import torch import sys import os import datetime import glob import json import re from distutils.util import strtobool from utils import ( HParams, plot_spectrogram_to_numpy, summarize, load_checkpoint, save_checkpoint, latest_checkpoint_path, ) from random import randint, shuffle from time import sleep from time import time as ttime from tqdm import tqdm from torch.cuda.amp import GradScaler, autocast from torch.nn import functional as F from torch.nn.parallel import DistributedDataParallel as DDP from torch.utils.data import DataLoader from torch.utils.tensorboard import SummaryWriter import torch.distributed as dist import torch.multiprocessing as mp now_dir = os.getcwd() sys.path.append(os.path.join(now_dir)) from data_utils import ( DistributedBucketSampler, TextAudioCollate, TextAudioCollateMultiNSFsid, TextAudioLoader, TextAudioLoaderMultiNSFsid, ) from losses import ( discriminator_loss, feature_loss, generator_loss, kl_loss, ) from mel_processing import mel_spectrogram_torch, spec_to_mel_torch from rvc.train.process.extract_model import extract_model from rvc.lib.algorithm import commons from rvc.lib.algorithm.discriminators import MultiPeriodDiscriminator from rvc.lib.algorithm.discriminators import MultiPeriodDiscriminatorV2 from rvc.lib.algorithm.synthesizers import Synthesizer # Parse command line arguments model_name = sys.argv[1] save_every_epoch = int(sys.argv[2]) total_epoch = int(sys.argv[3]) pretrainG = sys.argv[4] pretrainD = sys.argv[5] version = sys.argv[6] gpus = sys.argv[7] batch_size = int(sys.argv[8]) sample_rate = int(sys.argv[9]) pitch_guidance = strtobool(sys.argv[10]) save_only_latest = strtobool(sys.argv[11]) save_every_weights = strtobool(sys.argv[12]) cache_data_in_gpu = strtobool(sys.argv[13]) overtraining_detector = strtobool(sys.argv[14]) overtraining_threshold = int(sys.argv[15]) sync_graph = strtobool(sys.argv[16]) current_dir = os.getcwd() experiment_dir = os.path.join(current_dir, "logs", model_name) config_save_path = os.path.join(experiment_dir, "config.json") with open(config_save_path, "r") as f: config = json.load(f) config = HParams(**config) config.data.training_files = os.path.join(experiment_dir, "filelist.txt") os.environ["CUDA_VISIBLE_DEVICES"] = gpus.replace("-", ",") n_gpus = len(gpus.split("-")) torch.backends.cudnn.deterministic = False torch.backends.cudnn.benchmark = False global_step = 0 lowest_value = {"step": 0, "value": float("inf"), "epoch": 0} last_loss_gen_all = 0 # Disable logging import logging logging.getLogger("torch").setLevel(logging.ERROR) class EpochRecorder: """ Records the time elapsed per epoch. """ def __init__(self): self.last_time = ttime() def record(self): """ Records the elapsed time and returns a formatted string. """ now_time = ttime() elapsed_time = now_time - self.last_time self.last_time = now_time elapsed_time = round(elapsed_time, 1) elapsed_time_str = str(datetime.timedelta(seconds=int(elapsed_time))) current_time = datetime.datetime.now().strftime("%H:%M:%S") return f"time={current_time} | training_speed={elapsed_time_str}" def main(): """ Main function to start the training process. """ os.environ["MASTER_ADDR"] = "localhost" os.environ["MASTER_PORT"] = str(randint(20000, 55555)) def start(): """ Starts the training process with multi-GPU support. """ children = [] pid_file_path = os.path.join(experiment_dir, "train_pid.txt") with open(pid_file_path, "w") as pid_file: