TTSs/openai_tts.py

import io
import os

import gradio as gr
from openai import OpenAI
from pydub import AudioSegment

from TTSs.base_tts import Base_TTS


class OpenAI_TTS(Base_TTS):

    def get_name(self):
        return 'OpenAI'

    def _get_config_page(self):
        with gr.Group(visible=False) as config:
            openai_api_key = gr.Textbox(label="openai的API Key（默认为环境变量值）",
                                        placeholder="请输入openai的API Key",
                                        type="password",
                                        interactive=True,
                                        value=os.environ.get('OPENAI_API_KEY', ''))
            with gr.Row():
                openai_model_choices = ['tts-1', 'tts-1-hd']
                openai_voices = ['alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer']
                openai_model = gr.Dropdown(choices=openai_model_choices, value=openai_model_choices[0],
                                           label="模型选择",
                                           interactive=True)
                openai_speaker = gr.Dropdown(choices=openai_voices, value=openai_voices[0], label="音色选择",
                                             interactive=True)
                speed = gr.Slider(minimum=0.25, maximum=4.0, value=1, label='语速', step=0.1, interactive=True)

        inputs = [
            openai_api_key, openai_model, openai_speaker, speed
        ]

        return config, inputs

    def _generate(self, text, openai_api_key, openai_model, openai_speaker, speed):
        client = OpenAI(api_key=openai_api_key)
        response = client.audio.speech.create(
            model=openai_model,
            voice=openai_speaker,
            speed=speed,
            input=text
        )

        original_audio = AudioSegment.from_file(io.BytesIO(response.content), format="mp3")
        return original_audio