storyteller-ml/voice_conversion/so-vits-svc-michael/inference_main.py

import io
import logging
import time
from pathlib import Path

import librosa
import matplotlib.pyplot as plt
import numpy as np
import soundfile

from inference import infer_tool
from inference import slicer
from inference.infer_tool import Svc

logging.getLogger('numba').setLevel(logging.WARNING)
chunks_dict = infer_tool.read_temp("inference/chunks_temp.json")


def stitch_wav_tensors_with_crossfade(wav_tensor1,wav_tensor2,sample_rate:int):
    # Load two audio signals  as NumPy tensors

    # Compute the length of the cross-fade window in samples
    fade_len = int(0.01 * sample_rate)

    # Create a cross-fade window
    fade_window = np.hanning(2*fade_len)

    # Concatenate the signals with cross-fade
    audio_out = np.concatenate((wav_tensor1[:-fade_len], 
                                wav_tensor1[-fade_len:] * fade_window[:fade_len] + wav_tensor2[:fade_len] * fade_window[fade_len:], 
                                wav_tensor2[fade_len:]))
    
    #sf.write("audio_out.wav", audio_out, sample_rate)
    return audio_out

def main():
    import argparse

    parser = argparse.ArgumentParser(description='sovits4 inference')

    # 一定要设置的部分
    parser.add_argument('-m', '--model_path', type=str, default="logs/44k/G_0.pth", help='模型路径')
    parser.add_argument('-c', '--config_path', type=str, default="configs/config.json", help='配置文件路径')
    parser.add_argument('-n', '--clean_names', type=str, nargs='+', default=["君の知らない物語-src.wav"], help='wav文件名列表，放在raw文件夹下')
    parser.add_argument('-t', '--trans', type=int, nargs='+', default=[0], help='音高调整，支持正负（半音）')
    parser.add_argument('-s', '--spk_list', type=str, nargs='+', default=['nen'], help='合成目标说话人名称')

    # 可选项部分
    parser.add_argument('-a', '--auto_predict_f0', action='store_true', default=False,
                        help='语音转换自动预测音高，转换歌声时不要打开这个会严重跑调')
    parser.add_argument('-cm', '--cluster_model_path', type=str, default="logs/44k/kmeans_10000.pt", help='聚类模型路径，如果没有训练聚类则随便填')
    parser.add_argument('-cr', '--cluster_infer_ratio', type=float, default=0, help='聚类方案占比，范围0-1，若没有训练聚类模型则填0即可')

    # 不用动的部分
    parser.add_argument('-sd', '--slice_db', type=int, default=-40, help='默认-40，嘈杂的音频可以-30，干声保留呼吸可以-50')
    parser.add_argument('-d', '--device', type=str, default=None, help='推理设备，None则为自动选择cpu和gpu')
    parser.add_argument('-ns', '--noice_scale', type=float, default=0.4, help='噪音级别，会影响咬字和音质，较为玄学')
    parser.add_argument('-p', '--pad_seconds', type=float, default=0.5, help='推理音频pad秒数，由于未知原因开头结尾会有异响，pad一小段静音段后就不会出现')
    parser.add_argument('-wf', '--wav_format', type=str, default='flac', help='音频输出格式')

    args = parser.parse_args()

    svc_model = Svc(args.model_path, args.config_path, args.device, args.cluster_model_path)
    infer_tool.mkdir(["raw", "results"])
    clean_names = args.clean_names
    trans = args.trans
    spk_list = args.spk_list
    slice_db = args.slice_db
    wav_format = args.wav_format
    auto_predict_f0 = args.auto_predict_f0
    cluster_infer_ratio = args.cluster_infer_ratio
    noice_scale = args.noice_scale
    pad_seconds = args.pad_seconds

    infer_tool.fill_a_to_b(trans, clean_names)
    for clean_name, tran in zip(clean_names, trans):
        raw_audio_path = f"raw/{clean_name}"
        if "." not in raw_audio_path:
            raw_audio_path += ".wav"
        infer_tool.format_wav(raw_audio_path)
        wav_path = Path(raw_audio_path).with_suffix('.wav')
        chunks = slicer.cut(wav_path, db_thresh=slice_db)
        audio_data, audio_sr = slicer.chunks2audio(wav_path, chunks)

