stop": 118,
+			"res_type": "sinc_fastest"
+		},
+		"2": {
+			"sr": 32000,
+			"hl": 352,
+			"n_fft": 1024,
+			"crop_start": 22,
+			"crop_stop": 505,
+			"hpf_start": 44,
+			"hpf_stop": 23,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 32000,
+	"pre_filter_start": 710,
+	"pre_filter_stop": 731
+}

uvr5_pack/lib_v5/modelparams/2band_44100_lofi.json

@@ -0,0 +1,30 @@
+{
+	"bins": 512,
+	"unstable_bins": 7,
+	"reduction_bins": 510,
+	"band": {
+		"1": {
+			"sr": 11025,
+			"hl": 160,
+			"n_fft": 768,
+			"crop_start": 0,
+			"crop_stop": 192,
+			"lpf_start": 41,
+			"lpf_stop": 139,
+			"res_type": "sinc_fastest"
+		},
+		"2": {
+			"sr": 44100,
+			"hl": 640,
+			"n_fft": 1024,
+			"crop_start": 10,
+			"crop_stop": 320,
+			"hpf_start": 47,
+			"hpf_stop": 15,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 510,
+	"pre_filter_stop": 512
+}

uvr5_pack/lib_v5/modelparams/2band_48000.json

@@ -0,0 +1,30 @@
+{
+	"bins": 768,
+	"unstable_bins": 7,
+	"reduction_bins": 705,
+	"band": {
+		"1": {
+			"sr": 6000,
+			"hl": 66,
+			"n_fft": 512,
+			"crop_start": 0,
+			"crop_stop": 240,
+			"lpf_start": 60,
+			"lpf_stop": 240,
+			"res_type": "sinc_fastest"
+		},
+		"2": {
+			"sr": 48000,
+			"hl": 528,
+			"n_fft": 1536,
+			"crop_start": 22,
+			"crop_stop": 505,
+			"hpf_start": 82,
+			"hpf_stop": 22,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 48000,
+	"pre_filter_start": 710,
+	"pre_filter_stop": 731
+}

uvr5_pack/lib_v5/modelparams/3band_44100.json

@@ -0,0 +1,42 @@
+{
+	"bins": 768,
+	"unstable_bins": 5,
+	"reduction_bins": 733,
+	"band": {
+		"1": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 768,
+			"crop_start": 0,
+			"crop_stop": 278,
+			"lpf_start": 28,
+			"lpf_stop": 140,
+			"res_type": "polyphase"
+		},
+		"2": {
+			"sr": 22050,
+			"hl": 256,
+			"n_fft": 768,
+			"crop_start": 14,
+			"crop_stop": 322,
+			"hpf_start": 70,
+			"hpf_stop": 14,
+			"lpf_start": 283,
+			"lpf_stop": 314,
+			"res_type": "polyphase"
+		},	
+		"3": {
+			"sr": 44100,
+			"hl": 512,
+			"n_fft": 768,
+			"crop_start": 131,
+			"crop_stop": 313,
+			"hpf_start": 154,
+			"hpf_stop": 141,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 757,
+	"pre_filter_stop": 768
+}

uvr5_pack/lib_v5/modelparams/3band_44100_mid.json

@@ -0,0 +1,43 @@
+{
+	"mid_side": true,
+	"bins": 768,
+	"unstable_bins": 5,
+	"reduction_bins": 733,
+	"band": {
+		"1": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 768,
+			"crop_start": 0,
+			"crop_stop": 278,
+			"lpf_start": 28,
+			"lpf_stop": 140,
+			"res_type": "polyphase"
+		},
+		"2": {
+			"sr": 22050,
+			"hl": 256,
+			"n_fft": 768,
+			"crop_start": 14,
+			"crop_stop": 322,
+			"hpf_start": 70,
+			"hpf_stop": 14,
+			"lpf_start": 283,
+			"lpf_stop": 314,
+			"res_type": "polyphase"
+		},	
+		"3": {
+			"sr": 44100,
+			"hl": 512,
+			"n_fft": 768,
+			"crop_start": 131,
+			"crop_stop": 313,
+			"hpf_start": 154,
+			"hpf_stop": 141,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 757,
+	"pre_filter_stop": 768
+}

