TTSs/elevenlabs_tts.py

import io
import os

import gradio as gr
from elevenlabs import voices, generate, set_api_key, VoiceSettings, Voice
from pydub import AudioSegment

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=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",

        )

        # 将字节串转换为AudioSegment对象
        original_audio = AudioSegment.from_file(io.BytesIO(original_audio_bytes), format="mp3")
        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']