Electrical Engineering And Systems Science

One of the most important parts of an end-to-end speaker verification system is the speaker embedding generation. In our previous paper, we reported that shortcut connections-based multi-layer aggregation improves the representational power of the speaker embedding. However, the number of model parameters is relatively large and the unspecified variations increase in the multi-layer aggregation. Therefore, we propose a self-attentive multi-layer aggregation with feature recalibration and normalization for end-to-end speaker verification system. To reduce the number of model parameters, the ResNet, which scaled channel width and layer depth, is used as a baseline. To control the variability in the training, a self-attention mechanism is applied to perform the multi-layer aggregation with dropout regularizations and batch normalizations. Then, a feature recalibration layer is applied to the aggregated feature using fully-connected layers and nonlinear activation functions. Deep length normalization is also used on a recalibrated feature in the end-to-end training process. Experimental results using the VoxCeleb1 evaluation dataset showed that the performance of the proposed methods was comparable to that of state-of-the-art models (equal error rate of 4.95% and 2.86%, using the VoxCeleb1 and VoxCeleb2 training datasets, respectively).

Unsupervised spoken term discovery consists of two tasks: finding the acoustic segment boundaries and labeling acoustically similar segments with the same labels. We perform segmentation based on the assumption that the frame feature vectors are more similar within a segment than across the segments. Therefore, for strong segmentation performance, it is crucial that the features represent the phonetic properties of a frame more than other factors of variability. We achieve this via a self-expressing autoencoder framework. It consists of a single encoder and two decoders with shared weights. The encoder projects the input features into a latent representation. One of the decoders tries to reconstruct the input from these latent representations and the other from the self-expressed version of them. We use the obtained features to segment and cluster the speech data. We evaluate the performance of the proposed method in the Zero Resource 2020 challenge unit discovery task. The proposed system consistently outperforms the baseline, demonstrating the usefulness of the method in learning representations.

We propose a self-supervised representation learning model for the task of unsupervised phoneme boundary detection. The model is a convolutional neural network that operates directly on the raw waveform. It is optimized to identify spectral changes in the signal using the Noise-Contrastive Estimation principle. At test time, a peak detection algorithm is applied over the model outputs to produce the final boundaries. As such, the proposed model is trained in a fully unsupervised manner with no manual annotations in the form of target boundaries nor phonetic transcriptions. We compare the proposed approach to several unsupervised baselines using both TIMIT and Buckeye corpora. Results suggest that our approach surpasses the baseline models and reaches state-of-the-art performance on both data sets. Furthermore, we experimented with expanding the training set with additional examples from the Librispeech corpus. We evaluated the resulting model on distributions and languages that were not seen during the training phase (English, Hebrew and German) and showed that utilizing additional untranscribed data is beneficial for model performance.

The computing power of mobile devices limits the end-user applications in terms of storage size, processing, memory and energy consumption. These limitations motivate researchers for the design of more efficient deep models. On the other hand, self-attention networks based on Transformer architecture have attracted remarkable interests due to their high parallelization capabilities and strong performance on a variety of Natural Language Processing (NLP) applications. Inspired by the Transformer, we propose a tandem Self-Attention Encoding and Pooling (SAEP) mechanism to obtain a discriminative speaker embedding given non-fixed length speech utterances. SAEP is a stack of identical blocks solely relied on self-attention and position-wise feed-forward networks to create vector representation of speakers. This approach encodes short-term speaker spectral features into speaker embeddings to be used in text-independent speaker verification. We have evaluated this approach on both VoxCeleb1 & 2 datasets. The proposed architecture is able to outperform the baseline x-vector, and shows competitive performance to some other benchmarks based on convolutions, with a significant reduction in model size. It employs 94%, 95%, and 73% less parameters compared to ResNet-34, ResNet-50, and x-vector, respectively. This indicates that the proposed fully attention based architecture is more efficient in extracting time-invariant features from speaker utterances.

In this paper, we apply Semi-Supervised Learning (SSL) along with Data Augmentation (DA) for improving the accuracy of End-to-End ASR. We focus on the consistency regularization principle, which has been successfully applied to image classification tasks, and present sequence-to-sequence (seq2seq) versions of the FixMatch and Noisy Student algorithms. Specifically, we generate the pseudo labels for the unlabeled data on-the-fly with a seq2seq model after perturbing the input features with DA. We also propose soft label variants of both algorithms to cope with pseudo label errors, showing further performance improvements. We conduct SSL experiments on a conversational speech data set with 1.9kh manually transcribed training data, using only 25% of the original labels (475h labeled data). In the result, the Noisy Student algorithm with soft labels and consistency regularization achieves 10.4% word error rate (WER) reduction when adding 475h of unlabeled data, corresponding to a recovery rate of 92%. Furthermore, when iteratively adding 950h more unlabeled data, our best SSL performance is within 5% WER increase compared to using the full labeled training set (recovery rate: 78%).