uvr5_pack/lib_v5/modelparams/3band_44100_msb2.json

@@ -0,0 +1,43 @@
+{
+	"mid_side_b2": true,
+	"bins": 640,
+	"unstable_bins": 7,
+	"reduction_bins": 565,
+	"band": {
+		"1": {
+			"sr": 11025,
+			"hl": 108,
+			"n_fft": 1024,
+			"crop_start": 0,
+			"crop_stop": 187,
+			"lpf_start": 92,
+			"lpf_stop": 186,
+			"res_type": "polyphase"
+		},
+		"2": {
+			"sr": 22050,
+			"hl": 216,
+			"n_fft": 768,
+			"crop_start": 0,
+			"crop_stop": 212,
+			"hpf_start": 68,
+			"hpf_stop": 34,
+			"lpf_start": 174,
+			"lpf_stop": 209,
+			"res_type": "polyphase"
+		},	
+		"3": {
+			"sr": 44100,
+			"hl": 432,
+			"n_fft": 640,
+			"crop_start": 66,
+			"crop_stop": 307,
+			"hpf_start": 86,
+			"hpf_stop": 72,
+			"res_type": "kaiser_fast"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 639,
+	"pre_filter_stop": 640
+}

uvr5_pack/lib_v5/modelparams/4band_44100.json

@@ -0,0 +1,54 @@
+{
+	"bins": 768,
+	"unstable_bins": 7,
+	"reduction_bins": 668,
+	"band": {
+		"1": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 1024,
+			"crop_start": 0,
+			"crop_stop": 186,
+			"lpf_start": 37,
+			"lpf_stop": 73,
+			"res_type": "polyphase"
+		},
+		"2": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 512,
+			"crop_start": 4,
+			"crop_stop": 185,			
+			"hpf_start": 36,
+			"hpf_stop": 18,
+			"lpf_start": 93,
+			"lpf_stop": 185,
+			"res_type": "polyphase"
+		},
+		"3": {
+			"sr": 22050,
+			"hl": 256,
+			"n_fft": 512,
+			"crop_start": 46,
+			"crop_stop": 186,
+			"hpf_start": 93,
+			"hpf_stop": 46,
+			"lpf_start": 164,
+			"lpf_stop": 186,
+			"res_type": "polyphase"
+		},	
+		"4": {
+			"sr": 44100,
+			"hl": 512,
+			"n_fft": 768,
+			"crop_start": 121,
+			"crop_stop": 382,
+			"hpf_start": 138,
+			"hpf_stop": 123,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 740,
+	"pre_filter_stop": 768
+}

uvr5_pack/lib_v5/modelparams/4band_44100_mid.json

@@ -0,0 +1,55 @@
+{
+	"bins": 768,
+	"unstable_bins": 7,
+	"mid_side": true,
+	"reduction_bins": 668,
+	"band": {
+		"1": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 1024,
+			"crop_start": 0,
+			"crop_stop": 186,
+			"lpf_start": 37,
+			"lpf_stop": 73,
+			"res_type": "polyphase"
+		},
+		"2": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 512,
+			"crop_start": 4,
+			"crop_stop": 185,			
+			"hpf_start": 36,
+			"hpf_stop": 18,
+			"lpf_start": 93,
+			"lpf_stop": 185,
+			"res_type": "polyphase"
+		},
+		"3": {
+			"sr": 22050,
+			"hl": 256,
+			"n_fft": 512,
+			"crop_start": 46,
+			"crop_stop": 186,
+			"hpf_start": 93,
+			"hpf_stop": 46,
+			"lpf_start": 164,
+			"lpf_stop": 186,
+			"res_type": "polyphase"
+		},	
+		"4": {
+			"sr": 44100,
+			"hl": 512,
+			"n_fft": 768,
+			"crop_start": 121,
+			"crop_stop": 382,
+			"hpf_start": 138,
+			"hpf_stop": 123,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 740,
+	"pre_filter_stop": 768
+}

