DL-Art-School/codes/data/audio/unsupervised_audio_dataset.py

200 lines
7.9 KiB
Python
Raw Normal View History

2021-09-14 23:43:16 +00:00
import os
import pathlib
import random
import sys
from warnings import warn
2021-09-14 23:43:16 +00:00
import torch
import torch.utils.data
import torch.nn.functional as F
import torchaudio
from audio2numpy import open_audio
from tqdm import tqdm
from data.audio.wav_aug import WavAugmentor
from data.util import find_files_of_type, is_wav_file, is_audio_file, load_paths_from_cache
2021-09-14 23:43:16 +00:00
from models.tacotron2.taco_utils import load_wav_to_torch
from utils.util import opt_get
def load_audio(audiopath, sampling_rate):
2021-09-15 00:29:17 +00:00
if audiopath[-4:] == '.wav':
2021-09-14 23:43:16 +00:00
audio, lsr = load_wav_to_torch(audiopath)
elif audiopath[-4:] == '.mp3':
# https://github.com/neonbjb/pyfastmp3decoder - Definitely worth it.
from pyfastmp3decoder.mp3decoder import load_mp3
audio, lsr = load_mp3(audiopath, sampling_rate)
audio = torch.FloatTensor(audio)
2021-09-14 23:43:16 +00:00
else:
audio, lsr = open_audio(audiopath)
audio = torch.FloatTensor(audio)
# Remove any channel data.
if len(audio.shape) > 1:
if audio.shape[0] < 5:
audio = audio[0]
else:
assert audio.shape[1] < 5
audio = audio[:, 0]
if lsr != sampling_rate:
2022-01-05 22:47:22 +00:00
audio = torchaudio.functional.resample(audio, lsr, sampling_rate)
2021-09-14 23:43:16 +00:00
# Check some assumptions about audio range. This should be automatically fixed in load_wav_to_torch, but might not be in some edge cases, where we should squawk.
# '2' is arbitrarily chosen since it seems like audio will often "overdrive" the [-1,1] bounds.
if torch.any(audio > 2) or not torch.any(audio < 0):
print(f"Error with {audiopath}. Max={audio.max()} min={audio.min()}")
audio.clip_(-1, 1)
return audio.unsqueeze(0)
def load_similar_clips(path, sample_length, sample_rate, n=3, include_self=True, fallback_to_self=True):
sim_path = os.path.join(os.path.dirname(path), 'similarities.pth')
candidates = []
if os.path.exists(sim_path):
similarities = torch.load(sim_path)
fname = os.path.basename(path)
if fname in similarities.keys():
candidates = [os.path.join(os.path.dirname(path), s) for s in similarities[fname]]
else:
print(f'Similarities list found for {path} but {fname} was not in that list.')
if len(candidates) == 0:
if fallback_to_self:
candidates = [path]
else:
candidates = find_files_of_type('img', os.path.dirname(path), qualifier=is_audio_file)[0]
assert len(candidates) < 50000 # Sanity check to ensure we aren't loading "related files" that aren't actually related.
if not include_self:
candidates.remove(path)
if len(candidates) == 0:
print(f"No conditioning candidates found for {path}")
raise NotImplementedError()
# Sample with replacement. This can get repeats, but more conveniently handles situations where there are not enough candidates.
related_clips = []
for k in range(n):
rel_clip = load_audio(random.choice(candidates), sample_rate)
gap = rel_clip.shape[-1] - sample_length
if gap < 0:
rel_clip = F.pad(rel_clip, pad=(0, abs(gap)))
elif gap > 0:
rand_start = random.randint(0, gap)
rel_clip = rel_clip[:, rand_start:rand_start+sample_length]
related_clips.append(rel_clip)
if n > 1:
return torch.stack(related_clips, dim=0)
else:
return related_clips[0]
2021-09-14 23:43:16 +00:00
class UnsupervisedAudioDataset(torch.utils.data.Dataset):
def __init__(self, opt):
path = opt['path']
cache_path = opt['cache_path'] # Will fail when multiple paths specified, must be specified in this case.
exclusions = []
if 'exclusions' in opt.keys():
for exc in opt['exclusions']:
with open(exc, 'r') as f:
exclusions.extend(f.read().splitlines())
self.audiopaths = load_paths_from_cache(path, cache_path, exclusions)
2021-09-14 23:43:16 +00:00
# Parse options
self.sampling_rate = opt_get(opt, ['sampling_rate'], 22050)
self.pad_to = opt_get(opt, ['pad_to_seconds'], None)
if self.pad_to is not None:
self.pad_to *= self.sampling_rate
self.pad_to = opt_get(opt, ['pad_to_samples'], self.pad_to)
2021-12-28 23:18:12 +00:00
self.min_length = opt_get(opt, ['min_length'], 0)
2021-09-14 23:43:16 +00:00
2021-10-26 16:42:23 +00:00
# "Resampled clip" is audio data pulled from the basis of "clip" but with randomly different bounds. There are no
# guarantees that "clip_resampled" is different from "clip": in fact, if "clip" is less than pad_to_seconds/samples,
self.should_resample_clip = opt_get(opt, ['resample_clip'], False)
