Weakly labeled sound event detection with a capsule-transformer model

| openaire: EC/H2020/637422/EU//EVERYSOUND

Weakly labeled semi-supervised learning methods have recently drawn increasing attention from the research community for sound event detection tasks. Due to the weakness of the labelling, neural networks are often designed to perform sound event detection (SED) and audio tagging (AT) at the same time. In this paper, we propose a task-aware mean teacher method using a convolutional recurrent neural network (CRNN) with multi-branch structure to solve the SED and AT tasks differently. Specifically, a branch with coarse-level temporal resolution is designed for the AT task, while a branch with fine-level temporal resolution is designed for the SED task. The mean teacher based semi-supervised learning method is first adopted to improve the performance of the coarse-level AT branch by exploiting unlabeled data. Then the coarse-level AT branch is introduced as a teacher to guide the aggregated AT output of the fine-level SED branch, yielding an improvement in the SED performance. To further improve the AT and SED performance, information from multiple layers is exploited in the form of a multi-resolution feature. Experimental results on Task4 of the DCASE2018 challenge demonstrate the superiority of the proposed method, achieving 37.7% F1-score, which outperforms the winning system’s 32.4%.

<p>Joint sound event localization and detection (SELD) is an emerging audio signal processing task adding spatial dimensions to acoustic scene analysis and sound event detection. A popular approach to modeling SELD jointly is using convolutional recur- rent neural network (CRNN) models, where CNNs learn high-level features from multi-channel audio input and the RNNs learn temporal relationships from these high-level features. However, RNNs have some drawbacks, such as a limited capability to model long temporal dependencies and slow training and inference times due to their sequential processing nature. Recently, a few SELD studies used multi-head self-attention (MHSA), among other innovations in their models. MHSA and the related transformer networks have shown state-of-the-art performance in various domains. While they can model long temporal dependencies, they can also be parallelized efficiently. In this paper, we study in detail the effect of MHSA on the SELD task. Specifically, we examined the effects of replacing the RNN blocks with self-attention layers. We studied the influence of stacking multiple self-attention blocks, using multiple attention heads in each self-attention block, and the effect of position embeddings and layer normalization. Evaluation on the DCASE 2021 SELD (task 3) development data set shows a significant improvement in all employed metrics compared to the baseline CRNN accompanying the task.</p>

Sound event detection(SED) consists of two subtasks: predicting the classes of sound events within an audio clip (audio tagging) and indicating the onset and offset times for each event (localization). One of the common approaches for SED with weak label is multiple instance learning (MIL) method. However, the general MIL method only optimizes the global loss calculated from the aggregated clip-wise predictions and weak clip labels, lacking a direct constraint on the frame-wise predictions, which leads to a large number of unreasonable prediction values. To address this issue, we explore the deterministic information that can be used to constrain the framewise predictions and based on which we design a frame loss with two terms. Experimental results on the DCASE2017 Task4 dataset demonstrate that the proposed loss can improve the performance of general MIL method. While this article focuses on SED applications, the proposed methods could be applied widely to MIL problems. Code will be available at WSSED.

In this paper, we present a gated convolutional neural network and a temporal attention-based localization method for audio classification, which won the 1st place in the large-scale weakly supervised sound event detection task of Detection and Classification of Acoustic Scenes and Events (DCASE) 2017 challenge. The audio clips in this task, which are extracted from YouTube videos, are manually labelled with one or more audio tags, but without time stamps of the audio events, hence referred to as weakly labelled data. Two subtasks are defined in this challenge including audio tagging and sound event detection using this weakly labelled data. We propose a convolutional recurrent neural network (CRNN) with learnable gated linear units (GLUs) non-linearity applied on the log Mel spectrogram. In addition, we propose a temporal attention method along the frames to predict the locations of each audio event in a chunk from the weakly labelled data. The performances of our systems were ranked the 1st and the 2nd as a team in these two sub-tasks of DCASE 2017 challenge with F value 55.6% and Equal error 0.73, respectively.

