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

196 lines
7.5 KiB
Python
Raw Normal View History

import os
import random
import audio2numpy
import numpy as np
import torch
import torch.utils.data
2021-08-12 21:44:55 +00:00
import torch.nn.functional as F
2021-10-31 21:01:38 +00:00
import torchaudio
2021-07-28 02:33:30 +00:00
from tqdm import tqdm
import models.tacotron2.layers as layers
2021-10-31 21:01:38 +00:00
from data.audio.unsupervised_audio_dataset import load_audio
from models.tacotron2.taco_utils import load_wav_to_torch, load_filepaths_and_text
from models.tacotron2.text import text_to_sequence
2021-07-28 02:33:30 +00:00
from utils.util import opt_get
def load_mozilla_cv(filename):
with open(filename, encoding='utf-8') as f:
components = [line.strip().split('\t') for line in f][1:] # First line is the header
base = os.path.dirname(filename)
filepaths_and_text = [[os.path.join(base, f'clips/{component[1]}'), component[2]] for component in components]
return filepaths_and_text
def load_voxpopuli(filename):
with open(filename, encoding='utf-8') as f:
lines = [line.strip().split('\t') for line in f][1:] # First line is the header
base = os.path.dirname(filename)
filepaths_and_text = []
for line in lines:
if len(line) == 0:
continue
file, raw_text, norm_text, speaker_id, split, gender = line
year = file[:4]
filepaths_and_text.append([os.path.join(base, year, f'{file}.ogg.wav'), raw_text])
return filepaths_and_text
2021-10-31 21:01:38 +00:00
class TextWavLoader(torch.utils.data.Dataset):
def __init__(self, hparams):
self.path = hparams['path']
if not isinstance(self.path, list):
self.path = [self.path]
fetcher_mode = opt_get(hparams, ['fetcher_mode'], 'lj')
if not isinstance(fetcher_mode, list):
fetcher_mode = [fetcher_mode]
assert len(self.path) == len(fetcher_mode)
self.audiopaths_and_text = []
for p, fm in zip(self.path, fetcher_mode):
if fm == 'lj' or fm == 'libritts':
fetcher_fn = load_filepaths_and_text
elif fm == 'mozilla_cv':
fetcher_fn = load_mozilla_cv
elif fm == 'voxpopuli':
fetcher_fn = load_voxpopuli
else:
raise NotImplementedError()
self.audiopaths_and_text.extend(fetcher_fn(p))
self.text_cleaners = hparams.text_cleaners
2021-10-31 21:01:38 +00:00
self.sample_rate = hparams.sample_rate
random.seed(hparams.seed)
random.shuffle(self.audiopaths_and_text)
2021-10-31 21:01:38 +00:00
self.max_wav_len = opt_get(hparams, ['max_wav_length'], None)
2021-08-12 21:44:55 +00:00
self.max_text_len = opt_get(hparams, ['max_text_length'], None)
# If needs_collate=False, all outputs will be aligned and padded at maximum length.
self.needs_collate = opt_get(hparams, ['needs_collate'], True)
if not self.needs_collate:
2021-10-31 21:01:38 +00:00
assert self.max_wav_len is not None and self.max_text_len is not None
2021-10-31 21:01:38 +00:00
def get_wav_text_pair(self, audiopath_and_text):
# separate filename and text
audiopath, text = audiopath_and_text[0], audiopath_and_text[1]
2021-08-14 20:37:17 +00:00
text_seq = self.get_text(text)
2021-10-31 21:01:38 +00:00
wav = load_audio(audiopath, self.sample_rate)
return (text_seq, wav, text, audiopath_and_text[0])
def get_text(self, text):
text_norm = torch.IntTensor(text_to_sequence(text, self.text_cleaners))
return text_norm
def __getitem__(self, index):
2021-11-03 06:31:50 +00:00
try:
tseq, wav, text, path = self.get_wav_text_pair(self.audiopaths_and_text[index])
except:
2021-11-03 06:33:38 +00:00
print(f"error loadding {self.audiopaths_and_text[index][0]}")
2021-11-03 06:31:50 +00:00
return self[index+1]
2021-10-31 21:01:38 +00:00
if wav is None or \
(self.max_wav_len is not None and wav.shape[-1] > self.max_wav_len) or \
2021-08-14 20:37:17 +00:00
(self.max_text_len is not None and tseq.shape[0] > self.max_text_len):
2021-10-31 21:01:38 +00:00
# Basically, this audio file is nonexistent or too long to be supported by the dataset.
# It's hard to handle this situation properly. Best bet is to return the a random valid token and skew the dataset somewhat as a result.
