Sound event detection (SED) is essential for recognizing specific sounds and their temporal locations within acoustic signals. This becomes challenging particularly for on-device applications, where computational resources are limited. To address this issue, we introduce a novel framework referred to as dual knowledge distillation for developing efficient SED systems in this work. Our proposed dual knowledge distillation commences with temporal-averaging knowledge distillation (TAKD), utilizing a mean student model derived from the temporal averaging of the student model's parameters. This allows the student model to indirectly learn from a pre-trained teacher model, ensuring a stable knowledge distillation. Subsequently, we introduce embedding-enhanced feature distillation (EEFD), which involves incorporating an embedding distillation layer within the student model to bolster contextual learning. On DCASE 2023 Task 4A public evaluation dataset, our proposed SED system with dual knowledge distillation having merely one-third of the baseline model's parameters, demonstrates superior performance in terms of PSDS1 and PSDS2. This highlights the importance of proposed dual knowledge distillation for compact SED systems, which can be ideal for edge devices.
Traditional vision transformer consists of two parts: transformer encoder and multi-layer perception (MLP). The former plays the role of feature learning to obtain better representation, while the latter plays the role of classification. Here, the MLP is constituted of two fully connected (FC) layers, average value computing, FC layer and softmax layer. However, due to the use of average value computing module, some useful information may get lost, which we plan to preserve by the use of alternative framework. In this work, we propose a novel vision transformer referred to as adaptive-avg-pooling based attention vision transformer (AAViT) that uses modules of adaptive average pooling and attention to replace the module of average value computing. We explore the proposed AAViT for the studies on face anti-spoofing using Replay-Attack database. The experiments show that the AAViT outperforms vision transformer in face anti-spoofing by producing a reduced equal error rate. In addition, we found that the proposed AAViT can perform much better than some commonly used neural networks such as ResNet and some other known systems on the Replay-Attack corpus.
Sound event detection (SED) entails identifying the type of sound and estimating its temporal boundaries from acoustic signals. These events are uniquely characterized by their spatio-temporal features, which are determined by the way they are produced. In this study, we leverage some distinctive high-level acoustic characteristics of various sound events to assist the SED model training, without requiring additional labeled data. Specifically, we use the DCASE Task 4 2022 dataset and categorize the 10 classes into four subcategories based on their high-level acoustic characteristics. We then introduce a novel multi-task learning framework that jointly trains the SED and high-level acoustic characteristics classification tasks, using shared layers and weighted loss. Our method significantly improves the performance of the SED system, achieving a 36.3% improvement in terms of the polyphonic sound event detection score compared to the baseline on the DCASE 2022 Task 4 validation set.
Jointly learning from a small labeled set and a larger unlabeled set is an active research topic under semi-supervised learning (SSL). In this paper, we propose a novel SSL method based on a two-stage framework for leveraging a large unlabeled in-domain set. Stage-1 of our proposed framework focuses on audio-tagging (AT), which assists the sound event detection (SED) system in Stage-2. The AT system is trained utilizing a strongly labeled set converted into weak predictions referred to as weakified set, a weakly labeled set, and an unlabeled set. This AT system then infers on the unlabeled set to generate reliable pseudo-weak labels, which are used with the strongly and weakly labeled set to train a frequency dynamic convolutional recurrent neural network-based SED system at Stage-2 in a supervised manner. Our system outperforms the baseline by 45.5% in terms of polyphonic sound detection score on the DESED real validation set.
This manuscript describes the I4U submission to the 2020 NIST Speaker Recognition Evaluation (SRE'20) Conversational Telephone Speech (CTS) Challenge. The I4U's submission was resulted from active collaboration among researchers across eight research teams - I$^2$R (Singapore), UEF (Finland), VALPT (Italy, Spain), NEC (Japan), THUEE (China), LIA (France), NUS (Singapore), INRIA (France) and TJU (China). The submission was based on the fusion of top performing sub-systems and sub-fusion systems contributed by individual teams. Efforts have been spent on the use of common development and validation sets, submission schedule and milestone, minimizing inconsistency in trial list and score file format across sites.
We study a novel neural architecture and its training strategies of speaker encoder for speaker recognition without using any identity labels. The speaker encoder is trained to extract a fixed-size speaker embedding from a spoken utterance of various length. Contrastive learning is a typical self-supervised learning technique. However, the quality of the speaker encoder depends very much on the sampling strategy of positive and negative pairs. It is common that we sample a positive pair of segments from the same utterance. Unfortunately, such poor-man's positive pairs (PPP) lack necessary diversity for the training of a robust encoder. In this work, we propose a multi-modal contrastive learning technique with novel sampling strategies. By cross-referencing between speech and face data, we study a method that finds diverse positive pairs (DPP) for contrastive learning, thus improving the robustness of the speaker encoder. We train the speaker encoder on the VoxCeleb2 dataset without any speaker labels, and achieve an equal error rate (EER) of 2.89\%, 3.17\% and 6.27\% under the proposed progressive clustering strategy, and an EER of 1.44\%, 1.77\% and 3.27\% under the two-stage learning strategy with pseudo labels, on the three test sets of VoxCeleb1. This novel solution outperforms the state-of-the-art self-supervised learning methods by a large margin, at the same time, achieves comparable results with the supervised learning counterpart. We also evaluate our self-supervised learning technique on LRS2 and LRW datasets, where the speaker information is unknown. All experiments suggest that the proposed neural architecture and sampling strategies are robust across datasets.
The time delay neural network (TDNN) represents one of the state-of-the-art of neural solutions to text-independent speaker verification. However, they require a large number of filters to capture the speaker characteristics at any local frequency region. In addition, the performance of such systems may degrade under short utterance scenarios. To address these issues, we propose a multi-scale frequency-channel attention (MFA), where we characterize speakers at different scales through a novel dual-path design which consists of a convolutional neural network and TDNN. We evaluate the proposed MFA on the VoxCeleb database and observe that the proposed framework with MFA can achieve state-of-the-art performance while reducing parameters and computation complexity. Further, the MFA mechanism is found to be effective for speaker verification with short test utterances.
This work provides a brief description of Human Language Technology (HLT) Laboratory, National University of Singapore (NUS) system submission for 2020 NIST conversational telephone speech (CTS) speaker recognition evaluation (SRE). The challenge focuses on evaluation under CTS data containing multilingual speech. The systems developed at HLT-NUS consider time-delay neural network (TDNN) x-vector and ECAPA-TDNN systems. We also perform domain adaption of probabilistic linear discriminant analysis (PLDA) model and adaptive s-norm on our systems. The score level fusion of TDNN x-vector and ECAPA-TDNN systems is carried out, which improves the final system performance of our submission to 2020 NIST CTS SRE.
In self-supervised learning for speaker recognition, pseudo labels are useful as the supervision signals. It is a known fact that a speaker recognition model doesn't always benefit from pseudo labels due to their unreliability. In this work, we observe that a speaker recognition network tends to model the data with reliable labels faster than those with unreliable labels. This motivates us to study a loss-gated learning (LGL) strategy, which extracts the reliable labels through the fitting ability of the neural network during training. With the proposed LGL, our speaker recognition model obtains a 46.3% performance gain over the system without it. Further, the proposed self-supervised speaker recognition with LGL trained on the VoxCeleb2 dataset without any labels achieves an equal error rate of 1.66% on the VoxCeleb1 original test set. We plan to release the codes later for public use.