Polyphonic Sound Event Detection (SED) in real-world recordings is a challenging task because of the dynamic polyphony level, intensity, and duration of sound events. Current polyphonic SED systems fail to model the temporal structure of sound events explicitly and instead attempt to look at which sound events are present at each audio frame. Consequently, the event-wise detection performance is much lower than the segment-wise detection performance. In this work, we propose a joint model approach to improve the temporal localization of sound events using a multi-task learning setup. The first task predicts which sound events are present at each time frame; we call this branch `Sound Event Detection (SED) model', while the second task predicts if a sound event is present or not at each frame; we call this branch `Sound Activity Detection (SAD) model'. We verify the proposed joint model by comparing it with a separate implementation of both tasks aggregated together from individual task predictions. Our experiments on the URBAN-SED dataset show that the proposed joint model can alleviate False Positive (FP) and False Negative (FN) errors and improve both the segment-wise and the event-wise metrics.

Multi-granularity acoustic information fusion for sound event detection

Acoustic environments affect acoustic characteristics of sound to be recognized by physically interacting with sound wave propagation. Thus, training acoustic models for audio and speech tasks requires regularization on various acoustic environments in order to achieve robust performance in real life applications. We propose FilterAugment, a data augmentation method for regularization of acoustic models on various acoustic environments. FilterAugment mimics acoustic filters by applying different weights on frequency bands, therefore enables model to extract relevant information from wider frequency region. It is an improved version of frequency masking which masks information on random frequency bands. FilterAugment improved sound event detection (SED) model performance by 6.50% while frequency masking only improved 2.13% in terms of polyphonic sound detection score (PSDS). It achieved equal error rate (EER) of 1.22% when applied to a text-independent speaker verification model, outperforming model used frequency masking with EER of 1.26%. Prototype of FilterAugment was applied in our participation in DCASE 2021 challenge task 4, and played a major role in achieving the 3rd rank.

Each edition of the challenge on Detection and Classification of Acoustic Scenes and Events DCASE contained several tasks involving sound event detection in different setups. DCASE 2017 presented participants with three such tasks, each having specific datasets and detection requirements: Task 2, in which target sound events were very rare in both training and testing data, Task 3 having overlapping events annotated in real-life audio, and Task 4, in which only weakly labeled data were available for training. In this paper, we present three tasks, including the datasets and baseline systems, and analyze the challenge entries for each task. We observe the popularity of methods using deep neural networks, and the still widely used mel frequency-based representations, with only few approaches standing out as radically different. Analysis of the systems behavior reveals that task-specific optimization has a big role in producing good performance; however, often this optimization closely follows the ranking metric, and its maximization/minimization does not result in universally good performance. We also introduce the calculation of confidence intervals based on a jackknife resampling procedure to perform statistical analysis of the challenge results. The analysis indicates that while the 95% confidence intervals for many systems overlap, there are significant differences in performance between the top systems and the baseline for all tasks.

Audio pattern recognition is an important research topic in the machine learning area, and includes several tasks such as audio tagging, acoustic scene classification, music classification, speech emotion classification and sound event detection. Recently, neural networks have been applied to tackle audio pattern recognition problems. However, previous systems are built on specific datasets with limited durations. Recently, in computer vision and natural language processing, systems pretrained on large-scale datasets have generalized well to several tasks. However, there is limited research on pretraining systems on large-scale datasets for audio pattern recognition. In this paper, we propose pretrained audio neural networks (PANNs) trained on the large-scale AudioSet dataset. These PANNs are transferred to other audio related tasks. We investigate the performance and computational complexity of PANNs modeled by a variety of convolutional neural networks. We propose an architecture called Wavegram-Logmel-CNN using both log-mel spectrogram and waveform as input feature. Our best PANN system achieves a state-of-the-art mean average precision (mAP) of 0.439 on AudioSet tagging, outperforming the best previous system of 0.392. We transfer PANNs to six audio pattern recognition tasks, and demonstrate state-of-the-art performance in several of those tasks. We have released the source code and pretrained models of PANNs: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/qiuqiangkong/audioset_tagging_cnn</uri> .