uvr5_pack/lib_v5/modelparams/4band_44100_msb.json

@@ -0,0 +1,55 @@
+{
+	"mid_side_b": true,
+	"bins": 768,
+	"unstable_bins": 7,
+	"reduction_bins": 668,
+	"band": {
+		"1": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 1024,
+			"crop_start": 0,
+			"crop_stop": 186,
+			"lpf_start": 37,
+			"lpf_stop": 73,
+			"res_type": "polyphase"
+		},
+		"2": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 512,
+			"crop_start": 4,
+			"crop_stop": 185,			
+			"hpf_start": 36,
+			"hpf_stop": 18,
+			"lpf_start": 93,
+			"lpf_stop": 185,
+			"res_type": "polyphase"
+		},
+		"3": {
+			"sr": 22050,
+			"hl": 256,
+			"n_fft": 512,
+			"crop_start": 46,
+			"crop_stop": 186,
+			"hpf_start": 93,
+			"hpf_stop": 46,
+			"lpf_start": 164,
+			"lpf_stop": 186,
+			"res_type": "polyphase"
+		},	
+		"4": {
+			"sr": 44100,
+			"hl": 512,
+			"n_fft": 768,
+			"crop_start": 121,
+			"crop_stop": 382,
+			"hpf_start": 138,
+			"hpf_stop": 123,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 740,
+	"pre_filter_stop": 768
+}

uvr5_pack/lib_v5/modelparams/4band_44100_msb2.json

@@ -0,0 +1,55 @@
+{
+	"mid_side_b": true,
+	"bins": 768,
+	"unstable_bins": 7,
+	"reduction_bins": 668,
+	"band": {
+		"1": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 1024,
+			"crop_start": 0,
+			"crop_stop": 186,
+			"lpf_start": 37,
+			"lpf_stop": 73,
+			"res_type": "polyphase"
+		},
+		"2": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 512,
+			"crop_start": 4,
+			"crop_stop": 185,			
+			"hpf_start": 36,
+			"hpf_stop": 18,
+			"lpf_start": 93,
+			"lpf_stop": 185,
+			"res_type": "polyphase"
+		},
+		"3": {
+			"sr": 22050,
+			"hl": 256,
+			"n_fft": 512,
+			"crop_start": 46,
+			"crop_stop": 186,
+			"hpf_start": 93,
+			"hpf_stop": 46,
+			"lpf_start": 164,
+			"lpf_stop": 186,
+			"res_type": "polyphase"
+		},	
+		"4": {
+			"sr": 44100,
+			"hl": 512,
+			"n_fft": 768,
+			"crop_start": 121,
+			"crop_stop": 382,
+			"hpf_start": 138,
+			"hpf_stop": 123,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 740,
+	"pre_filter_stop": 768
+}

uvr5_pack/lib_v5/modelparams/4band_44100_reverse.json

@@ -0,0 +1,55 @@
+{
+	"reverse": true,
+	"bins": 768,
+	"unstable_bins": 7,
+	"reduction_bins": 668,
+	"band": {
+		"1": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 1024,
+			"crop_start": 0,
+			"crop_stop": 186,
+			"lpf_start": 37,
+			"lpf_stop": 73,
+			"res_type": "polyphase"
+		},
+		"2": {
+			"sr": 11025,
+			"hl": 128,
+			"n_fft": 512,
+			"crop_start": 4,
+			"crop_stop": 185,			
+			"hpf_start": 36,
+			"hpf_stop": 18,
+			"lpf_start": 93,
+			"lpf_stop": 185,
+			"res_type": "polyphase"
+		},
+		"3": {
+			"sr": 22050,
+			"hl": 256,
+			"n_fft": 512,
+			"crop_start": 46,
+			"crop_stop": 186,
+			"hpf_start": 93,
+			"hpf_stop": 46,
+			"lpf_start": 164,
+			"lpf_stop": 186,
+			"res_type": "polyphase"
+		},	
+		"4": {
+			"sr": 44100,
+			"hl": 512,
+			"n_fft": 768,
+			"crop_start": 121,
+			"crop_stop": 382,
+			"hpf_start": 138,
+			"hpf_stop": 123,
+			"res_type": "sinc_medium"
+		}
+	},
+	"sr": 44100,
+	"pre_filter_start": 740,
+	"pre_filter_stop": 768
+}