Many state-of-the-art systems for audio tagging and sound event detection employ convolutional recurrent neural architectures. Typically, they are trained in a mean teacher setting to deal with the heterogeneous annotation of the available data. In this work, we present a thorough analysis of how changing the temporal resolution of these convolutional recurrent neural networks - which can be done by simply adapting their pooling operations - impacts their performance. By using a variety of evaluation metrics, we investigate the effects of adapting this design parameter under several sound recognition scenarios involving different needs in terms of temporal localization.
Large-scale sound recognition data sets typically consist of acoustic recordings obtained from multimedia libraries. As a consequence, modalities other than audio can often be exploited to improve the outputs of models designed for associated tasks. Frequently, however, not all contents are available for all samples of such a collection: For example, the original material may have been removed from the source platform at some point, and therefore, non-auditory features can no longer be acquired. We demonstrate that a multi-encoder framework can be employed to deal with this issue by applying this method to attention-based deep learning systems, which are currently part of the state of the art in the domain of sound recognition. More specifically, we show that the proposed model extension can successfully be utilized to incorporate partially available visual information into the operational procedures of such networks, which normally only use auditory features during training and inference. Experimentally, we verify that the considered approach leads to improved predictions in a number of evaluation scenarios pertaining to audio tagging and sound event detection. Additionally, we scrutinize some properties and limitations of the presented technique.
Standard machine learning models for tagging and classifying acoustic signals cannot handle classes that were not seen during training. Zero-Shot (ZS) learning overcomes this restriction by predicting classes based on adaptable class descriptions. This study sets out to investigate the effectiveness of self-attention-based audio embedding architectures for ZS learning. To this end, we compare the very recent patchout spectrogram transformer with two classic convolutional architectures. We evaluate these three architectures on three tasks and on three different benchmark datasets: general-purpose tagging on AudioSet, environmental sound classification on ESC-50, and instrument tagging on OpenMIC. Our results show that the self-attention-based embedding methods outperform both compared convolutional architectures in all of these settings. By designing training and test data accordingly, we observe that prediction performance suffers significantly when the `semantic distance' between training and new test classes is large, an effect that will deserve more detailed investigations.
It is easier to hear birds than see them, however, they still play an essential role in nature and they are excellent indicators of deteriorating environmental quality and pollution. Recent advances in Machine Learning and Convolutional Neural Networks allow us to detect and classify bird sounds, by doing this, we can assist researchers in monitoring the status and trends of bird populations and biodiversity in ecosystems. We propose a sound detection and classification pipeline for analyzing complex soundscape recordings and identify birdcalls in the background. Our pipeline learns from weak labels, classifies fine-grained bird vocalizations in the wild, and is robust against background sounds (e.g., airplanes, rain, etc). Our solution achieved 10th place of 816 teams at the BirdCLEF 2021 Challenge hosted on Kaggle.
Machines that can represent and describe environmental soundscapes have practical potential, e.g., for audio tagging and captioning systems. Prevailing learning paradigms have been relying on parallel audio-text data, which is, however, scarcely available on the web. We propose VIP-ANT that induces \textbf{A}udio-\textbf{T}ext alignment without using any parallel audio-text data. Our key idea is to share the image modality between bi-modal image-text representations and bi-modal image-audio representations; the image modality functions as a pivot and connects audio and text in a tri-modal embedding space implicitly. In a difficult zero-shot setting with no paired audio-text data, our model demonstrates state-of-the-art zero-shot performance on the ESC50 and US8K audio classification tasks, and even surpasses the supervised state of the art for Clotho caption retrieval (with audio queries) by 2.2\% R@1. We further investigate cases of minimal audio-text supervision, finding that, e.g., just a few hundred supervised audio-text pairs increase the zero-shot audio classification accuracy by 8\% on US8K. However, to match human parity on some zero-shot tasks, our empirical scaling experiments suggest that we would need about $2^{21} \approx 2M$ supervised audio-caption pairs. Our work opens up new avenues for learning audio-text connections with little to no parallel audio-text data.
