TTSs/genshin_local/genshin_local_tts.py

import io
import logging
import os
import traceback

import gradio as gr
import scipy.io.wavfile as wavfile
from pydub import AudioSegment

from TTSs.base_tts import Base_TTS


class genshin_local_TTS(Base_TTS):
    def __init__(self):
        if self.is_show():
            try:
                from .genshin_bg import speakers
                from .genshin_bg import languages
                from .genshin_bg import tts_fn
                from .genshin_bg import change_dir

                self.speakers_genshin_local = speakers
                self.languages = languages
                self.genshin_tts_fn = tts_fn
                self.change_dir = change_dir

                logging.info('导入原神本地语音合成模块成功')
            except Exception as e:
                traceback.print_exc()

                logging.error('导入原神本地语音合成模块失败')
                logging.error(e)
        else:
            logging.info('未启用原神本地语音合成模块')

    def is_show(self):
        if os.environ.get('USING_LOCAL', 'false').lower() == 'true':
            return True
        else:
            return False

    def get_name(self):
        return '原神语音合成-local'

    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="原神语音合成（本地版）", variant="primary", interactive=True, visible=False)
        return btn

    def _get_config_page(self):

        with gr.Group(visible=False) as config_genshin_local:
            gr.Markdown('原作者：红血球AE3803\nhttps://space.bilibili.com/6589795/dynamic')
            with gr.Row():
                speaker_local = gr.Dropdown(choices=self.speakers_genshin_local, value=self.speakers_genshin_local[0],
                                            label="角色",
                                            interactive=True)
                search_local = 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_local = gr.Slider(minimum=0, maximum=1, value=0.5, step=0.1,
                                                label="SDP Ratio",
                                                interactive=True)
                    noise_scale_local = gr.Slider(minimum=0.1, maximum=2, value=0.6, step=0.1,
                                                  label="Noise",
                                                  interactive=True)
                with gr.Row():
                    noise_scale_w_local = gr.Slider(minimum=0.1, maximum=2, value=0.9, step=0.1,
                                                    label="Noise_W",
                                                    interactive=True)
                    length_scale_local = gr.Slider(minimum=0.1, maximum=2, value=1.0, step=0.1,
                                                   label="Length",
                                                   interactive=True)

            language_local = gr.Dropdown(
                choices=self.languages, value=self.languages[0], label="Language", interactive=True
            )

            btn2.click(self.search_speaker, inputs=[search_local], outputs=[speaker_local])

        inputs = [
            speaker_local,
            sdp_ratio_local,
            noise_scale_local,
            noise_scale_w_local,
            length_scale_local,
            language_local
        ]

        return config_genshin_local, inputs

    def _generate(self, text, speaker_local,
                  sdp_ratio_local,
                  noise_scale_local,
                  noise_scale_w_local,
                  length_scale_local,
                  language_local):

        with self.change_dir():
            ori_audio_data = self.genshin_tts_fn(text, speaker_local,
                                                 sdp_ratio_local,
                                                 noise_scale_local,
                                                 noise_scale_w_local,
                                                 length_scale_local,
                                                 language_local,
                                                 None, 'Happy',
                                                 'Text prompt',
                                                 'style_text',
                                                 0.7)[1]

        wav_io = io.BytesIO()
        wavfile.write(wav_io, ori_audio_data[0], ori_audio_data[1])
        wav_io.seek(0)
        original_audio = AudioSegment.from_wav(wav_io)
        return original_audio

    def search_speaker(self, search_value):
        for s in self.speakers_genshin_local:
            if search_value == s:
                return s
        for s in self.speakers_genshin_local:
            if search_value in s:
                return s