代码重钩，将功能模块化，使代码高内聚，第耦合

TTSs/base_tts.py

@@ -0,0 +1,141 @@
+import abc
+import io
+import scipy.io.wavfile as wavfile
+from pydub import AudioSegment
+
+import gradio as gr
+
+
+
+def mix_background_music(original_audio, 背景音乐, TTS_up, bg_up):
+    original_audio = original_audio + TTS_up
+
+    ori_out = io.BytesIO()
+    original_audio.export(ori_out, format="mp3")
+    ori_out.seek(0)
+    original_audio_bytes = ori_out.read()
+
+    if 背景音乐 is None:  # 没有背景音乐
+        return original_audio_bytes, None
+
+    else:  # 有背景音乐
+        # 读取用户上传的背景音乐文件
+        wav_io = io.BytesIO()
+        wavfile.write(wav_io, 背景音乐[0], 背景音乐[1])
+        wav_io.seek(0)
+        background_music = AudioSegment.from_wav(wav_io)
+
+        background_music = background_music + bg_up
+
+        # 将背景音乐合并到原始音频中
+        combined = original_audio.overlay(background_music)
+
+        # 使用BytesIO导出合并后的音频为字节串
+        mix_out = io.BytesIO()
+        combined.export(mix_out, format="mp3")
+        mix_out.seek(0)  # 重置指针到开始位置
+        combined_audio_bytes = mix_out.read()
+
+        return original_audio_bytes, combined_audio_bytes
+
+class Base_TTS(metaclass=abc.ABCMeta):
+    default_show = False
+
+    @abc.abstractmethod
+    def _get_config_page(self):
+        """要求返回2个参数：config和inputs
+        config: gr.Group对象，用于包裹所有的输入项
+        inputs: 一个列表，包含所有的输入项，用于后续的事件绑定，列表的顺序必须和后续_generate方法中参数args的顺序一致
+
+        例如：
+        :return: config, inputs
+        """
+
+    @abc.abstractmethod
+    def _generate(self, text, *args):
+        """要求返回一个AudioSegment对象
+
+        :param text: str，要转换为语音的文本
+        :param args: 其他参数，与_get_config_page方法中的inputs一一对应
+
+        :return: original_audio
+
+        示例：
+        original_audio = AudioSegment.from_file(io.BytesIO(original_audio_bytes), format="mp3")
+        return original_audio
+        """
+
+    @abc.abstractmethod
+    def get_name(self):
+        """要求返回一个str，是TTS的名称
+
+        :return: name
+
+        示例：
+        return '原神语音合成引擎'
+        """
+
+    def is_show(self):
+        """
+        是否显示这个TTS
+
+        :return: True or False
+        """
+        return True
+
+    def _get_submit_button(self):
+        """
+        要求返回1个参数：btn
+        btn: gr.Button对象，是提交按钮
+
+        :return: btn
+        """
+        btn = gr.Button(value="test提交", variant="primary", interactive=True, visible=False)
+        return btn
+
+    def get_config_page(self):
+        self.config, self.inputs = self._get_config_page()
+        if self.default_show:
+            self.config.visible = True
+        else:
+            self.config.visible = False
+
+    def get_submit_button(self):
+        self.btn = self._get_submit_button()
+        if self.default_show:
+            self.btn.visible = True
+        else:
+            self.btn.visible = False
+
+    def set_visible(self, visible: bool):
+        # 检查是否已经获取了config和btn
+        if not hasattr(self, 'config') or not hasattr(self, 'btn'):
+            raise Exception('请先调用get_config_page和get_submit_button方法')
+
+        self.btn.visible = visible
+        self.config.visible = visible
+
+    def create_interface_event(self, text, audio, TTS_up, bg_up, text_output, ori_audio_output, mix_audio_output):
+        """如果交互方法只是的点击提交按钮，可以不用重写这个方法"""
+
+        # 检查是否已经获取了config和btn
+        if not hasattr(self, 'config') or not hasattr(self, 'btn'):
+            raise Exception('请先调用get_config_page和get_submit_button方法')
+
+        self.btn.click(self.generate, inputs=[
+                                                 text,
+                                                 audio,
+                                                 TTS_up,
+                                                 bg_up
+                                             ] + self.inputs,
+                       outputs=[text_output, ori_audio_output, mix_audio_output])
+
+    def generate(self, text, 背景音乐, TTS_up, bg_up, *args):
+        try:
+            original_audio = self._generate(text, *args)
+
+            return None, *mix_background_music(original_audio, 背景音乐, TTS_up, bg_up)
+        except Exception as e:
+            return str(e), None, None
+
+

TTSs/elevenlabs_tts.py

@@ -1,20 +1,68 @@
 import io
-import json
+import os
 
-import requests
+import gradio as gr
 from elevenlabs import voices, generate, set_api_key, VoiceSettings, Voice
 from pydub import AudioSegment
 
