train_first.py

from torch.utils.tensorboard import SummaryWriter

import os
import os.path as osp
import re
import sys
import yaml
import shutil
import numpy as np
import torch
import click
import warnings
warnings.simplefilter('ignore')

# load packages
import random
import yaml
from munch import Munch
import numpy as np
import torch
from torch import nn
import torch.nn.functional as F
import torchaudio
import librosa

from models import *
from meldataset import build_dataloader
from utils import *
from optimizers import build_optimizer
import time


# simple fix for dataparallel that allows access to class attributes
class MyDataParallel(torch.nn.DataParallel):
    def __getattr__(self, name):
        try:
            return super().__getattr__(name)
        except AttributeError:
            return getattr(self.module, name)

import logging
from logging import StreamHandler
logger = logging.getLogger(__name__)
logger.setLevel(logging.DEBUG)
handler = StreamHandler()
handler.setLevel(logging.DEBUG)
logger.addHandler(handler)

@click.command()
@click.option('-p', '--config_path', default='Configs/config.yml', type=str)
def main(config_path):

    config = yaml.safe_load(open(config_path))

    log_dir = config['log_dir']
    if not osp.exists(log_dir): os.makedirs(log_dir, exist_ok=True)
    shutil.copy(config_path, osp.join(log_dir, osp.basename(config_path)))
    writer = SummaryWriter(log_dir + "/tensorboard")

    # write logs
    file_handler = logging.FileHandler(osp.join(log_dir, 'train.log'))
    file_handler.setLevel(logging.DEBUG)
    file_handler.setFormatter(logging.Formatter('%(levelname)s:%(asctime)s: %(message)s'))
    logger.addHandler(file_handler)

    batch_size = config.get('batch_size', 10)
    device = config.get('device', 'cpu')
    epochs = config.get('epochs_1st', 200)
    save_freq = config.get('save_freq', 2)
    train_path = config.get('train_data', None)
    val_path = config.get('val_data', None)
    multigpu = config.get('multigpu', False)
    log_interval = config.get('log_interval', 10)
    saving_epoch = config.get('save_freq', 2)

    # load data
    train_list, val_list = get_data_path_list(train_path, val_path)

    train_dataloader = build_dataloader(train_list,
                                        batch_size=batch_size,
                                        num_workers=8,
                                        dataset_config={},
                                        device=device)

    val_dataloader = build_dataloader(val_list,
                                      batch_size=batch_size,
                                      validation=True,
                                      num_workers=2,
                                      device=device,
                                      dataset_config={})
    # load pretrained ASR model
    ASR_config = config.get('ASR_config', False)
    ASR_path = config.get('ASR_path', False)
    text_aligner = load_ASR_models(ASR_path, ASR_config)

    # load pretrained F0 model
    F0_path = config.get('F0_path', False)
    pitch_extractor = load_F0_models(F0_path)

    scheduler_params = {
        "max_lr": float(config['optimizer_params'].get('lr', 1e-4)),
        "pct_start": float(config['optimizer_params'].get('pct_start', 0.0)),
        "epochs": epochs,
        "steps_per_epoch": len(train_dataloader),
    }

    model = build_model(Munch(config['model_params']), text_aligner, pitch_extractor)

    _ = [model[key].to(device) for key in model]

    optimizer = build_optimizer({key: model[key].parameters() for key in model},
                                      scheduler_params_dict= {key: scheduler_params.copy() for key in model})

    # multi-GPU support
    if multigpu:
        for key in model:
            model[key] = MyDataParallel(model[key])

    if config.get('pretrained_model', '') != '':
        model, optimizer, start_epoch, iters = load_checkpoint(model,  optimizer, config['pretrained_model'],
                                    load_only_params=config.get('load_only_params', True))
    else:
        start_epoch = 0
        iters = 0

    best_loss = float('inf')  # best test loss
    loss_train_record = list([])
    loss_test_record = list([])

    loss_params = Munch(config['loss_params'])
    TMA_epoch = loss_params.TMA_epoch
    TMA_CEloss = loss_params.TMA_CEloss
    VC_epoch = loss_params.VC_epoch
    
    if VC_epoch < 0:
        VC_epoch = epochs + VC_epoch

    for epoch in range(start_epoch, epochs):
        running_loss = 0
        start_time = time.time()
        criterion = nn.L1Loss() 

