app.py

import os
import io
import gradio as gr
import librosa
import numpy as np
import logging
import soundfile
import asyncio
import argparse
import gradio.processing_utils as gr_processing_utils
logging.getLogger('numba').setLevel(logging.WARNING)
logging.getLogger('markdown_it').setLevel(logging.WARNING)
logging.getLogger('urllib3').setLevel(logging.WARNING)
logging.getLogger('matplotlib').setLevel(logging.WARNING)

limitation = os.getenv("SYSTEM") == "spaces"  # limit audio length in huggingface spaces

audio_postprocess_ori = gr.Audio.postprocess

def audio_postprocess(self, y):
    data = audio_postprocess_ori(self, y)
    if data is None:
        return None
    return gr_processing_utils.encode_url_or_file_to_base64(data["name"])

gr.Audio.postprocess = audio_postprocess
def create_vc_fn(model, sid):
    def vc_fn(input_audio, vc_transform, auto_f0):
        if input_audio is None:
            return "You need to upload an audio", None
        sampling_rate, audio = input_audio
        duration = audio.shape[0] / sampling_rate
        if duration > 20 and limitation:
            return "Please upload an audio file that is less than 20 seconds. If you need to generate a longer audio file, please use Colab.", None
        audio = (audio / np.iinfo(audio.dtype).max).astype(np.float32)
        if len(audio.shape) > 1:
            audio = librosa.to_mono(audio.transpose(1, 0))
        if sampling_rate != 16000:
            audio = librosa.resample(audio, orig_sr=sampling_rate, target_sr=16000)
        raw_path = io.BytesIO()
        soundfile.write(raw_path, audio, 16000, format="wav")
        raw_path.seek(0)
        out_audio, out_sr = model.infer(sid, vc_transform, raw_path,
                                       auto_predict_f0=auto_f0,
                                       )
        return "Success", (44100, out_audio.cpu().numpy())
    return vc_fn

def get_speakers():
  speakers = []

  for _,dirs,_ in os.walk(CUSTOM_MODELS_DIR):
    for folder in dirs:
      cur_speaker = {}
      # Look for G_****.pth
      g = glob.glob(os.path.join(CUSTOM_MODELS_DIR,folder,'G_*.pth'))
      if not len(g):
        continue
      cur_speaker["model_path"] = g[0]
      cur_speaker["model_folder"] = folder

      # Look for *.pt (clustering model)
      clst = glob.glob(os.path.join(CUSTOM_MODELS_DIR,folder,'*.pt'))
      if not len(clst):
        cur_speaker["cluster_path"] = ""
      else:
        cur_speaker["cluster_path"] = clst[0]

      # Look for config.json
      cfg = glob.glob(os.path.join(CUSTOM_MODELS_DIR,folder,'*.json'))
      if not len(cfg):
        continue
      cur_speaker["cfg_path"] = cfg[0]
      with open(cur_speaker["cfg_path"]) as f:
        try:
          cfg_json = json.loads(f.read())
        except Exception as e:
          print("Malformed config json in "+folder)
        for name, i in cfg_json["spk"].items():
          cur_speaker["name"] = name
          cur_speaker["id"] = i
          if not name.startswith('.'):
            speakers.append(copy.copy(cur_speaker))

  return sorted(speakers, key=lambda x:x["name"].lower())

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--device', type=str, default='cpu')
    parser.add_argument('--api', action="store_true", default=False)
    parser.add_argument("--share", action="store_true", default=False, help="share gradio app")
    args = parser.parse_args()

    speakers = get_speakers()
    speaker_list = [x["name"] for x in speakers]

    models = []
    voices = []

    # !svc infer {NAME}.wav -c config.json -m G_riri_220.pth
    #  display(Audio(f"{NAME}.out.wav", autoplay=True))
    with gr.Blocks() as app:
        gr.Markdown(
            "# <center> Sovits Chapay\n"
            "## <center> The input audio should be clean and pure voice without background music.\n"
        )

        with gr.Row():
            with gr.Column():
                vc_input = gr.Audio(label="Input audio"+' (less than 20 seconds)' if limitation else '')

                vc_transform = gr.Number(label="vc_transform", value=0)

                voice = gr.Dropdown(choices=speaker_list, visible=True)

                vc_submit = gr.Button("Generate", variant="primary")
            with gr.Column():
                vc_output1 = gr.Textbox(label="Output Message")
                vc_output2 = gr.Audio(label="Output Audio")
            vc_submit.click(vc_fn, [vc_input, vc_transform, auto_f0], [vc_output1, vc_output2])
        app.queue(concurrency_count=1, api_open=args.api).launch(share=args.share)