# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import argparse import os import paddle from paddleaudio.datasets import ESC50 from paddleaudio.features import LogMelSpectrogram from paddleaudio.utils import logger from paddleaudio.utils import Timer from paddlespeech.cls.models import cnn14 from paddlespeech.cls.models import SoundClassifier # yapf: disable parser = argparse.ArgumentParser(__doc__) parser.add_argument("--epochs", type=int, default=50, help="Number of epoches for fine-tuning.") parser.add_argument("--feat_backend", type=str, choices=['numpy', 'paddle'], default='numpy', help="Choose backend to extract features from audio files.") parser.add_argument("--learning_rate", type=float, default=5e-5, help="Learning rate used to train with warmup.") parser.add_argument("--batch_size", type=int, default=16, help="Total examples' number in batch for training.") parser.add_argument("--num_workers", type=int, default=0, help="Number of workers in dataloader.") parser.add_argument("--checkpoint_dir", type=str, default='./checkpoint', help="Directory to save model checkpoints.") parser.add_argument("--save_freq", type=int, default=10, help="Save checkpoint every n epoch.") parser.add_argument("--log_freq", type=int, default=10, help="Log the training infomation every n steps.") args = parser.parse_args() # yapf: enable if __name__ == "__main__": nranks = paddle.distributed.get_world_size() if paddle.distributed.get_world_size() > 1: paddle.distributed.init_parallel_env() local_rank = paddle.distributed.get_rank() backbone = cnn14(pretrained=True, extract_embedding=True) model = SoundClassifier(backbone, num_class=len(ESC50.label_list)) model = paddle.DataParallel(model) optimizer = paddle.optimizer.Adam( learning_rate=args.learning_rate, parameters=model.parameters()) criterion = paddle.nn.loss.CrossEntropyLoss() if args.feat_backend == 'numpy': train_ds = ESC50(mode='train', feat_type='melspectrogram') dev_ds = ESC50(mode='dev', feat_type='melspectrogram') else: train_ds = ESC50(mode='train') dev_ds = ESC50(mode='dev') feature_extractor = LogMelSpectrogram(sr=16000) train_sampler = paddle.io.DistributedBatchSampler( train_ds, batch_size=args.batch_size, shuffle=True, drop_last=False) train_loader = paddle.io.DataLoader( train_ds, batch_sampler=train_sampler, num_workers=args.num_workers, return_list=True, use_buffer_reader=True, ) steps_per_epoch = len(train_sampler) timer = Timer(steps_per_epoch * args.epochs) timer.start() for epoch in range(1, args.epochs + 1): model.train() avg_loss = 0 num_corrects = 0 num_samples = 0 for batch_idx, batch in enumerate(train_loader): if args.feat_backend == 'numpy': feats, labels = batch else: waveforms, labels = batch feats = feature_extractor( waveforms ) # Need a padding when lengths of waveforms differ in a batch. feats = paddle.transpose(feats, [0, 2, 1]) # To [N, length, n_mels] logits = model(feats) loss = criterion(logits, labels) loss.backward() optimizer.step() if isinstance(optimizer._learning_rate, paddle.optimizer.lr.LRScheduler): optimizer._learning_rate.step() optimizer.clear_grad() # Calculate loss avg_loss += loss.numpy()[0] # Calculate metrics preds = paddle.argmax(logits, axis=1) num_corrects += (preds == labels).numpy().sum() num_samples += feats.shape[0] timer.count() if (batch_idx + 1) % args.log_freq == 0 and local_rank == 0: lr = optimizer.get_lr() avg_loss /= args.log_freq avg_acc = num_corrects / num_samples print_msg = 'Epoch={}/{}, Step={}/{}'.format( epoch, args.epochs, batch_idx + 1, steps_per_epoch) print_msg += ' loss={:.4f}'.format(avg_loss) print_msg += ' acc={:.4f}'.format(avg_acc) print_msg += ' lr={:.6f} step/sec={:.2f} | ETA {}'.format( lr, timer.timing, timer.eta) logger.train(print_msg) avg_loss = 0 num_corrects = 0 num_samples = 0 if epoch % args.save_freq == 0 and batch_idx + 1 == steps_per_epoch and local_rank == 0: dev_sampler = paddle.io.BatchSampler( dev_ds, batch_size=args.batch_size, shuffle=False, drop_last=False) dev_loader = paddle.io.DataLoader( dev_ds, batch_sampler=dev_sampler, num_workers=args.num_workers, return_list=True, ) model.eval() num_corrects = 0 num_samples = 0 with logger.processing('Evaluation on validation dataset'): for batch_idx, batch in enumerate(dev_loader): if args.feat_backend == 'numpy': feats, labels = batch else: waveforms, labels = batch feats = feature_extractor(waveforms) feats = paddle.transpose(feats, [0, 2, 1]) logits = model(feats) preds = paddle.argmax(logits, axis=1) num_corrects += (preds == labels).numpy().sum() num_samples += feats.shape[0] print_msg = '[Evaluation result]' print_msg += ' dev_acc={:.4f}'.format(num_corrects / num_samples) logger.eval(print_msg) # Save model save_dir = os.path.join(args.checkpoint_dir, 'epoch_{}'.format(epoch)) logger.info('Saving model checkpoint to {}'.format(save_dir)) paddle.save(model.state_dict(), os.path.join(save_dir, 'model.pdparams')) paddle.save(optimizer.state_dict(), os.path.join(save_dir, 'model.pdopt'))