data_utils.py

# modified from https://github.com/jaywalnut310/vits
import os
import random
import torch
import torch.utils.data

import commons
from mel_processing import spectrogram_torch
from utils import load_wav_to_torch, load_filepaths_and_text
from text import text_to_sequence
from analysis import Pitch
""" Modified from Multi speaker version of VITS"""


class TextAudioSpeakerLoader(torch.utils.data.Dataset):
    """
        1) loads audio, speaker_id, text pairs
        2) normalizes text and converts them to sequences of integers
        3) computes spectrograms from audio files.
    """

    def __init__(self, audiopaths_sid_text, hparams, pt_run=False):
        self.audiopaths_sid_text = load_filepaths_and_text(audiopaths_sid_text)
        self.text_cleaners = hparams.text_cleaners
        self.sampling_rate = hparams.sampling_rate
        self.filter_length = hparams.filter_length
        self.hop_length = hparams.hop_length
        self.win_length = hparams.win_length

        self.lang = hparams.languages

        self.add_blank = hparams.add_blank
        self.min_text_len = getattr(hparams, "min_text_len", 1)
        self.max_text_len = getattr(hparams, "max_text_len", 190)

        self.speaker_dict = {
            speaker: idx
            for idx, speaker in enumerate(hparams.speakers)
        }
        self.data_path = hparams.data_path

        self.pitch = Pitch(sr=hparams.sampling_rate,
                           W=hparams.tau_max,
                           tau_max=hparams.tau_max,
                           midi_start=hparams.midi_start,
                           midi_end=hparams.midi_end,
                           octave_range=hparams.octave_range)

        random.seed(1234)
        random.shuffle(self.audiopaths_sid_text)
        self._filter()
        if pt_run:
            for _audiopaths_sid_text in self.audiopaths_sid_text:
                _ = self.get_audio_text_speaker_pair(_audiopaths_sid_text,
                                                     True)

    def _filter(self):
        """
        Filter text & store spec lengths
        """
        # Store spectrogram lengths for Bucketing
        # wav_length ~= file_size / (wav_channels * Bytes per dim) = file_size / (1 * 2)
        # spec_length = wav_length // hop_length

        audiopaths_sid_text_new = []
        lengths = []
        for audiopath, text, spk in self.audiopaths_sid_text:
            if self.min_text_len <= len(text) and len(
                    text) <= self.max_text_len:
                audiopath = os.path.join(self.data_path, audiopath)
                audiopaths_sid_text_new.append([audiopath, text, spk])
                lengths.append(
                    os.path.getsize(audiopath) // (2 * self.hop_length))
        self.audiopaths_sid_text = audiopaths_sid_text_new
        self.lengths = lengths

    def get_audio_text_speaker_pair(self, audiopath_sid_text, pt_run=False):
        # separate filename, speaker_id and text
        audiopath, text, spk = audiopath_sid_text[0], audiopath_sid_text[
            1], audiopath_sid_text[2]
        text, tone = self.get_text(text)
        spec, ying, wav = self.get_audio(audiopath, pt_run)
        sid = self.get_sid(self.speaker_dict[spk])
        return (text, spec, ying, wav, sid, tone)

    def get_audio(self, filename, pt_run=False):
        audio, sampling_rate = load_wav_to_torch(filename)
        if sampling_rate != self.sampling_rate:
            raise ValueError("{} {} SR doesn't match target {} SR".format(
                sampling_rate, self.sampling_rate))
        audio_norm = audio.unsqueeze(0)
        spec_filename = filename.replace(".wav", ".spec.pt")
        ying_filename = filename.replace(".wav", ".ying.pt")
        if os.path.exists(spec_filename) and not pt_run:
            spec = torch.load(spec_filename, map_location='cpu')
        else:
            spec = spectrogram_torch(audio_norm,
                                     self.filter_length,
                                     self.sampling_rate,
                                     self.hop_length,
                                     self.win_length,
                                     center=False)
            spec = torch.squeeze(spec, 0)
            torch.save(spec, spec_filename)
        if os.path.exists(ying_filename) and not pt_run:
            ying = torch.load(ying_filename, map_location='cpu')
        else:
            wav = torch.nn.functional.pad(
                audio_norm.unsqueeze(0),
                (self.filter_length - self.hop_length,
                 self.filter_length - self.hop_length +
                 (-audio_norm.shape[1]) % self.hop_length + self.hop_length * (audio_norm.shape[1] % self.hop_length == 0)),
                mode='constant').squeeze(0)
            ying = self.pitch.yingram(wav)[0]
            torch.save(ying, ying_filename)
        return spec, ying, audio_norm

