import io import os #os.system("wget -P hubert/ https://huggingface.co/spaces/innnky/nanami/resolve/main/checkpoint_best_legacy_500.pt") import gradio as gr import librosa import numpy as np import soundfile from inference.infer_tool import Svc import logging logging.getLogger('numba').setLevel(logging.WARNING) logging.getLogger('markdown_it').setLevel(logging.WARNING) logging.getLogger('urllib3').setLevel(logging.WARNING) logging.getLogger('matplotlib').setLevel(logging.WARNING) model = Svc("logs/44k/06_21_27_28/G_61000.pth", "logs/44k/06_21_27_28/config.json", cluster_model_path="llogs/44k/06_21_27_28/kmeans_10000.pt") def vc_fn(sid, input_audio, vc_transform, auto_f0,cluster_ratio, noise_scale): if input_audio is None: return "You need to upload an audio", None sampling_rate, audio = input_audio # print(audio.shape,sampling_rate) duration = audio.shape[0] / sampling_rate if duration > 45: return "请上传小于45s的音频,需要转换长音频请本地进行转换", None audio = (audio / np.iinfo(audio.dtype).max).astype(np.float32) 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) print(audio.shape) out_wav_path = "temp.wav" soundfile.write(out_wav_path, audio, 16000, format="wav") print( cluster_ratio, auto_f0, noise_scale) out_audio, out_sr = model.infer(sid, vc_transform, out_wav_path, cluster_infer_ratio=cluster_ratio, auto_predict_f0=auto_f0, noice_scale=noise_scale ) return "Success", (44100, out_audio.cpu().numpy()) app = gr.Blocks() with app: with gr.Tabs(): with gr.TabItem("Basic"): gr.Markdown(value=""" datealive角色语音合成 so-vits-svc-4.0 在线试用 使用凛绪轮回中的语音数据训练而成。 目前可用角色为凛祢、凛绪、鞠亚、鞠奈。 其它角色在训练中,效果还不够好,日后还会更新。 可以下载其它角色的模型以试用,文件夹名称对应角色编号。 """) spks = list(model.spk2id.keys()) sid = gr.Dropdown(label="音色", choices=spks, value=spks[0]) vc_input3 = gr.Audio(label="上传音频(长度小于45秒)") vc_transform = gr.Number(label="变调(整数,可以正负,半音数量,升高八度就是12),使用变调请务必同时将伴奏变调相同半音数!", value=0) cluster_ratio = gr.Number(label="聚类模型混合比例,0-1之间,默认为0不启用聚类,能提升音色相似度,但会导致咬字下降(如果使用建议0.5左右)", value=0) auto_f0 = gr.Checkbox(label="自动f0预测,配合聚类模型f0预测效果更好,会导致变调功能失效(仅限转换语音,歌声不要勾选此项会究极跑调)", value=False) noise_scale = gr.Number(label="noise_scale 建议不要动,会影响音质,玄学参数", value=0.4) vc_submit = gr.Button("转换", variant="primary") vc_output1 = gr.Textbox(label="Output Message") vc_output2 = gr.Audio(label="Output Audio") vc_submit.click(vc_fn, [sid, vc_input3, vc_transform,auto_f0,cluster_ratio, noise_scale], [vc_output1, vc_output2]) gr.HTML("""
仅供学习交流试用,不可用于商业或非法用途。
由此项目产出的任何结果公开发表需指明输入源,注明原作者及代码来源
""") app.launch()