for i in range(n_gpus): subproc = mp.Process( target=run, args=( i, n_gpus, experiment_dir, pretrainG, pretrainD, pitch_guidance, custom_total_epoch, custom_save_every_weights, config, ), ) children.append(subproc) subproc.start() pid_file.write(str(subproc.pid) + "\n") for i in range(n_gpus): children[i].join() n_gpus = torch.cuda.device_count() if torch.cuda.is_available() == False and torch.backends.mps.is_available() == True: n_gpus = 1 if n_gpus < 1: print("GPU not detected, reverting to CPU (not recommended)") n_gpus = 1 if sync_graph == True: print( "Sync graph is now activated! With sync graph enabled, the model undergoes a single epoch of training. Once the graphs are synchronized, training proceeds for the previously specified number of epochs." ) custom_total_epoch = 1 custom_save_every_weights = True start() # Synchronize graphs by modifying config files model_config_file = os.path.join(experiment_dir, "config.json") rvc_config_file = os.path.join( now_dir, "rvc", "configs", version, str(sample_rate) + ".json" ) if not os.path.exists(rvc_config_file): rvc_config_file = os.path.join( now_dir, "rvc", "configs", "v1", str(sample_rate) + ".json" ) pattern = rf"{os.path.basename(model_name)}_1e_(\d+)s\.pth" for filename in os.listdir(experiment_dir): match = re.match(pattern, filename) if match: steps = int(match.group(1)) def edit_config(config_file): """ Edits the config file to synchronize graphs. Args: config_file (str): Path to the config file. """ with open(config_file, "r", encoding="utf8") as json_file: config_data = json.load(json_file) config_data["train"]["log_interval"] = steps with open(config_file, "w", encoding="utf8") as json_file: json.dump( config_data, json_file, indent=2, separators=(",", ": "), ensure_ascii=False, ) edit_config(model_config_file) edit_config(rvc_config_file) # Clean up unnecessary files for root, dirs, files in os.walk( os.path.join(now_dir, "logs", model_name), topdown=False ): for name in files: file_path = os.path.join(root, name) file_name, file_extension = os.path.splitext(name) if file_extension == ".0": os.remove(file_path) elif ("D" in name or "G" in name) and file_extension == ".pth": os.remove(file_path) elif ( "added" in name or "trained" in name ) and file_extension == ".index": os.remove(file_path) for name in dirs: if name == "eval": folder_path = os.path.join(root, name) for item in os.listdir(folder_path): item_path = os.path.join(folder_path, item) if os.path.isfile(item_path): os.remove(item_path) os.rmdir(folder_path) print("Successfully synchronized graphs!") custom_total_epoch = total_epoch custom_save_every_weights = save_every_weights start() else: custom_total_epoch = total_epoch custom_save_every_weights = save_every_weights start() def run( rank, n_gpus, experiment_dir, pretrainG, pretrainD, pitch_guidance, custom_total_epoch, custom_save_every_weights, config, ): """ Runs the training loop on a specific GPU. Args: rank (int): Rank of the current GPU. n_gpus (int): Total number of GPUs. """ global global_step if rank == 0: writer = SummaryWriter(log_dir=experiment_dir) writer_eval = SummaryWriter(log_dir=os.path.join(experiment_dir, "eval")) dist.init_process_group( backend="gloo", init_method="env://", world_size=n_gpus, rank=rank ) torch.manual_seed(config.train.seed) if torch.cuda.is_available(): torch.cuda.set_device(rank) # Create datasets and dataloaders if pitch_guidance == True: