webui.py

import os
os.environ['TMPDIR'] = './temps' # avoid the system default temp folder not having access permissions
# os.environ['HF_ENDPOINT'] = 'https://hf-mirror.com' # use huggingfacae mirror for users that could not login to huggingface

import re
import numpy as np
import matplotlib.pyplot as plt

import torch
import gradio as gr

from api import StableTTSAPI

device = 'cuda' if torch.cuda.is_available() else 'cpu'

tts_model_path = './checkpoints/checkpoint_0.pt'
vocoder_model_path = './vocoders/pretrained/firefly-gan-base-generator.ckpt'
vocoder_type = 'ffgan'

model = StableTTSAPI(tts_model_path, vocoder_model_path, vocoder_type).to(device)

@ torch.inference_mode()
def inference(text, ref_audio, language, step, temperature, length_scale, solver, cfg):
    text = remove_newlines_after_punctuation(text)
    
    if language == 'chinese':
        text = text.replace(' ', '')
        
    audio, mel = model.inference(text, ref_audio, language, step, temperature, length_scale, solver, cfg)
    
    max_val = torch.max(torch.abs(audio))
    if max_val > 1:
        audio = audio / max_val
    
    audio_output = (model.mel_config.sample_rate, (audio.cpu().squeeze(0).numpy() * 32767).astype(np.int16)) # (samplerate, int16 audio) for gr.Audio
    mel_output = plot_mel_spectrogram(mel.cpu().squeeze(0).numpy()) # get the plot of mel
    
    return audio_output, mel_output

def plot_mel_spectrogram(mel_spectrogram):
    plt.close() # prevent memory leak
    fig, ax = plt.subplots(figsize=(20, 8))
    ax.imshow(mel_spectrogram, aspect='auto', origin='lower')
    plt.axis('off')
    fig.subplots_adjust(left=0, right=1, top=1, bottom=0) # remove white edges
    return fig

def remove_newlines_after_punctuation(text):
    pattern = r'([，。！？、“”‘’《》【】；：,.!?\'\"<>()\[\]{}])\n'
    return re.sub(pattern, r'\1', text)

def main():
    
    from pathlib import Path
    examples = list(Path('./audios').rglob('*.wav'))

    # gradio wabui, reference: https://huggingface.co/spaces/fishaudio/fish-speech-1
    gui_title = 'StableTTS'
    gui_description = """Next-generation TTS model using flow-matching and DiT, inspired by Stable Diffusion 3."""
    example_text = """你指尖跳动的电光，是我永恒不变的信仰。唯我超电磁炮永世长存！"""
    
    with gr.Blocks(theme=gr.themes.Base()) as demo:
        demo.load(None, None, js="() => {const params = new URLSearchParams(window.location.search);if (!params.has('__theme')) {params.set('__theme', 'light');window.location.search = params.toString();}}")

        with gr.Row():
            with gr.Column():
                gr.Markdown(f"# {gui_title}")
                gr.Markdown(gui_description)

        with gr.Row():
            with gr.Column():
                input_text_gr = gr.Textbox(
                    label="Input Text",
                    info="Put your text here",
                    value=example_text,
                )
             
                ref_audio_gr = gr.Audio(
                    label="Reference Audio",
                    type="filepath"
                )
                
                language_gr = gr.Dropdown(
                    label='Language',
                    choices=list(model.supported_languages),
                    value = 'chinese'
                )
                
                step_gr = gr.Slider(
                    label='Step',
                    minimum=1,
                    maximum=100,
                    value=25,
                    step=1
                )
                
                temperature_gr = gr.Slider(
                    label='Temperature',
                    minimum=0,
                    maximum=2,
                    value=1,
                )
                
                length_scale_gr = gr.Slider(
                    label='Length_Scale',
                    minimum=0,
                    maximum=5,
                    value=1,
                )
                
                solver_gr = gr.Dropdown(
                    label='ODE Solver',
                    choices=['euler', 'midpoint', 'dopri5', 'rk4', 'implicit_adams', 'bosh3', 'fehlberg2', 'adaptive_heun'],
                    value = 'dopri5'
                )
                
                cfg_gr = gr.Slider(
                    label='CFG',
                    minimum=0,
                    maximum=10,
                    value=3,
                )
                
            with gr.Column():
                mel_gr = gr.Plot(label="Mel Visual")
                audio_gr = gr.Audio(label="Synthesised Audio", autoplay=True)
                tts_button = gr.Button("\U0001F3A7 Generate / 合成", elem_id="send-btn", visible=True, variant="primary")
                examples = gr.Examples(examples, ref_audio_gr)

        tts_button.click(inference, [input_text_gr, ref_audio_gr, language_gr, step_gr, temperature_gr, length_scale_gr, solver_gr, cfg_gr], outputs=[audio_gr, mel_gr])

    demo.queue()  
    demo.launch(debug=True, show_api=True)


if __name__ == '__main__':
    main()