        for spk in spk_list:
            audio = []
            for (slice_tag, data) in audio_data:
                print(f'#=====segment start, {round(len(data) / audio_sr, 3)}s======')

                length = int(np.ceil(len(data) / audio_sr * svc_model.target_sample))
                if slice_tag:
                    print('jump empty segment')
                    _audio = np.zeros(length)
                    
                elif length >= 15:
                    # pad
                    pad_len = int(audio_sr * pad_seconds)
                    # if chunk size is too large more than 15 seconds break it into smaller pieces because sliced by large amounts
                    # final piece
                    final_cross_faded_chunk_wav_tensor = None
                    # chunk into smaller pieces cross fade 
                    # Set the length of each chunk in seconds
                    chunk_length = 15
                    # Calculate the number of samples in each chunk
                    chunk_samples = chunk_length * audio_sr
                    # Calculate the total number of chunks
                    total_chunks = int(len(data) / chunk_samples)
                    # Calculate the number of remaining samples
                    remaining_samples = len(data) - (total_chunks * chunk_samples)
                    # Add an additional chunk for any remaining samples
                    if remaining_samples > 0:
                        total_chunks += 1
                    # Slice the WAV tensor into chunks
                    chunks = [data[i * chunk_samples:min((i + 1) * chunk_samples, len(data))] for i in range(total_chunks)]
                    
                    for chunk in chunks:
                        # prepare for inference
                        data = np.concatenate([np.zeros([pad_len]), chunk, np.zeros([pad_len])])
                        
                        # Testing writes to the audio to do a match up.
                        
                        soundfile.write(raw_path, data, audio_sr, format="wav")
                        raw_path.seek(0)     

                        out_audio, out_sr = svc_model.infer(spk, tran, raw_path,
                                                            cluster_infer_ratio=cluster_infer_ratio,
                                                            auto_predict_f0=auto_predict_f0,
                                                            noice_scale=noice_scale
                                                            )
                        #trim padded audio by new sample length and cross fade
                        pad_len = int(svc_model.target_sample * pad_seconds)
                        _audio = _audio[pad_len:-pad_len]
                        _audio = out_audio.cpu().numpy()
                        
                        if final_cross_faded_chunk_wav_tensor == None:
                            final_cross_faded_chunk_wav_tensor = _audio
                        else:
                            final_cross_faded_chunk_wav_tensor = stitch_wav_tensors_with_crossfade(wav_tensor1=final_cross_faded_chunk_wav_tensor,wav_tensor2=_audio,sample_rate=svc_model.target_sample)
                        
                        _audio = final_cross_faded_chunk_wav_tensor
                        
                    else:
                        data = np.concatenate([np.zeros([pad_len]), data, np.zeros([pad_len])])
                        raw_path = io.BytesIO()
                        
                        soundfile.write(raw_path, data, audio_sr, format="wav")
                        raw_path.seek(0)          
                        
                        out_audio, out_sr = svc_model.infer(spk, tran, raw_path,
                                                            cluster_infer_ratio=cluster_infer_ratio,
                                                            auto_predict_f0=auto_predict_f0,
                                                            noice_scale=noice_scale
                                                            )
                        _audio = out_audio.cpu().numpy()
                        
                        #trim padded audio by new sample length
                        pad_len = int(svc_model.target_sample * pad_seconds)
                        # trim padded both sides with new sample ra
                        _audio = _audio[pad_len:-pad_len]

                audio.extend(list(infer_tool.pad_array(_audio, length)))
            key = "auto" if auto_predict_f0 else f"{tran}key"
            cluster_name = "" if cluster_infer_ratio == 0 else f"_{cluster_infer_ratio}"
            res_path = f'./results/{clean_name}_{key}_{spk}{cluster_name}.{wav_format}'
            soundfile.write(res_path, audio, svc_model.target_sample, format=wav_format)

if __name__ == '__main__':
    main()