This paper proposes a sound event detection (SED) model using an EfficientNet-B2 and an attention and aggregation-based feature pyramid network (A2-FPN). In particular, the EfficientNet-B2 is first obtained from the pretrained model on the basis of the pretraining, sampling, labeling, and aggregation (PSLA) framework. Then, the A2-FPN module is applied to the outputs of the layers of the EfficientNet-B2 to deal with the different time and frequency resolutions from acoustic features. The aggregated feature map from the A2-FPN module is used as input features to two bidirectional gated recurrent unit layers. Specifically, the proposed A2-FPN-based SED model is trained by the mean-teacher approach to utilize weakly labeled and unlabeled data. Finally, the proposed A2-FPN-based SED model is applied to the detection and classification of acoustic scenes and events (DCASE) 2021 Challenge Task 4. Consequently, it is shown that the polyphonic sound event detection score (PSDS) 1 and 2 of the proposed A2-FPN-based SED model are the higher of 0.03 and 0.172, respectively, than those of the DCASE 2021 Challenge Task 4 baseline.

Polyphonic sound event localization and detection (SELD) aims at detecting types of sound events with corresponding temporal activities and spatial locations. In this paper, a trackwise ensemble event independent network with a novel data augmentation method is proposed. The proposed model is based on our previous proposed Event-Independent Network V2 and is extended by conformer blocks and dense blocks. The track-wise ensemble model with track-wise output format is proposed to solve an ensemble model problem for track-wise output format that track permutation may occur among different models. The data augmentation approach contains several data augmentation chains, which are composed of random combinations of several data augmentation operations. The method also utilizes log-mel spectrograms, intensity vectors, and Spatial Cues-Augmented Log-Spectrogram (SALSA) for different models. We evaluate our proposed method in the Task of the L3DAS22 challenge and obtain the top ranking solution with a location-dependent F-score to be 0.699. Source code is released <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> .

Recent literature has demonstrated that the use of per-channel energy normalization (PCEN), has significant performance improvements over traditional log-scaled mel-frequency spectrograms in acoustic sound event detection (SED) in a multi-class setting with overlapping events. However, the configuration of PCEN's parameters is sensitive to the recording environment, the characteristics of the class of events of interest, and the presence of multiple overlapping events. This leads to improvements on a class-by-class basis, but poor cross-class performance. In this article, we experiment using PCEN spectrograms as an alternative method for SED in urban audio using the UrbanSED dataset, demonstrating per-class improvements based on parameter configuration. Furthermore, we address cross-class performance with PCEN using a novel method, Multi-Rate PCEN (MRPCEN). We demonstrate cross-class SED performance with MRPCEN, demonstrating improvements to cross-class performance compared to traditional single-rate PCEN.

Convolutional recurrent neural network (CRNN) with attention mechanisms has achieved the state-of-the-art performance in weakly labelled sound event detection (SED). We find the state-of-the-art CRNN only use single time-frequency scale to extract features of sound events. In this paper, we propose a multi-scale gated attention (MGA) CRNN for weakly labelled SED to extract more useful representations. MGA is concatenated by multi-scale context gating (CG), "Squeeze and Excitation" (SE) block and two convolutional layers. Multi-scale CG extracts multi-scale information of sound events and calculates attention weights at multiple resolutions for each feature map. The subsequent SE block learns the inter-dependencies between the channels of feature maps. MGA extract discriminative representations by selecting significant spatial feature and channel feature. Our method has achieved state-of-the-art results with 61.1% audio tagging F1 score on DCASE2017 Task 4 evaluation set and 32.16% event-based F1 score on DCASE2018 Task 4 evaluation set.

Sound event localization and detection have been applied in various fields. Due to the polyphony and noise interference, it becomes challenging to accurately predict the sound event and their occurrence locations. Aiming at this problem, we propose a Multiple Attention Fusion ResNet, which uses ResNet34 as the base network. Given the situation that the sound duration is not fixed, and there are multiple polyphonic and noise, we introduce the Gated Channel Transform to enhance the residual basic block. This enables the model to capture contextual information, evaluate channel weights, and reduce the interference caused by polyphony and noise. Furthermore, Split Attention is introduced to the model for capturing cross-channel information, which enhances the ability to distinguish the polyphony. Finally, Coordinate Attention is introduced to the model so that the model can focus on both the channel information and spatial location information of sound events. Experiments were conducted on two different datasets, TAU-NIGENS Spatial Sound Events 2020, and TAU-NIGENS Spatial Sound Events 2021. The results demonstrate that the proposed model significantly outperforms state-of-the-art methods under multiple polyphonic and noise-directional interference environments and it achieves competitive performance under a single polyphonic environment.