    def get_text(self, text):
        text_norm, tone = text_to_sequence(text, self.lang)
        if self.add_blank:
            text_norm = commons.intersperse(text_norm, 0)
            tone = commons.intersperse(tone, 0)
        text_norm = torch.LongTensor(text_norm)
        tone = torch.LongTensor(tone)
        return text_norm, tone

    def get_sid(self, sid):
        sid = torch.LongTensor([int(sid)])
        return sid

    def __getitem__(self, index):
        return self.get_audio_text_speaker_pair(
            self.audiopaths_sid_text[index])

    def __len__(self):
        return len(self.audiopaths_sid_text)


class TextAudioSpeakerCollate():
    """ Zero-pads model inputs and targets"""

    def __init__(self, return_ids=False):
        self.return_ids = return_ids

    def __call__(self, batch):
        """Collate's training batch from normalized text, audio and speaker identities
        PARAMS
        ------
        batch: [text_normalized, spec_normalized, wav_normalized, sid]
        """
        # Right zero-pad all one-hot text sequences to max input length
        _, ids_sorted_decreasing = torch.sort(torch.LongTensor(
            [x[1].size(1) for x in batch]),
                                              dim=0,
                                              descending=True)

        max_text_len = max([len(x[0]) for x in batch])
        max_spec_len = max([x[1].size(1) for x in batch])
        max_ying_len = max([x[2].size(1) for x in batch])
        max_wav_len = max([x[3].size(1) for x in batch])

        text_lengths = torch.LongTensor(len(batch))
        spec_lengths = torch.LongTensor(len(batch))
        ying_lengths = torch.LongTensor(len(batch))
        wav_lengths = torch.LongTensor(len(batch))
        sid = torch.LongTensor(len(batch))

        text_padded = torch.LongTensor(len(batch), max_text_len)
        tone_padded = torch.LongTensor(len(batch), max_text_len)
        spec_padded = torch.FloatTensor(len(batch), batch[0][1].size(0),
                                        max_spec_len)
        ying_padded = torch.FloatTensor(len(batch), batch[0][2].size(0),
                                        max_ying_len)
        wav_padded = torch.FloatTensor(len(batch), 1, max_wav_len)
        text_padded.zero_()
        tone_padded.zero_()
        spec_padded.zero_()
        ying_padded.zero_()
        wav_padded.zero_()
        for i in range(len(ids_sorted_decreasing)):
            row = batch[ids_sorted_decreasing[i]]

            text = row[0]
            text_padded[i, :text.size(0)] = text
            text_lengths[i] = text.size(0)

            spec = row[1]
            spec_padded[i, :, :spec.size(1)] = spec
            spec_lengths[i] = spec.size(1)

            ying = row[2]
            ying_padded[i, :, :ying.size(1)] = ying
            ying_lengths[i] = ying.size(1)

            wav = row[3]
            wav_padded[i, :, :wav.size(1)] = wav
            wav_lengths[i] = wav.size(1)

            tone = row[5]
            tone_padded[i, :text.size(0)] = tone

            sid[i] = row[4]

        if self.return_ids:
            return text_padded, text_lengths, spec_padded, spec_lengths, wav_padded, wav_lengths, sid, ids_sorted_decreasing
        return text_padded, text_lengths, spec_padded, spec_lengths, ying_padded, ying_lengths, wav_padded, wav_lengths, sid, tone_padded


class DistributedBucketSampler(torch.utils.data.distributed.DistributedSampler
                               ):
    """
    Maintain similar input lengths in a batch.
    Length groups are specified by boundaries.
    Ex) boundaries = [b1, b2, b3] -> any batch is included either {x | b1 < length(x) <=b2} or {x | b2 < length(x) <= b3}.