In this paper, a combinative approach using Nonnegative Matrix Factorization (NMF) and Convolutional Neural Network (CNN) is proposed for audio clip Sound Event Detection (SED). The main idea begins with the use of NMF to approximate strong labels for the weakly labeled data. Subsequently, using the approximated strongly labeled data, two different CNNs are trained in a semi-supervised framework where one CNN is used for clip-level prediction and the other for frame-level prediction. Based on this idea, our model can achieve an event-based F1-score of 45.7% on the Detection and Classification of Acoustic Scenes and Events (DCASE) 2020 Challenge Task 4 validation dataset. By ensembling models through averaging the posterior outputs, event-based F1-score can be increased to 48.6%. By comparing with the baseline model, our proposed models outperform the baseline model by over 8%. By testing our models on the DCASE 2020 Challenge Task 4 test set, our models can achieve an event-based F1-score of 44.4% while our ensembled system can achieve an event-based F1-score of 46.3%. Such results have a minimum margin of 7% over the baseline system which demonstrates the robustness of our proposed method on different datasets.

Recent progress in singing voice separation has primarily focused on supervised deep learning methods. However, the scarcity of ground-truth data with clean musical sources has been a problem for long. Given a limited set of labeled data, we present a method to leverage a large volume of unlabeled data to improve the model's performance. Following the noisy self-training framework, we first train a teacher network on the small labeled dataset and infer pseudo-labels from the large corpus of unlabeled mixtures. Then, a larger student network is trained on combined ground-truth and self-labeled datasets. Empirical results show that the proposed self-training scheme, along with data augmentation methods, effectively leverage the large unlabeled corpus and obtain superior performance compared to supervised methods.

Automatic emotion recognition is an active research topic with wide range of applications. Due to the high manual annotation cost and inevitable label ambiguity, the development of emotion recognition dataset is limited in both scale and quality. Therefore, one of the key challenges is how to build effective models with limited data resource. Previous works have explored different approaches to tackle this challenge including data enhancement, transfer learning, and semi-supervised learning etc. However, the weakness of these existing approaches includes such as training instability, large performance loss during transfer, or marginal improvement. In this work, we propose a novel semi-supervised multi-modal emotion recognition model based on cross-modality distribution matching, which leverages abundant unlabeled data to enhance the model training under the assumption that the inner emotional status is consistent at the utterance level across modalities. We conduct extensive experiments to evaluate the proposed model on two benchmark datasets, IEMOCAP and MELD. The experiment results prove that the proposed semi-supervised learning model can effectively utilize unlabeled data and combine multi-modalities to boost the emotion recognition performance, which outperforms other state-of-the-art approaches under the same condition. The proposed model also achieves competitive capacity compared with existing approaches which take advantage of additional auxiliary information such as speaker and interaction context.

Sound event detection is a core module for acoustic environmental analysis. Semi-supervised learning technique allows to largely scale up the dataset without increasing the annotation budget, and recently attracts lots of research attention. In this work, we study on two advanced semi-supervised learning techniques for sound event detection. Data augmentation is important for the success of recent deep learning systems. This work studies the audio-signal random augmentation method, which provides an augmentation strategy that can handle a large number of different audio transformations. In addition, consistency regularization is widely adopted in recent state-of-the-art semi-supervised learning methods, which exploits the unlabelled data by constraining the prediction of different transformations of one sample to be identical to the prediction of this sample. This work finds that, for semi-supervised sound event detection, consistency regularization is an effective strategy, especially the best performance is achieved when it is combined with the MeanTeacher model.

The lack of labeled data is a major obstacle in many music information retrieval tasks such as melody extraction, where labeling is extremely laborious or costly. Semi-supervised learning (SSL) provides a solution to alleviate the issue by leveraging a large amount of unlabeled data. In this paper, we propose an SSL method using teacher-student models for vocal melody extraction. The teacher model is pre-trained with labeled data and guides the student model to make identical predictions given unlabeled input in a self-training setting. We examine three setups of teacher-student models with different data augmentation schemes and loss functions. Also, considering the scarcity of labeled data in the test phase, we artificially generate large-scale testing data with pitch labels from unlabeled data using an analysis-synthesis method. The results show that the SSL method significantly increases the performance against supervised learning only and the improvement depends on the teacher-student models, the size of unlabeled data, the number of self-training iterations, and other training details. We also find that it is essential to ensure that the unlabeled audio has vocal parts. Finally, we show that the proposed SSL method enables a baseline convolutional recurrent neural network model to achieve performance comparable to state-of-the-arts.