# "Extra samples" are other audio clips pulled from wav files in the same directory as the 'clip' wav file.
2021-09-14 23:43:16 +00:00
self.extra_samples = opt_get(opt, ['extra_samples'], 0)
self.extra_sample_len = opt_get(opt, ['extra_sample_length'], 44000)
2021-09-14 23:43:16 +00:00
2021-12-28 23:18:12 +00:00
self.debug_loading_failures = opt_get(opt, ['debug_loading_failures'], True)
2021-09-14 23:43:16 +00:00
def get_audio_for_index(self, index):
audiopath = self.audiopaths[index]
audio = load_audio(audiopath, self.sampling_rate)
2021-12-28 23:18:12 +00:00
assert audio.shape[1] > self.min_length
2021-09-14 23:43:16 +00:00
return audio, audiopath
def get_related_audio_for_index(self, index):
if self.extra_samples <= 0:
return None, 0
2021-09-14 23:43:16 +00:00
audiopath = self.audiopaths[index]
return load_similar_clips(audiopath, self.extra_sample_len, self.sampling_rate, n=self.extra_samples)
2021-09-14 23:43:16 +00:00
def __getitem__(self, index):
try:
# Split audio_norm into two tensors of equal size.
audio_norm, filename = self.get_audio_for_index(index)
alt_files = self.get_related_audio_for_index(index)
except:
2021-12-28 23:18:12 +00:00
if self.debug_loading_failures:
print(f"Error loading audio for file {self.audiopaths[index]} {sys.exc_info()}")
return self[index+1]
2021-09-14 23:43:16 +00:00
2021-10-26 16:42:23 +00:00
# When generating resampled clips, skew is a bias that tries to spread them out from each other, reducing their
# influence on one another.
skew = [-1, 1] if self.should_resample_clip else [0]
# To increase variability, which skew is applied to the clip and resampled_clip is randomized.
random.shuffle(skew)
clips = []
for sk in skew:
if self.pad_to is not None:
if audio_norm.shape[-1] <= self.pad_to:
clips.append(torch.nn.functional.pad(audio_norm, (0, self.pad_to - audio_norm.shape[-1])))
else:
gap = audio_norm.shape[-1] - self.pad_to
start = min(max(random.randint(0, gap-1) + sk * gap // 2, 0), gap-1)
clips.append(audio_norm[:, start:start+self.pad_to])
else:
clips.append(audio_norm)
2021-09-14 23:43:16 +00:00
output = {
2021-10-26 16:42:23 +00:00
'clip': clips[0],
2022-01-01 07:23:46 +00:00
'clip_lengths': torch.tensor(audio_norm.shape[-1]),
2021-09-14 23:43:16 +00:00
'path': filename,
}
2021-10-26 16:42:23 +00:00
if self.should_resample_clip:
output['resampled_clip'] = clips[1]
if self.extra_samples > 0:
output['alt_clips'] = alt_files
2021-09-14 23:43:16 +00:00
return output
def __len__(self):
return len(self.audiopaths)
if __name__ == '__main__':
params = {
'mode': 'unsupervised_audio',
'path': ['\\\\192.168.5.3\\rtx3080_audio\\split\\cleaned\\books0'],
'cache_path': 'E:\\audio\\remote-cache3.pth',
2021-09-14 23:43:16 +00:00
'sampling_rate': 22050,
2021-10-17 23:32:46 +00:00
'pad_to_samples': 40960,
2021-09-14 23:43:16 +00:00
'phase': 'train',
2021-10-17 23:32:46 +00:00
'n_workers': 1,
2021-09-14 23:43:16 +00:00
'batch_size': 16,
'extra_samples': 4,
2021-10-26 16:42:23 +00:00
'resample_clip': True,
2021-09-14 23:43:16 +00:00
}
from data import create_dataset, create_dataloader, util
ds = create_dataset(params)
dl = create_dataloader(ds, params)
i = 0
for b in tqdm(dl):
2021-10-17 23:32:46 +00:00
for b_ in range(b['clip'].shape[0]):
#pass
torchaudio.save(f'{i}_clip_{b_}.wav', b['clip'][b_], ds.sampling_rate)
2021-10-26 16:42:23 +00:00
torchaudio.save(f'{i}_resampled_clip_{b_}.wav', b['resampled_clip'][b_], ds.sampling_rate)
2021-10-17 23:32:46 +00:00
i += 1