Deep learning techniques for separating audio into different sound sources face several challenges. Standard architectures require training separate models for different types of audio sources. Although some universal separators employ a single model to target multiple sources, they have difficulty generalizing to unseen sources. In this paper, we propose a three-component pipeline to train a universal audio source separator from a large, but weakly-labeled dataset: AudioSet. First, we propose a transformer-based sound event detection system for processing weakly-labeled training data. Second, we devise a query-based audio separation model that leverages this data for model training. Third, we design a latent embedding processor to encode queries that specify audio targets for separation, allowing for zero-shot generalization. Our approach uses a single model for source separation of multiple sound types, and relies solely on weakly-labeled data for training. In addition, the proposed audio separator can be used in a zero-shot setting, learning to separate types of audio sources that were never seen in training. To evaluate the separation performance, we test our model on MUSDB18, while training on the disjoint AudioSet. We further verify the zero-shot performance by conducting another experiment on audio source types that are held-out from training. The model achieves comparable Source-to-Distortion Ratio (SDR) performance to current supervised models in both cases.
The great success of transformer-based models in natural language processing (NLP) has led to various attempts at adapting these architectures to other domains such as vision and audio. Recent work has shown that transformers can outperform Convolutional Neural Networks (CNNs) on vision and audio tasks. However, one of the main shortcomings of transformer models, compared to the well-established CNNs, is the computational complexity. In transformers, the compute and memory complexity is known to grow quadratically with the input length. Therefore, there has been extensive work on optimizing transformers, but often at the cost of degrading predictive performance. In this work, we propose a novel method to optimize and regularize transformers on audio spectrograms. Our proposed models achieve a new state-of-the-art performance on Audioset and can be trained on a single consumer-grade GPU. Furthermore, we propose a transformer model that outperforms CNNs in terms of both performance and training speed. Source code: https://github.com/kkoutini/PaSST
Sound event detection (SED) has gained increasing attention with its wide application in surveillance, video indexing, etc. Existing models in SED mainly generate frame-level prediction, converting it into a sequence multi-label classification problem. A critical issue with the frame-based model is that it pursues the best frame-level prediction rather than the best event-level prediction. Besides, it needs post-processing and cannot be trained in an end-to-end way. This paper firstly presents the one-dimensional Detection Transformer (1D-DETR), inspired by Detection Transformer for image object detection. Furthermore, given the characteristics of SED, the audio query branch and a one-to-many matching strategy for fine-tuning the model are added to 1D-DETR to form Sound Event Detection Transformer (SEDT). To our knowledge, SEDT is the first event-based and end-to-end SED model. Experiments are conducted on the URBAN-SED dataset and the DCASE2019 Task4 dataset, and both show that SEDT can achieve competitive performance.
We study the merit of transfer learning for two sound recognition problems, i.e., audio tagging and sound event detection. Employing feature fusion, we adapt a baseline system utilizing only spectral acoustic inputs to also make use of pretrained auditory and visual features, extracted from networks built for different tasks and trained with external data. We perform experiments with these modified models on an audiovisual multi-label data set, of which the training partition contains a large number of unlabeled samples and a smaller amount of clips with weak annotations, indicating the clip-level presence of 10 sound categories without specifying the temporal boundaries of the active auditory events. For clip-based audio tagging, this transfer learning method grants marked improvements. Addition of the visual modality on top of audio also proves to be advantageous in this context. When it comes to generating transcriptions of audio recordings, the benefit of pretrained features depends on the requested temporal resolution: for coarse-grained sound event detection, their utility remains notable. But when more fine-grained predictions are required, performance gains are strongly reduced due to a mismatch between the problem at hand and the goals of the models from which the pretrained vectors were obtained.
Sound event detection is an important facet of audio tagging that aims to identify sounds of interest and define both the sound category and time boundaries for each sound event in a continuous recording. With advances in deep neural networks, there has been tremendous improvement in the performance of sound event detection systems, although at the expense of costly data collection and labeling efforts. In fact, current state-of-the-art methods employ supervised training methods that leverage large amounts of data samples and corresponding labels in order to facilitate identification of sound category and time stamps of events. As an alternative, the current study proposes a semi-supervised method for generating pseudo-labels from unsupervised data using a student-teacher scheme that balances self-training and cross-training. Additionally, this paper explores post-processing which extracts sound intervals from network prediction, for further improvement in sound event detection performance. The proposed approach is evaluated on sound event detection task for the DCASE2020 challenge. The results of these methods on both "validation" and "public evaluation" sets of DESED database show significant improvement compared to the state-of-the art systems in semi-supervised learning.