train_dataset = TextAudioLoaderMultiNSFsid(config.data) elif pitch_guidance == False: train_dataset = TextAudioLoader(config.data) else: raise ValueError(f"Unexpected value for pitch_guidance: {pitch_guidance}") train_sampler = DistributedBucketSampler( train_dataset, batch_size * n_gpus, [100, 200, 300, 400, 500, 600, 700, 800, 900], num_replicas=n_gpus, rank=rank, shuffle=True, ) if pitch_guidance == True: collate_fn = TextAudioCollateMultiNSFsid() elif pitch_guidance == False: collate_fn = TextAudioCollate() train_loader = DataLoader( train_dataset, num_workers=4, shuffle=False, pin_memory=True, collate_fn=collate_fn, batch_sampler=train_sampler, persistent_workers=True, prefetch_factor=8, ) # Initialize models and optimizers net_g = Synthesizer( config.data.filter_length // 2 + 1, config.train.segment_size // config.data.hop_length, **config.model, use_f0=pitch_guidance == True, is_half=config.train.fp16_run, sr=sample_rate, ) if torch.cuda.is_available(): net_g = net_g.cuda(rank) if version == "v1": net_d = MultiPeriodDiscriminator(config.model.use_spectral_norm) else: net_d = MultiPeriodDiscriminatorV2(config.model.use_spectral_norm) if torch.cuda.is_available(): net_d = net_d.cuda(rank) optim_g = torch.optim.AdamW( net_g.parameters(), config.train.learning_rate, betas=config.train.betas, eps=config.train.eps, ) optim_d = torch.optim.AdamW( net_d.parameters(), config.train.learning_rate, betas=config.train.betas, eps=config.train.eps, ) # Wrap models with DDP if torch.cuda.is_available(): net_g = DDP(net_g, device_ids=[rank]) net_d = DDP(net_d, device_ids=[rank]) else: net_g = DDP(net_g) net_d = DDP(net_d) # Load checkpoint if available try: print("Starting training...") _, _, _, epoch_str = load_checkpoint( latest_checkpoint_path(experiment_dir, "D_*.pth"), net_d, optim_d ) _, _, _, epoch_str = load_checkpoint( latest_checkpoint_path(experiment_dir, "G_*.pth"), net_g, optim_g ) global_step = (epoch_str - 1) * len(train_loader) except: epoch_str = 1 global_step = 0 if pretrainG != "": if rank == 0: print(f"Loaded pretrained (G) '{pretrainG}'") if hasattr(net_g, "module"): net_g.module.load_state_dict( torch.load(pretrainG, map_location="cpu")["model"] ) else: net_g.load_state_dict( torch.load(pretrainG, map_location="cpu")["model"] ) if pretrainD != "": if rank == 0: print(f"Loaded pretrained (D) '{pretrainD}'") if hasattr(net_d, "module"): net_d.module.load_state_dict( torch.load(pretrainD, map_location="cpu")["model"] ) else: net_d.load_state_dict( torch.load(pretrainD, map_location="cpu")["model"] ) # Initialize schedulers and scaler scheduler_g = torch.optim.lr_scheduler.ExponentialLR( optim_g, gamma=config.train.lr_decay, last_epoch=epoch_str - 2 ) scheduler_d = torch.optim.lr_scheduler.ExponentialLR( optim_d, gamma=config.train.lr_decay, last_epoch=epoch_str - 2 ) scaler = GradScaler(enabled=config.train.fp16_run) cache = [] for epoch in range(epoch_str, total_epoch + 1): if rank == 0: train_and_evaluate( rank, epoch, config, [net_g, net_d], [optim_g, optim_d], scaler, [train_loader, None], [writer, writer_eval], cache, custom_save_every_weights, custom_total_epoch, ) else: train_and_evaluate( rank, epoch, config, [net_g, net_d], [optim_g, optim_d], scaler, [train_loader, None], None, cache, custom_save_every_weights, custom_total_epoch, ) scheduler_g.step() scheduler_d.step() def