paddle-文本合成语音

依赖

paddlepaddle==2.4.2
paddleaudio==1.1.0
paddlespeech==1.4.1
gradio==3.20.1

试验

## https://www.paddlepaddle.org.cn/modelsDetail?modelId=26
from paddlespeech.cli.tts import TTSExecutor
## 如果找不到nltk的zip，就手动执行下 nltk_download.py

tts_executor = TTSExecutor()
wav_file = tts_executor(
    text="热烈欢迎您在 Discussions 中提交问题，并在 Issues 中指出发现的 bug。此外，我们非常希望您参与到 Paddle Speech 的开发中！",
    output='output.wav',
    am='fastspeech2_mix',
    voc='hifigan_csmsc',
    lang='mix',
    spk_id=170)

完整代码

import os

import gradio as gr
from paddlespeech.cli.tts import TTSExecutor

tts_executor = TTSExecutor()

def speech_generate(text: str) -> os.PathLike:
    assert isinstance(text,str) and len(text) > 0, 'Input Chinese-English text...'
    wav_file = tts_executor(
        text=text,
        output='output.wav',
        am='fastspeech2_mix',
        voc='hifigan_csmsc',
        lang='zh',
        spk_id=174)
    return wav_file

def clear_all():
    return None, None

with gr.Blocks() as demo:
    gr.Markdown("文本合成语音")

    with gr.Column(scale=1, min_width=50):
        text_input = gr.Textbox(placeholder="Type here...", lines=5)

        with gr.Row():
            btn1 = gr.Button("Clear")
            btn2 = gr.Button("Submit")

        audio_output = gr.Audio(type="filepath", label="Output")

    btn2.click(
        fn=speech_generate,
        inputs=text_input,
        outputs=audio_output,
        scroll_to_output=True)
    btn1.click(fn=clear_all, inputs=None, outputs=[text_input, audio_output])

    gr.Button.style(1)

demo.launch(server_port=7001)