fix(data_utils): allow float32 audio to be processed properly (#115)

src/so_vits_svc_fork/data_utils.py

@@ -4,10 +4,11 @@
 import numpy as np
 import torch
 import torch.utils.data
+import torchaudio
 
 from . import utils
 from .modules.mel_processing import spectrogram_torch
-from .utils import load_filepaths_and_text, load_wav_to_torch
+from .utils import load_filepaths_and_text
 
 # import h5py
 
@@ -24,7 +25,6 @@ class TextAudioSpeakerLoader(torch.utils.data.Dataset):
 
     def __init__(self, audiopaths, hparams):
         self.audiopaths = load_filepaths_and_text(audiopaths)
-        self.max_wav_value = hparams.data.max_wav_value
         self.sampling_rate = hparams.data.sampling_rate
         self.filter_length = hparams.data.filter_length
         self.hop_length = hparams.data.hop_length
@@ -39,15 +39,13 @@ def __init__(self, audiopaths, hparams):
 
     def get_audio(self, filename):
         filename = filename.replace("\\", "/")
-        audio, sampling_rate = load_wav_to_torch(filename)
+        audio_norm, sampling_rate = torchaudio.load(filename)
         if sampling_rate != self.sampling_rate:
             raise ValueError(
                 "{} SR doesn't match target {} SR".format(
                     sampling_rate, self.sampling_rate
                 )
             )
-        audio_norm = audio / self.max_wav_value
-        audio_norm = audio_norm.unsqueeze(0)
         spec_filename = filename.replace(".wav", ".spec.pt")
         if os.path.exists(spec_filename):
             spec = torch.load(spec_filename)

src/so_vits_svc_fork/preprocess_flist_config.py

@@ -3,26 +3,17 @@
 import json
 import os
 import re
-import wave
 from copy import deepcopy
 from logging import getLogger
 from pathlib import Path
 from random import shuffle
 
+from librosa import get_duration
 from tqdm import tqdm
 
 LOG = getLogger(__name__)
 
 
-def _get_wav_duration(filepath: Path):
-    with open(filepath, "rb") as f:
-        with wave.open(f) as wav_file:
-            n_frames = wav_file.getnframes()
-            framerate = wav_file.getframerate()
-            duration = n_frames / float(framerate)
-    return duration
-
-
 def preprocess_config(
     input_dir: Path | str,
     train_list_path: Path | str,
@@ -48,7 +39,7 @@ def preprocess_config(
             pattern = re.compile(r"^[\.a-zA-Z0-9_\/]+$")
             if not pattern.match(path.name):
                 LOG.warning(f"file name {path} contains non-alphanumeric characters.")
-            if _get_wav_duration(path) < 0.3:
+            if get_duration(filename=path) < 0.3:
                 LOG.warning(f"skip {path} because it is too short.")
                 continue
             paths.append(path)