train_and_evaluate( rank, epoch, hps, nets, optims, scaler, loaders, writers, cache, custom_save_every_weights, custom_total_epoch, ): """ Trains and evaluates the model for one epoch. Args: rank (int): Rank of the current GPU. epoch (int): Current epoch number. hps (Namespace): Hyperparameters. nets (list): List of models [net_g, net_d]. optims (list): List of optimizers [optim_g, optim_d]. scaler (GradScaler): Gradient scaler for mixed precision training. loaders (list): List of dataloaders [train_loader, eval_loader]. writers (list): List of TensorBoard writers [writer, writer_eval]. cache (list): List to cache data in GPU memory. """ global global_step, last_loss_gen_all, lowest_value if epoch == 1: lowest_value = {"step": 0, "value": float("inf"), "epoch": 0} last_loss_gen_all = 0.0 net_g, net_d = nets optim_g, optim_d = optims train_loader = loaders[0] if loaders is not None else None if writers is not None: writer = writers[0] train_loader.batch_sampler.set_epoch(epoch) net_g.train() net_d.train() # Data caching if cache_data_in_gpu == True: data_iterator = cache if cache == []: for batch_idx, info in enumerate(train_loader): if pitch_guidance == True: ( phone, phone_lengths, pitch, pitchf, spec, spec_lengths, wave, wave_lengths, sid, ) = info elif pitch_guidance == False: ( phone, phone_lengths, spec, spec_lengths, wave, wave_lengths, sid, ) = info if torch.cuda.is_available(): phone = phone.cuda(rank, non_blocking=True) phone_lengths = phone_lengths.cuda(rank, non_blocking=True) if pitch_guidance == True: pitch = pitch.cuda(rank, non_blocking=True) pitchf = pitchf.cuda(rank, non_blocking=True) sid = sid.cuda(rank, non_blocking=True) spec = spec.cuda(rank, non_blocking=True) spec_lengths = spec_lengths.cuda(rank, non_blocking=True) wave = wave.cuda(rank, non_blocking=True) wave_lengths = wave_lengths.cuda(rank, non_blocking=True) if pitch_guidance == True: cache.append( ( batch_idx, ( phone, phone_lengths, pitch, pitchf, spec, spec_lengths, wave, wave_lengths, sid, ), ) ) elif pitch_guidance == False: cache.append( ( batch_idx, ( phone, phone_lengths, spec, spec_lengths, wave, wave_lengths, sid, ), ) ) else: shuffle(cache) else: data_iterator = enumerate(train_loader) epoch_recorder = EpochRecorder() with tqdm(total=len(train_loader), leave=False) as pbar: for batch_idx, info in data_iterator: if pitch_guidance == True: ( phone, phone_lengths, pitch, pitchf, spec, spec_lengths, wave, wave_lengths, sid, ) = info elif pitch_guidance == False: phone, phone_lengths, spec, spec_lengths, wave, wave_lengths, sid = info if (cache_data_in_gpu == False) and torch.cuda.is_available(): phone = phone.cuda(rank, non_blocking=True) phone_lengths = phone_lengths.cuda(rank, non_blocking=True) if pitch_guidance == True: pitch = pitch.cuda(rank, non_blocking=True) pitchf = pitchf.cuda(rank, non_blocking=True) sid = sid.cuda(rank, non_blocking=True) spec = spec.cuda(rank, non_blocking=True) spec_lengths = spec_lengths.cuda(rank, non_blocking=True) wave = wave.cuda(rank, non_blocking=True) # Forward pass with autocast(enabled=config.train.fp16_run): if pitch_guidance == True: ( y_hat, ids_slice, x_mask, z_mask, (z, z_p, m_p, logs_p, m_q, logs_q), ) = net_g( phone, phone_lengths, pitch, pitchf, spec, spec_lengths, sid ) elif pitch_guidance == False: ( y_hat, ids_slice, x_mask, z_mask, (z, z_p, m_p, logs_p, m_q, logs_q), ) = net_g(phone, phone_lengths, spec, spec_lengths, sid) mel = spec_to_mel_torch( spec, config.data.filter_length, config.data.n_mel_channels, config.data.sample_rate, config.data.mel_fmin, config.data.mel_fmax, ) y_mel = commons.slice_segments( mel, ids_slice, config.train.segment_size // config.data.hop_length ) with autocast(enabled=False): y_hat_mel = mel_spectrogram_torch( y_hat.float().squeeze(1), config.data.filter_length, config.data.n_mel_channels, config.data.sample_rate, config.data.hop_length, config.data.win_length, config.data.mel_fmin, config.data.mel_fmax, ) if config.train.fp16_run == True: y_hat_mel = y_hat_mel.half() wave = commons.slice_segments( wave, ids_slice * config.data.hop_length, config.train.segment_size ) y_d_hat_r, y_d_hat_g, _, _ = net_d(wave, y_hat.detach()) with autocast(enabled=False): loss_disc, losses_disc_r, losses_disc_g = discriminator_loss( y_d_hat_r, y_d_hat_g ) # Discriminator backward and update optim_d.zero_grad() scaler.scale(loss_disc).backward() scaler.unscale_(optim_d) grad_norm_d = commons.clip_grad_value(net_d.parameters(), None) scaler.step(optim_d) # Generator backward and update with autocast(enabled=config.train.fp16_run): y_d_hat_r, y_d_hat_g, fmap_r, fmap_g = net_d(wave, y_hat) with autocast(enabled=False): loss_mel = F.l1_loss(y_mel, y_hat_mel) * config.train.c_mel loss_kl = ( kl_loss(z_p, logs_q, m_p, logs_p, z_mask) * config.train.c_kl ) loss_fm = feature_loss(fmap_r, fmap_g) loss_gen, losses_gen = generator_loss(y_d_hat_g) loss_gen_all = loss_gen + loss_fm + loss_mel + loss_kl if loss_gen_all < lowest_value["value"]: lowest_value["value"] = loss_gen_all lowest_value["step"] = global_step lowest_value["epoch"] = epoch # print(f'Lowest generator loss updated: {lowest_value["value"]} at epoch {epoch}, step {global_step}') if epoch > lowest_value["epoch"]: print( "Alert: The lower generating loss has been exceeded by a lower loss in a subsequent epoch." ) optim_g.zero_grad() scaler.scale(loss_gen_all).backward() scaler.unscale_(optim_g) grad_norm_g = commons.clip_grad_value(net_g.parameters(), None) scaler.step(optim_g) scaler.update() # Logging and checkpointing if rank == 0: if global_step % config.train.log_interval == 0: lr = optim_g.param_groups[0]["lr"] # print("Epoch: {} [{:.0f}%]".format(epoch, 100.0 * batch_idx / len(train_loader))) if loss_mel > 75: loss_mel = 75 if loss_kl > 9: loss_kl = 9 scalar_dict = { "loss/g/total": loss_gen_all, "loss/d/total": loss_disc, "learning_rate": lr, "grad_norm_d": grad_norm_d, "grad_norm_g": grad_norm_g, } scalar_dict.update( { "loss/g/fm": loss_fm, "loss/g/mel": loss_mel, "loss/g/kl": loss_kl, } ) scalar_dict.update( {"loss/g/{}".format(i): v for i, v in enumerate(losses_gen)} ) scalar_dict.update( { "loss/d_r/{}".format(i): v for i, v in enumerate(losses_disc_r) } ) scalar_dict.update( { "loss/d_g/{}".format(i): v for i, v in enumerate(losses_disc_g) } ) image_dict = { "slice/mel_org": plot_spectrogram_to_numpy( y_mel[0].data.cpu().numpy() ), "slice/mel_gen": plot_spectrogram_to_numpy( y_hat_mel[0].data.cpu().numpy() ), "all/mel": plot_spectrogram_to_numpy(mel[0].data.cpu().numpy()), } summarize( writer=writer, global_step=global_step, images=image_dict, scalars=scalar_dict, ) global_step += 1 pbar.update(1) # Save checkpoint if epoch % save_every_epoch == False and rank == 0: checkpoint_suffix = ( f"{global_step if save_only_latest == False else 