    It removes samples which are not included in the boundaries.
    Ex) boundaries = [b1, b2, b3] -> any x s.t. length(x) <= b1 or length(x) > b3 are discarded.
    """

    def __init__(self,
                 dataset,
                 batch_size,
                 boundaries,
                 num_replicas=None,
                 rank=None,
                 shuffle=True):
        super().__init__(dataset,
                         num_replicas=num_replicas,
                         rank=rank,
                         shuffle=shuffle)
        self.lengths = dataset.lengths
        self.batch_size = batch_size
        self.boundaries = boundaries

        self.buckets, self.num_samples_per_bucket = self._create_buckets()
        self.total_size = sum(self.num_samples_per_bucket)
        self.num_samples = self.total_size // self.num_replicas

    def _create_buckets(self):
        buckets = [[] for _ in range(len(self.boundaries) - 1)]
        for i in range(len(self.lengths)):
            length = self.lengths[i]
            idx_bucket = self._bisect(length)
            if idx_bucket != -1:
                buckets[idx_bucket].append(i)

        for i in range(len(buckets) - 1, -1, -1):
            if len(buckets[i]) == 0:
                buckets.pop(i)
                self.boundaries.pop(i + 1)

        num_samples_per_bucket = []
        for i in range(len(buckets)):
            len_bucket = len(buckets[i])
            total_batch_size = self.num_replicas * self.batch_size
            rem = (total_batch_size -
                   (len_bucket % total_batch_size)) % total_batch_size
            num_samples_per_bucket.append(len_bucket + rem)
        return buckets, num_samples_per_bucket

    def __iter__(self):
        # deterministically shuffle based on epoch
        g = torch.Generator()
        g.manual_seed(self.epoch)

        indices = []
        if self.shuffle:
            for bucket in self.buckets:
                indices.append(
                    torch.randperm(len(bucket), generator=g).tolist())
        else:
            for bucket in self.buckets:
                indices.append(list(range(len(bucket))))

        batches = []
        for i in range(len(self.buckets)):
            bucket = self.buckets[i]
            len_bucket = len(bucket)
            ids_bucket = indices[i]
            num_samples_bucket = self.num_samples_per_bucket[i]

            # add extra samples to make it evenly divisible
            rem = num_samples_bucket - len_bucket
            ids_bucket = ids_bucket + ids_bucket * \
                (rem // len_bucket) + ids_bucket[:(rem % len_bucket)]

            # subsample
            ids_bucket = ids_bucket[self.rank::self.num_replicas]

            # batching
            for j in range(len(ids_bucket) // self.batch_size):
                batch = [
                    bucket[idx]
                    for idx in ids_bucket[j * self.batch_size:(j + 1) *
                                          self.batch_size]
                ]
                batches.append(batch)

        if self.shuffle:
            batch_ids = torch.randperm(len(batches), generator=g).tolist()
            batches = [batches[i] for i in batch_ids]
        self.batches = batches

        assert len(self.batches) * self.batch_size == self.num_samples
        return iter(self.batches)

    def _bisect(self, x, lo=0, hi=None):
        if hi is None:
            hi = len(self.boundaries) - 1

        if hi > lo:
            mid = (hi + lo) // 2
            if self.boundaries[mid] < x and x <= self.boundaries[mid + 1]:
                return mid
            elif x <= self.boundaries[mid]:
                return self._bisect(x, lo, mid)
            else:
                return self._bisect(x, mid + 1, hi)
        else:
            return -1

    def __len__(self):
        return self.num_samples // self.batch_size


def create_spec(audiopaths_sid_text, hparams):
    audiopaths_sid_text = load_filepaths_and_text(audiopaths_sid_text)
    for audiopath, _, _ in audiopaths_sid_text:
        audiopath = os.path.join(hparams.data_path, audiopath)
        audio, sampling_rate = load_wav_to_torch(audiopath)
        if sampling_rate != hparams.sampling_rate:
            raise ValueError("{} {} SR doesn't match target {} SR".format(
                sampling_rate, hparams.sampling_rate))
        audio_norm = audio.unsqueeze(0)
        specpath = audiopath.replace(".wav", ".spec.pt")

        if not os.path.exists(specpath):
            spec = spectrogram_torch(audio_norm,
                                     hparams.filter_length,
                                     hparams.sampling_rate,
                                     hparams.hop_length,
                                     hparams.win_length,
                                     center=False)
            spec = torch.squeeze(spec, 0)
            torch.save(spec, specpath)