        _ = [model[key].train() for key in model]

        for i, batch in enumerate(train_dataloader):

            batch = [b.to(device) for b in batch]
            texts, input_lengths, mels, mel_input_length, labels, ref_mels, ref_labels = batch

            mask = length_to_mask(mel_input_length // (2 ** model.text_aligner.n_down)).to('cuda')
            m = length_to_mask(input_lengths)

            text_mask = length_to_mask(input_lengths).to(texts.device)
            ppgs, s2s_pred, s2s_attn_feat = model.text_aligner(mels, mask, texts)

            s2s_attn_feat = s2s_attn_feat.transpose(-1, -2)
            s2s_attn_feat = s2s_attn_feat[..., 1:]
            s2s_attn_feat = s2s_attn_feat.transpose(-1, -2)
            
            with torch.no_grad():
                text_mask = length_to_mask(input_lengths).to(texts.device)
                attn_mask = (~mask).unsqueeze(-1).expand(mask.shape[0], mask.shape[1], text_mask.shape[-1]).float().transpose(-1, -2)
                attn_mask = attn_mask.float() * (~text_mask).unsqueeze(-1).expand(text_mask.shape[0], text_mask.shape[1], mask.shape[-1]).float()
                attn_mask = (attn_mask < 1)
                
            s2s_attn_feat.masked_fill_(attn_mask, -float("inf"))
            
            if TMA_CEloss:
                s2s_attn = F.softmax(s2s_attn_feat, dim=1) # along the mel dimension
            else:
                s2s_attn = F.softmax(s2s_attn_feat, dim=-1) # along the text dimension

            # get monotonic version 
            with torch.no_grad():
                mask_ST = mask_from_lens(s2s_attn, input_lengths, mel_input_length // (2 ** model.text_aligner.n_down))
                s2s_attn_mono = maximum_path(s2s_attn, mask_ST)

            s2s_attn = torch.nan_to_num(s2s_attn)

            # encode
            t_en = model.text_encoder(texts, input_lengths, m)

            # 50% of chance of using monotonic version
            if bool(random.getrandbits(1)):
                asr = (t_en @ s2s_attn)
            else:
                asr = (t_en @ s2s_attn_mono)

            # get clips
            mel_len = int(mel_input_length.min().item() / 2 - 1)
            en = []
            gt = []
            
            for bib in range(len(mel_input_length)):
                mel_length = int(mel_input_length[bib].item() / 2)
                
                random_start = np.random.randint(0, mel_length - mel_len)
                en.append(asr[bib, :, random_start:random_start+mel_len])
                gt.append(mels[bib, :, (random_start * 2):((random_start+mel_len) * 2)])
                
            mask = length_to_mask(torch.LongTensor([mel_len] * batch_size)).to('cuda')
            en = torch.stack(en)
            gt = torch.stack(gt).detach()

            # clip too short to be used by the style encoder
            if gt.shape[-1] < 80:
                continue
            
            real_norm = log_norm(gt.unsqueeze(1)).squeeze(1).detach()
            F0_real, _, _ = model.pitch_extractor(gt.unsqueeze(1))
            s = model.style_encoder(gt.unsqueeze(1))
            
            mel_rec = model.decoder(en, F0_real, real_norm, s)
            
            s_trg = model.style_encoder(ref_mels.unsqueeze(1))
            
            mel_fake = model.decoder(en, 
                                     F0_real, 
                                     real_norm, 
                                     s_trg)
            # discriminator loss
            optimizer.zero_grad()
            gt.requires_grad_()
            out, _ = model.discriminator(gt.unsqueeze(1), labels)
            loss_real = adv_loss(out, 1)
            loss_reg = r1_reg(out, gt)

            out, _ = model.discriminator(mel_rec.detach().unsqueeze(1), labels)
            loss_fake = adv_loss(out, 0)

            d_loss = loss_real + loss_fake + loss_reg * loss_params.lambda_reg
            d_loss.backward()
            optimizer.step('discriminator')

            # generator loss
            optimizer.zero_grad()
            loss_mel = criterion(mel_rec, gt)

            if epoch > TMA_epoch and epoch < VC_epoch: # start TMA training
                loss_s2s = 0
                for _s2s_pred, _text_input, _text_length in zip(s2s_pred, texts, input_lengths):
                    loss_s2s += F.cross_entropy(_s2s_pred[:_text_length], _text_input[:_text_length])
                loss_s2s /= texts.size(0)
                
                if TMA_CEloss:
                    # cross entropy loss for monotonic alignment
                    log_attn = torch.nan_to_num(F.log_softmax(s2s_attn_feat, dim=1)) # along the mel dimension
                    loss_mono = -(torch.mul(log_attn, s2s_attn_mono).sum(axis=[-1, -2]) / input_lengths).mean()
                else:
                    # L1 loss for monotonic alignment
                    loss_mono = F.l1_loss(s2s_attn, s2s_attn_mono) * 10
            else:
                loss_s2s = 0
                loss_mono = 0