2021-10-31 21:01:38 +00:00
#if wav is not None:
# print(f"Exception {index} wav_len:{wav.shape[-1]} text_len:{tseq.shape[0]} fname: {path}")
rv = random.randint(0,len(self)-1)
return self[rv]
2021-10-31 21:01:38 +00:00
orig_output = wav.shape[-1]
2021-08-14 20:37:17 +00:00
orig_text_len = tseq.shape[0]
2021-08-12 21:44:55 +00:00
if not self.needs_collate:
2021-10-31 21:01:38 +00:00
if wav.shape[-1] != self.max_wav_len:
wav = F.pad(wav, (0, self.max_wav_len - wav.shape[-1]))
2021-08-14 20:37:17 +00:00
if tseq.shape[0] != self.max_text_len:
tseq = F.pad(tseq, (0, self.max_text_len - tseq.shape[0]))
2021-08-12 21:51:23 +00:00
return {
2021-08-14 20:37:17 +00:00
'real_text': text,
'padded_text': tseq,
2021-08-12 21:51:23 +00:00
'input_lengths': torch.tensor(orig_text_len, dtype=torch.long),
2021-10-31 21:01:38 +00:00
'wav': wav,
2021-08-12 21:51:23 +00:00
'output_lengths': torch.tensor(orig_output, dtype=torch.long),
2021-08-14 20:37:17 +00:00
'filenames': path
2021-08-12 21:51:23 +00:00
}
2021-10-31 21:01:38 +00:00
return tseq, wav, path, text
def __len__(self):
return len(self.audiopaths_and_text)
class TextMelCollate():
2021-10-31 21:01:38 +00:00
""" Zero-pads model inputs and targets based on number of frames per step
"""
def __call__(self, batch):
2021-10-31 21:01:38 +00:00
"""Collate's training batch from normalized text and wav
PARAMS
------
2021-10-31 21:01:38 +00:00
batch: [text_normalized, wav, filename, text]
"""
# Right zero-pad all one-hot text sequences to max input length
input_lengths, ids_sorted_decreasing = torch.sort(
torch.LongTensor([len(x[0]) for x in batch]),
dim=0, descending=True)
max_input_len = input_lengths[0]
text_padded = torch.LongTensor(len(batch), max_input_len)
text_padded.zero_()
filenames = []
2021-08-14 20:37:17 +00:00
real_text = []
for i in range(len(ids_sorted_decreasing)):
text = batch[ids_sorted_decreasing[i]][0]
text_padded[i, :text.size(0)] = text
filenames.append(batch[ids_sorted_decreasing[i]][2])
2021-08-14 20:37:17 +00:00
real_text.append(batch[ids_sorted_decreasing[i]][3])
2021-10-31 21:01:38 +00:00
# Right zero-pad wav
num_wavs = batch[0][1].size(0)
max_target_len = max([x[1].size(1) for x in batch])
# include mel padded and gate padded
2021-10-31 21:01:38 +00:00
wav_padded = torch.FloatTensor(len(batch), num_wavs, max_target_len)
wav_padded.zero_()
output_lengths = torch.LongTensor(len(batch))
for i in range(len(ids_sorted_decreasing)):
2021-10-31 21:01:38 +00:00
wav = batch[ids_sorted_decreasing[i]][1]
wav_padded[i, :, :wav.size(1)] = wav
output_lengths[i] = wav.size(1)
2021-07-09 04:13:44 +00:00
return {
'padded_text': text_padded,
'input_lengths': input_lengths,
2021-10-31 21:01:38 +00:00
'wav': wav_padded,
'output_lengths': output_lengths,
2021-08-14 20:37:17 +00:00
'filenames': filenames,
'real_text': real_text,
2021-07-09 04:13:44 +00:00
}
if __name__ == '__main__':
2021-10-31 21:01:38 +00:00
batch_sz = 32
params = {
'mode': 'nv_tacotron',
2021-10-31 21:01:38 +00:00
'path': 'E:\\audio\\MozillaCommonVoice\\en\\test.tsv',
'phase': 'train',
2021-10-31 21:01:38 +00:00
'n_workers': 0,
'batch_size': batch_sz,
'fetcher_mode': 'mozilla_cv',
2021-10-31 21:01:38 +00:00
'needs_collate': True,
#'max_wav_length': 256000,
#'max_text_length': 200,
'sample_rate': 22050,
}
from data import create_dataset, create_dataloader
ds, c = create_dataset(params, return_collate=True)
dl = create_dataloader(ds, params, collate_fn=c)
i = 0
2021-08-06 18:03:46 +00:00
m = None
2021-08-13 15:36:31 +00:00
for k in range(1000):
for i, b in tqdm(enumerate(dl)):
2021-10-31 21:01:38 +00:00
w = b['wav']
for ib in range(batch_sz):
print(f'{i} {ib} {b["real_text"][ib]}')
torchaudio.save(f'{i}_clip_{ib}.wav', b['wav'][ib], ds.sample_rate)