2333333}.pth" ) save_checkpoint( net_g, optim_g, config.train.learning_rate, epoch, os.path.join(experiment_dir, "G_" + checkpoint_suffix), ) save_checkpoint( net_d, optim_d, config.train.learning_rate, epoch, os.path.join(experiment_dir, "D_" + checkpoint_suffix), ) if rank == 0 and custom_save_every_weights == True: if hasattr(net_g, "module"): ckpt = net_g.module.state_dict() else: ckpt = net_g.state_dict() extract_model( ckpt=ckpt, sr=sample_rate, pitch_guidance=pitch_guidance == True, name=model_name, model_dir=os.path.join( experiment_dir, f"{model_name}_{epoch}e_{global_step}s.pth", ), epoch=epoch, step=global_step, version=version, hps=hps, ) # Overtraining detection and best model saving if overtraining_detector == True: if epoch >= (lowest_value["epoch"] + overtraining_threshold): print( "Stopping training due to possible overtraining. Lowest generator loss: {} at epoch {}, step {}".format( lowest_value["value"], lowest_value["epoch"], lowest_value["step"] ) ) os._exit(2333333) best_epoch = lowest_value["epoch"] + overtraining_threshold - epoch if best_epoch == overtraining_threshold: old_model_files = glob.glob( os.path.join( experiment_dir, "{}_{}e_{}s_best_epoch.pth".format(model_name, "*", "*"), ) ) for file in old_model_files: os.remove(file) if hasattr(net_g, "module"): ckpt = net_g.module.state_dict() else: ckpt = net_g.state_dict() extract_model( ckpt=ckpt, sr=sample_rate, pitch_guidance=pitch_guidance == True, name=model_name, model_dir=os.path.join( experiment_dir, f"{model_name}_{epoch}e_{global_step}s_best_epoch.pth", ), epoch=epoch, step=global_step, version=version, hps=hps, ) # Print training progress if rank == 0: lowest_value_rounded = float(lowest_value["value"]) # Convert to float lowest_value_rounded = round( lowest_value_rounded, 3 ) # Round to 3 decimal place if epoch > 1 and overtraining_detector == True: print( f"{model_name} | epoch={epoch} | step={global_step} | {epoch_recorder.record()} | lowest_value={lowest_value_rounded} (epoch {lowest_value['epoch']} and step {lowest_value['step']}) | Number of epochs remaining for overtraining: {lowest_value['epoch'] + overtraining_threshold - epoch}" ) elif epoch > 1 and overtraining_detector == False: print( f"{model_name} | epoch={epoch} | step={global_step} | {epoch_recorder.record()} | lowest_value={lowest_value_rounded} (epoch {lowest_value['epoch']} and step {lowest_value['step']})" ) else: print( f"{model_name} | epoch={epoch} | step={global_step} | {epoch_recorder.record()}" ) last_loss_gen_all = loss_gen_all # Save the final model if epoch >= custom_total_epoch and rank == 0: lowest_value_rounded = float(lowest_value["value"]) # Convert to float lowest_value_rounded = round( lowest_value_rounded, 3 ) # Round to 3 decimal place print( f"Training has been successfully completed with {epoch} epoch, {global_step} steps and {round(loss_gen_all.item(), 3)} loss gen." ) print( f"Lowest generator loss: {lowest_value_rounded} at epoch {lowest_value['epoch']}, step {lowest_value['step']}" ) pid_file_path = os.path.join(experiment_dir, "train_pid.txt") os.remove(pid_file_path) if hasattr(net_g, "module"): ckpt = net_g.module.state_dict() else: ckpt = net_g.state_dict() extract_model( ckpt=ckpt, sr=sample_rate, pitch_guidance=pitch_guidance == True, name=model_name, model_dir=os.path.join( experiment_dir, f"{model_name}_{epoch}e_{global_step}s.pth", ), epoch=epoch, step=global_step, version=version, hps=hps, ) sleep(1) os._exit(2333333) if __name__ == "__main__": torch.multiprocessing.set_start_method("spawn") main()