            # adversarial loss
            with torch.no_grad():
                _, f_real = model.discriminator(gt.unsqueeze(1), labels)
            out_rec, f_fake = model.discriminator(mel_rec.unsqueeze(1), labels) # reconstructed sample
            loss_adv = adv_loss(out_rec, 1)
            
            # feature matching loss
            loss_fm = 0
            for m in range(len(f_real)):
                for k in range(len(f_real[m])):
                    loss_fm += torch.mean(torch.abs(f_real[m][k] - f_fake[m][k])) 

            # style loss
            s_pred = model.style_encoder(mel_fake.unsqueeze(1))
            loss_sty = torch.mean(torch.abs(s_pred - s_trg))
            
            # norm and F0 loss
            if epoch >= VC_epoch: # start VC training (for norm and F0 matching loss)
                # norm loss
                fake_norm = log_norm(mel_fake.unsqueeze(1)).squeeze(1)
                loss_norm = ((torch.nn.ReLU()(torch.abs(fake_norm - real_norm) - 0.5))**2).mean()

                # F0 loss
                F0_fake, _, _ = model.pitch_extractor(mel_fake.unsqueeze(1))
                loss_f0 = f0_loss(F0_fake, F0_real)
                
                # ASR loss
                asr_real = model.text_aligner.get_feature(gt)
                asr_fake = model.text_aligner.get_feature(mel_fake)
                loss_asr = torch.mean(torch.abs(asr_real - asr_fake))
                
                loss_feat = loss_norm + loss_f0 + loss_asr
                
                mel_cyc = model.decoder(en, F0_fake, fake_norm, s)
                
                loss_cyc = criterion(mel_cyc, gt)
                
            else:
                loss_feat = 0
                loss_cyc = 0
                        
            g_loss = loss_params.lambda_mel * loss_mel + \
                loss_params.lambda_adv * loss_adv + \
                loss_params.lambda_fm * loss_fm + \
                loss_params.lambda_mono * loss_mono + \
                loss_params.lambda_s2s * loss_s2s + \
                loss_params.lambda_sty * loss_sty + \
                loss_params.lambda_feat * loss_feat + \
                loss_params.lambda_cyc * loss_cyc

            running_loss += loss_mel.item()
            g_loss.backward()

            optimizer.step('text_encoder')
            optimizer.step('style_encoder')
            optimizer.step('decoder')

            # freeze text aligner and pitch extractor for VC training for simplicity
            if epoch > TMA_epoch and epoch < VC_epoch: 
                optimizer.step('text_aligner')
                optimizer.step('pitch_extractor')
                        
            iters = iters + 1
            if (i+1)%log_interval == 0:
                logger.info ('Epoch [%d/%d], Step [%d/%d], Mel Loss: %.5f, Adv Loss: %.5f, Disc Loss: %.5f, Mono Loss: %.5f, S2S Loss: %.5f, Style Loss: %.5f, Feat Loss: %.5f, Cycle Loss: %.5f'
                        %(epoch+1, epochs, i+1, len(train_list)//batch_size, running_loss / log_interval, loss_adv.item(), d_loss.item(), loss_mono, loss_s2s, loss_sty, loss_feat, loss_cyc))
                
                writer.add_scalar('train/mel_loss', running_loss / log_interval, iters)
                writer.add_scalar('train/adv_loss', loss_adv.item(), iters)
                writer.add_scalar('train/d_loss', d_loss.item(), iters)
                writer.add_scalar('train/mono_loss', loss_mono, iters)
                writer.add_scalar('train/s2s_loss', loss_s2s, iters)
                writer.add_scalar('train/style_loss', loss_sty, iters)
                writer.add_scalar('train/cycle_loss', loss_cyc, iters)
                writer.add_scalar('train/feat_loss', loss_feat, iters)

                running_loss = 0
                print('Time elasped:', time.time()-start_time)

        loss_test = 0

        _ = [model[key].eval() for key in model]

        with torch.no_grad():
            iters_test = 0
            for batch_idx, batch in enumerate(val_dataloader):
                optimizer.zero_grad()

                batch = [b.to(device) for b in batch]
                texts, input_lengths, mels, mel_input_length, labels, ref_mels, ref_labels = batch
                
                with torch.no_grad():
                    mask = length_to_mask(mel_input_length // (2 ** model.text_aligner.n_down)).to('cuda')
                    m = length_to_mask(input_lengths)
                    ppgs, s2s_pred, s2s_attn_feat = model.text_aligner(mels, mask, texts)