-from .tts_utils import mix_background_music
-
-
-def merge_audio(token, text, 背景音乐, 音色选择, stability, similarity_boost, style, use_speaker_boost, TTS_up, bg_up):
-    try:
+from TTSs.base_tts import Base_TTS
+
+
+class eleven_TTS(Base_TTS):
+    def __init__(self):
+        self.voices_eleven, self.speakers_eleven, self.select_key = self.get_eleven_spk()
+
+    def get_name(self):
+        return 'ElevenLabs'
+
+    def _get_config_page(self):
+        with gr.Group(visible=False) as config:
+            with gr.Row():
+                elevenlabs_api_key = gr.Textbox(label="elevenlabs的API Key（默认为环境变量值）",
+                                                placeholder="请输入elevenlabs的API Key",
+                                                type="password",
+                                                interactive=True,
+                                                value=os.environ.get('ELEVENLABS_API_KEY', ''))
+                speaker_eleven = gr.Dropdown(choices=self.select_key,
+                                             value=self.select_key[0], label="音色选择",
+                                             interactive=True)
+            with gr.Blocks():
+                with gr.Row():
+                    stability = gr.Slider(minimum=0, maximum=1, value=0.5, label='stability', interactive=True)
+                    similarity_boost = gr.Slider(minimum=0, maximum=1, value=0.75, label='similarity_boost',
+                                                 interactive=True)
+
+                with gr.Row():
+                    style = gr.Slider(minimum=0, maximum=1, value=0, label='style', interactive=True)
+                    use_speaker_boost = gr.Checkbox(label="Use Speaker Boost", value=True, interactive=True)
+
+        inputs = [
+            elevenlabs_api_key,
+            speaker_eleven,
+            stability,
+            similarity_boost,
+            style,
+            use_speaker_boost
+        ]
+
+        return config, inputs
+
+    def _get_submit_button(self):
+        """
+        要求返回1个参数：btn
+        btn: gr.Button对象，是提交按钮
+
+        :return: btn
+        """
+        btn = gr.Button(value="ElevenLabs提交", variant="primary", interactive=True, visible=False)
+        return btn
+
+    def _generate(self, text, token, 音色选择, stability, similarity_boost, style, use_speaker_boost):
         set_api_key(token)
 
         original_audio_bytes = generate(
             text=text,
-            voice=Voice(voice_id=voices_eleven[speakers_eleven[音色选择]].voice_id,
+            voice=Voice(voice_id=self.voices_eleven[self.speakers_eleven[音色选择]].voice_id,
                         settings=VoiceSettings(stability=stability, similarity_boost=similarity_boost, style=style,
                                                use_speaker_boost=use_speaker_boost)),
             model="eleven_multilingual_v2",
@@ -23,44 +71,36 @@ def merge_audio(token, text, 背景音乐, 音色选择, stability, similarity_b
 
         # 将字节串转换为AudioSegment对象
         original_audio = AudioSegment.from_file(io.BytesIO(original_audio_bytes), format="mp3")
-
-        return None, *mix_background_music(original_audio, 背景音乐, TTS_up, bg_up)
-    except Exception as e:
-        return str(e), None, None
-
-
-def get_eleven_spk():
-    try:
-        voices_eleven = voices()
-        speakers_eleven = {}
-        for i, v in enumerate(voices_eleven):
-            label = {}
-            label['口音'] = v.labels.get('accent')
-
-            if 'description ' in v.labels:
-                label['描述'] = v.labels['description ']
-            elif 'description' in v.labels:
-                label['描述'] = v.labels['description']
-
-            label['年龄'] = v.labels['age']
-            label['性别'] = v.labels['gender']
-
-            if 'use case' in v.labels:
-                label['用例'] = v.labels['use case']
-            elif 'usecase' in v.labels:
-                label['用例'] = v.labels['usecase']
-            else:
-                raise Exception('use case not found')
-
-            key = v.name + '-' + str(label)
-            speakers_eleven[key] = i
-        select_key = list(speakers_eleven.keys())
-
-        return voices_eleven, speakers_eleven, select_key
-    except Exception as e:
-        print(e)
-        return [], {'获取音色失败': '获取音色失败'}, ['获取音色失败']
-
-
-
-voices_eleven, speakers_eleven, select_key = get_eleven_spk()
+        return original_audio
+
+    def get_eleven_spk(self):
+        try:
+            voices_eleven = voices()
+            speakers_eleven = {}
+            for i, v in enumerate(voices_eleven):
+                label = {}
+                label['口音'] = v.labels.get('accent')
+
+                if 'description ' in v.labels:
+                    label['描述'] = v.labels['description ']
+                elif 'description' in v.labels:
+                    label['描述'] = v.labels['description']
+
+                label['年龄'] = v.labels['age']
+                label['性别'] = v.labels['gender']
+
+                if 'use case' in v.labels:
+                    label['用例'] = v.labels['use case']
+                elif 'usecase' in v.labels:
+                    label['用例'] = v.labels['usecase']
+                else:
+                    raise Exception('use case not found')
+
+                key = v.name + '-' + str(label)
+                speakers_eleven[key] = i
+            select_key = list(speakers_eleven.keys())
+
+            return voices_eleven, speakers_eleven, select_key
+        except Exception as e:
+            print(e)
+            return [], {'获取音色失败': '获取音色失败'}, ['获取音色失败']

