app.py

import spaces
import os
import random
import argparse

import torch
import gradio as gr
import numpy as np

import ChatTTS

import se_extractor
from api import BaseSpeakerTTS, ToneColorConverter
import soundfile

from tts_voice import tts_order_voice
import edge_tts
import tempfile
import anyio

print("loading ChatTTS model...")
chat = ChatTTS.Chat()
chat.load_models()


def generate_seed():
    new_seed = random.randint(1, 100000000)
    return {
        "__type__": "update",
        "value": new_seed
        }

@spaces.GPU
def chat_tts(text, temperature, top_P, top_K, audio_seed_input, text_seed_input, refine_text_flag, refine_text_input, output_path=None):

    torch.manual_seed(audio_seed_input)
    rand_spk = torch.randn(768)
    params_infer_code = {
        'spk_emb': rand_spk, 
        'temperature': temperature,
        'top_P': top_P,
        'top_K': top_K,
        }
    params_refine_text = {'prompt': '[oral_2][laugh_0][break_6]'}
    
    torch.manual_seed(text_seed_input)

    if refine_text_flag:
        if refine_text_input:
           params_refine_text['prompt'] = refine_text_input
        text = chat.infer(text, 
                          skip_refine_text=False,
                          refine_text_only=True,
                          params_refine_text=params_refine_text,
                          params_infer_code=params_infer_code
                          )
        print("Text has been refined!")
    
    wav = chat.infer(text, 
                     skip_refine_text=True, 
                     params_refine_text=params_refine_text, 
                     params_infer_code=params_infer_code
                     )
    
    audio_data = np.array(wav[0]).flatten()
    sample_rate = 24000
    text_data = text[0] if isinstance(text, list) else text

    if output_path is None:
        return [(sample_rate, audio_data), text_data]
    else:
        soundfile.write(output_path, audio_data, sample_rate)

# OpenVoice

ckpt_base_en = 'checkpoints/base_speakers/EN'
ckpt_converter_en = 'checkpoints/converter'
device = 'cuda:0'

#device = "cpu"

base_speaker_tts = BaseSpeakerTTS(f'{ckpt_base_en}/config.json', device=device)
base_speaker_tts.load_ckpt(f'{ckpt_base_en}/checkpoint.pth')

tone_color_converter = ToneColorConverter(f'{ckpt_converter_en}/config.json', device=device)
tone_color_converter.load_ckpt(f'{ckpt_converter_en}/checkpoint.pth')


def generate_audio(text, audio_ref, temperature, top_P, top_K, audio_seed_input, text_seed_input, refine_text_flag, refine_text_input):
    source_se = torch.load(f'{ckpt_base_en}/en_default_se.pth').to(device)
    reference_speaker = audio_ref
    target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, target_dir='processed', vad=True)
    save_path = "output.wav"

    # Run the base speaker tts
    src_path = "tmp.wav"
    chat_tts(text, temperature, top_P, top_K, audio_seed_input, text_seed_input, refine_text_flag, refine_text_input, src_path)
    print("Ready for voice cloning!")
    
    source_se, audio_name = se_extractor.get_se(src_path, tone_color_converter, target_dir='processed', vad=True)
    print("Get source segment!")
    
    # Run the tone color converter
    encode_message = "@Hilley"
    # convert from file
    tone_color_converter.convert(
        audio_src_path=src_path,
        src_se=source_se,
        tgt_se=target_se,
        output_path=save_path,
        message=encode_message)

    ''' 
    # convert from data
    src_path = None
    sample_rate, audio = chat_tts(text, temperature, top_P, top_K, audio_seed_input, text_seed_input, refine_text_flag, refine_text_input, src_path)[0]
    print("Ready for voice cloning!")
    tone_color_converter.convert_data(
        audio=audio,
        sample_rate=sample_rate,
        src_se=source_se,
        tgt_se=target_se,
        output_path=save_path,
        message=encode_message)
    '''
    print("Finished!")

    return "output.wav"

def vc_en(text, audio_ref, style_mode):
    if style_mode=="default":
        source_se = torch.load(f'{ckpt_base_en}/en_default_se.pth').to(device)
        reference_speaker = audio_ref
        target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, target_dir='processed', vad=True)
        save_path = "output.wav"

        # Run the base speaker tts
        src_path = "tmp.wav"
        base_speaker_tts.tts(text, src_path, speaker='default', language='English', speed=1.0)

        # Run the tone color converter
        encode_message = "@MyShell"
        tone_color_converter.convert(
            audio_src_path=src_path,
            src_se=source_se,
            tgt_se=target_se,
            output_path=save_path,
            message=encode_message)

    else:
        source_se = torch.load(f'{ckpt_base_en}/en_style_se.pth').to(device)
        reference_speaker = audio_ref
        target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, target_dir='processed', vad=True)

        save_path = "output.wav"