                    s2s_attn_feat = s2s_attn_feat.transpose(-1, -2)
                    s2s_attn_feat = s2s_attn_feat[..., 1:]
                    s2s_attn_feat = s2s_attn_feat.transpose(-1, -2)

                    with torch.no_grad():
                        text_mask = length_to_mask(input_lengths).to(texts.device)
                        attn_mask = (~mask).unsqueeze(-1).expand(mask.shape[0], mask.shape[1], text_mask.shape[-1]).float().transpose(-1, -2)
                        attn_mask = attn_mask.float() * (~text_mask).unsqueeze(-1).expand(text_mask.shape[0], text_mask.shape[1], mask.shape[-1]).float()
                        attn_mask = (attn_mask < 1)

                    s2s_attn_feat.masked_fill_(attn_mask, -float("inf"))
                    
                    if TMA_CEloss:
                        s2s_attn = F.softmax(s2s_attn_feat, dim=1) # along the mel dimension
                    else:
                        s2s_attn = F.softmax(s2s_attn_feat, dim=-1) # along the text dimension

                    # get monotonic version 
                    with torch.no_grad():
                        mask_ST = mask_from_lens(s2s_attn, input_lengths, mel_input_length // (2 ** model.text_aligner.n_down))
                        s2s_attn_mono = maximum_path(s2s_attn, mask_ST)

                    s2s_attn = torch.nan_to_num(s2s_attn)

                # encode
                t_en = model.text_encoder(texts, input_lengths, m)
                asr = (t_en @ s2s_attn_mono)

                # get clips
                mel_len = int(mel_input_length.min().item() / 2 - 1)
                en = []
                gt = []
                for bib in range(len(mel_input_length)):
                    mel_length = int(mel_input_length[bib].item() / 2)

                    random_start = np.random.randint(0, mel_length - mel_len)
                    en.append(asr[bib, :, random_start:random_start+mel_len])
                    gt.append(mels[bib, :, (random_start * 2):((random_start+mel_len) * 2)])
                en = torch.stack(en)
                gt = torch.stack(gt).detach()

                with torch.no_grad():
                    F0_real, _, F0 = model.pitch_extractor(gt.unsqueeze(1))
                    F0 = F0.reshape(F0.shape[0], F0.shape[1] * 2, F0.shape[2], 1).squeeze()

                # conversion
                s = model.style_encoder(gt.unsqueeze(1))
                s_trg = model.style_encoder(ref_mels.unsqueeze(1))
                
                real_norm = log_norm(gt.unsqueeze(1)).squeeze(1)
                mel_fake = model.decoder(en, F0_real, real_norm, s_trg)
                
                fake_norm = log_norm(mel_fake.unsqueeze(1)).squeeze(1).detach()
                F0_fake, _, _ = model.pitch_extractor(mel_fake.unsqueeze(1))
                mel_cyc = model.decoder(en, F0_fake, fake_norm, s)
                loss_cyc = criterion(mel_cyc, gt)                

                loss_test += loss_cyc
                iters_test += 1

        print('Epochs:', epoch + 1)
        logger.info('Validation cycle loss: %.3f' % (loss_test / iters_test))
        print('\n\n\n')

        writer.add_scalar('eval/cycle_loss', loss_test / iters_test, epoch + 1)
        attn_image = get_image(s2s_attn[0].cpu().numpy().squeeze())
        writer.add_figure('eval/attn', attn_image, epoch)
        mel_image = get_image(mel_cyc[0].cpu().numpy().squeeze())
        writer.add_figure('eval/mel_rec', mel_image, epoch)

        if epoch % saving_epoch == 0:
            if (loss_test / iters_test) < best_loss:
                best_loss = loss_test / iters_test
            print('Saving..')
            state = {
                'net':  {key: model[key].state_dict() for key in model}, 
                'optimizer': optimizer.state_dict(),
                'iters': iters,
                'val_loss': loss_test / iters_test,
                'epoch': epoch,
            }
            save_path = osp.join(log_dir, 'epoch_1st_%05d.pth' % epoch)
            torch.save(state, save_path)            
            
    print('Saving..')
    state = {
        'net':  {key: model[key].state_dict() for key in model}, 
        'optimizer': optimizer.state_dict(),
        'iters': iters,
        'val_loss': loss_test / iters_test,
        'epoch': epoch,
    }
    save_path = osp.join(log_dir, config.get('first_stage_path', 'first_stage.pth'))
    torch.save(state, save_path)
    
if __name__=="__main__":
    main()