TTSs/genshin_api_tts.py

@@ -1,46 +1,95 @@
 import io
 import json
+import os
 
+import gradio as gr
 import requests
 from pydub import AudioSegment
 
-from .tts_utils import mix_background_music
+from TTSs.base_tts import Base_TTS
 
 
-def infer(token, appid, speaker, text, sdp, noise, noisew, length, 背景音乐, speaker_up, back_up):
-    try:
+class genshin_api_TTS(Base_TTS):
+    def __init__(self):
+        self.genshin_api = "https://tirs.ai-lab.top"
+        self.speakers_genshin = self.get_spk()
+
+    def get_name(self):
+        return '原神语音合成-api'
+
+    def _get_submit_button(self):
+        """
+        要求返回1个参数：btn
+        btn: gr.Button对象，是提交按钮
+
+        :return: btn
+        """
+        btn = gr.Button(value="原神提交", variant="primary", interactive=True, visible=False)
+        return btn
+
+    def _get_config_page(self):
+        with gr.Group(visible=False) as config_genshin_api:
+            gr.Markdown('原作者：https://modelscope.cn/studios/erythrocyte/Bert-VITS2_Genshin_TTS/summary')
+            with gr.Row():
+                token = gr.Textbox(label="AccessToken", placeholder="请输入AccessToken",
+                                   type="password",
+                                   interactive=True, value=os.environ.get('GENSHIN_API_KEY', ''))
+                appid = gr.Textbox(label="AppID", placeholder="请输入AppID", interactive=True,
+                                   value='ig8t76x6036h3hpw')
+            with gr.Row():
+                gr.Markdown(
+                    "AppID不要改，AccessToken最好换自己的。")
+
+            with gr.Row():
+                speaker = gr.Dropdown(choices=self.speakers_genshin,
+                                      value=self.speakers_genshin[0], label="角色",
+                                      interactive=True)
+                search = gr.Textbox(label="搜索角色", lines=1, interactive=True)
+                btn2 = gr.Button(value="搜索")
+            with gr.Column():
+                with gr.Row():
+                    sdp_ratio = gr.Slider(minimum=0, maximum=1, value=0.2, step=0.1, label="SDP/DP 混合比",
+                                          interactive=True)
+                    noise_scale = gr.Slider(minimum=0.1, maximum=2, value=0.6, step=0.1, label="感情",
+                                            interactive=True)
+                with gr.Row():
+                    noise_scale_w = gr.Slider(minimum=0.1, maximum=2, value=0.8, step=0.1, label="音素长度",
+                                              interactive=True)
+                    length_scale = gr.Slider(minimum=-99, maximum=99, value=0, step=0.1, label="语速(%)",
+                                             interactive=True)
+
+            btn2.click(self.search_speaker, inputs=[search], outputs=[speaker])
+
+        inputs = [
+            token, appid, speaker, sdp_ratio, noise_scale, noise_scale_w,
+            length_scale
+        ]
+
+        return config_genshin_api, inputs
+
+    def _generate(self, text, token, appid, speaker, sdp, noise, noisew, length):
         speed = (100 - length) / 100
         headers = {'Content-Type': 'application/json'}
         infer_info = {'lang': 'zh', 'appid': appid, 'token': token, 'speaker': speaker, 'text': text,
                       'sdp_ratio': sdp, 'noise': noise, 'noisew': noisew, 'length': speed}
-        resp = requests.post(url=f"{genshin_api}/api/ex/vits", headers=headers,
+        resp = requests.post(url=f"{self.genshin_api}/api/ex/vits", headers=headers,
                              data=json.dumps(infer_info))
         data = json.loads(resp.text)
 
         wmv_file = requests.get(data["audio"]).content
 
         original_audio = AudioSegment.from_wav(io.BytesIO(wmv_file))
+        return original_audio
 
-        return data["message"].replace("\\n", "\n"), *mix_background_music(original_audio, 背景音乐, speaker_up,
-                                                                           back_up)
-    except Exception as e:
-        return str(e), None, None
-
-
-def get_spk():
-    resp = requests.get(url=f"{genshin_api}/spklist/spks.json")
-    data = json.loads(resp.text)
-    return data
-
-
-def search_speaker(search_value):
-    for s in speakers_genshin:
-        if search_value == s:
-            return s
-    for s in speakers_genshin:
-        if search_value in s:
-            return s
-
+    def get_spk(self):
+        resp = requests.get(url=f"{self.genshin_api}/spklist/spks.json")
+        data = json.loads(resp.text)
+        return data
 
-genshin_api = "https://tirs.ai-lab.top"
-speakers_genshin = get_spk()
+    def search_speaker(self, search_value):
+        for s in self.speakers_genshin:
+            if search_value == s:
+                return s
+        for s in self.speakers_genshin:
+            if search_value in s:
+                return s

TTSs/genshin_local/__init__.py

@@ -0,0 +1 @@
+from .genshin_local_tts import genshin_local_TTS