        # Run the base speaker tts
        src_path = "tmp.wav"
        base_speaker_tts.tts(text, src_path, speaker=style_mode, language='English', speed=0.9)

        # Run the tone color converter
        encode_message = "@MyShell"
        tone_color_converter.convert(
            audio_src_path=src_path,
            src_se=source_se,
            tgt_se=target_se,
            output_path=save_path,
            message=encode_message)

    return "output.wav"

language_dict = tts_order_voice

base_speaker = "base_audio.mp3"
source_se, audio_name = se_extractor.get_se(base_speaker, tone_color_converter, vad=True)

async def text_to_speech_edge(text, audio_ref, language_code):
    voice = language_dict[language_code]
    communicate = edge_tts.Communicate(text, voice)
    with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp_file:
        tmp_path = tmp_file.name

    await communicate.save(tmp_path)

    reference_speaker = audio_ref
    target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, target_dir='processed', vad=True)
    save_path = "output.wav"

    # Run the tone color converter
    encode_message = "@MyShell"
    tone_color_converter.convert(
        audio_src_path=tmp_path,
        src_se=source_se,
        tgt_se=target_se,
        output_path=save_path,
        message=encode_message)

    return "output.wav"


with gr.Blocks() as demo:
    gr.Markdown("# Enjoy chatting with your ai friends on website, telegram and so on! (https://linkin.love)")

    default_text = "Today a man knocked on my door and asked for a small donation toward the local swimming pool. I gave him a glass of water."        
    text_input = gr.Textbox(label="Input Text", lines=4, placeholder="Please Input Text...", value=default_text)
    voice_ref = gr.Audio(label="Reference Audio", info="Click on the ✎ button to upload your own target speaker audio", type="filepath", value="base_audio.mp3")
    
    with gr.Tab("💕Super Natural"):
        default_refine_text = "[oral_2][laugh_0][break_6]"    
        refine_text_checkbox = gr.Checkbox(label="Refine text", info="'oral' means add filler words, 'laugh' means add laughter, and 'break' means add a pause. (0-10) ", value=True)
        refine_text_input = gr.Textbox(label="Refine Prompt", lines=1, placeholder="Please Refine Prompt...", value=default_refine_text)

        with gr.Row():
            temperature_slider = gr.Slider(minimum=0.00001, maximum=1.0, step=0.00001, value=0.3, label="Audio temperature")
            top_p_slider = gr.Slider(minimum=0.1, maximum=0.9, step=0.05, value=0.7, label="top_P")
            top_k_slider = gr.Slider(minimum=1, maximum=20, step=1, value=20, label="top_K")

        with gr.Row():
            audio_seed_input = gr.Number(value=42, label="Speaker Seed")
            generate_audio_seed = gr.Button("\U0001F3B2")
            text_seed_input = gr.Number(value=42, label="Text Seed")
            generate_text_seed = gr.Button("\U0001F3B2")

        generate_button = gr.Button("Generate!")
        
        #text_output = gr.Textbox(label="Refined Text", interactive=False)
        audio_output = gr.Audio(label="Output Audio")

        generate_audio_seed.click(generate_seed, 
                                  inputs=[], 
                                  outputs=audio_seed_input)
        
        generate_text_seed.click(generate_seed, 
                                 inputs=[], 
                                 outputs=text_seed_input)
        
        generate_button.click(generate_audio, 
                              inputs=[text_input, voice_ref, temperature_slider, top_p_slider, top_k_slider, audio_seed_input, text_seed_input, refine_text_checkbox, refine_text_input], 
                              outputs=audio_output)
        
    with gr.Tab("💕Emotion Control"):
        emo_pick = gr.Dropdown(label="Emotion", info="🙂default😊friendly🤫whispering😄cheerful😱terrified😡angry😢sad", choices=["default", "friendly", "whispering", "cheerful", "terrified", "angry", "sad"], value="default")
        generate_button_emo = gr.Button("Generate!", variant="primary")
        audio_emo = gr.Audio(label="Output Audio", type="filepath")
        generate_button_emo.click(vc_en, [text_input, voice_ref, emo_pick], audio_emo)

    with gr.Tab("💕multilingual"):
        language = gr.Dropdown(choices=list(language_dict.keys()), value=list(language_dict.keys())[15], label="请选择文本对应的语言及说话人")
        generate_button_ml = gr.Button("开始语音情感真实复刻吧！", variant="primary")
        audio_ml = gr.Audio(label="为您合成的专属语音", type="filepath")
        generate_button_ml.click(text_to_speech_edge, [text_input, voice_ref, language], audio_ml)

parser = argparse.ArgumentParser(description='ChatTTS demo Launch')
parser.add_argument('--server_name', type=str, default='0.0.0.0', help='Server name')
parser.add_argument('--server_port', type=int, default=8080, help='Server port')
args = parser.parse_args()

    # demo.launch(server_name=args.server_name, server_port=args.server_port, inbrowser=True)




if __name__ == '__main__':
    demo.launch()