Convolutional Neural Networks have been extensively explored in the task of automatic music tagging. The problem can be approached by using either engineered time-frequency features or raw audio as input. Modulation filter bank representations that have been actively researched as a basis for timbre perception have the potential to facilitate the extraction of perceptually salient features. We explore end-to-end learned front-ends for audio representation learning, ModNet and SincModNet, that incorporate a temporal modulation processing block. The structure is effectively analogous to a modulation filter bank, where the FIR filter center frequencies are learned in a data-driven manner. The expectation is that a perceptually motivated filter bank can provide a useful representation for identifying music features. Our experimental results provide a fully visualisable and interpretable front-end temporal modulation decomposition of raw audio. We evaluate the performance of our model against the state-of-the-art of music tagging on the MagnaTagATune dataset. We analyse the impact on performance for particular tags when time-frequency bands are subsampled by the modulation filters at a progressively reduced rate. We demonstrate that modulation filtering provides promising results for music tagging and feature representation, without using extensive musical domain knowledge in the design of this front-end.
Having access to multi-modal cues (e.g. vision and audio) empowers some cognitive tasks to be done faster compared to learning from a single modality. In this work, we propose to transfer knowledge across heterogeneous modalities, even though these data modalities may not be semantically correlated. Rather than directly aligning the representations of different modalities, we compose audio, image, and video representations across modalities to uncover richer multi-modal knowledge. Our main idea is to learn a compositional embedding that closes the cross-modal semantic gap and captures the task-relevant semantics, which facilitates pulling together representations across modalities by compositional contrastive learning. We establish a new, comprehensive multi-modal distillation benchmark on three video datasets: UCF101, ActivityNet, and VGGSound. Moreover, we demonstrate that our model significantly outperforms a variety of existing knowledge distillation methods in transferring audio-visual knowledge to improve video representation learning. Code is released here: https://github.com/yanbeic/CCL.
Crowdsourcing has become a common approach for annotating large amounts of data. It has the advantage of harnessing a large workforce to produce large amounts of data in a short time, but comes with the disadvantage of employing non-expert annotators with different backgrounds. This raises the problem of data reliability, in addition to the general question of how to combine the opinions of multiple annotators in order to estimate the ground truth. This paper presents a study of the annotations and annotators' reliability for audio tagging. We adapt the use of Krippendorf's alpha and multi-annotator competence estimation (MACE) for a multi-labeled data scenario, and present how MACE can be used to estimate a candidate ground truth based on annotations from non-expert users with different levels of expertise and competence.
In the past decade, convolutional neural networks (CNNs) have been widely adopted as the main building block for end-to-end audio classification models, which aim to learn a direct mapping from audio spectrograms to corresponding labels. To better capture long-range global context, a recent trend is to add a self-attention mechanism on top of the CNN, forming a CNN-attention hybrid model. However, it is unclear whether the reliance on a CNN is necessary, and if neural networks purely based on attention are sufficient to obtain good performance in audio classification. In this paper, we answer the question by introducing the Audio Spectrogram Transformer (AST), the first convolution-free, purely attention-based model for audio classification. We evaluate AST on various audio classification benchmarks, where it achieves new state-of-the-art results of 0.485 mAP on AudioSet, 95.6% accuracy on ESC-50, and 98.1% accuracy on Speech Commands V2.
A good joint training framework is very helpful to improve the performances of weakly supervised audio tagging (AT) and acoustic event detection (AED) simultaneously. In this study, we propose three methods to improve the best teacher-student framework in the IEEE AASP Challenge on Detection and Classification of Acoustic Scenes and Events (DCASE) 2019 Task 4 for both audio tagging and acoustic events detection tasks. A frame-level target-events based deep feature distillation is first proposed, which aims to leverage the potential of limited strong-labeled data in weakly supervised framework to learn better intermediate feature maps. Then, we propose an adaptive focal loss and two-stage training strategy to enable an effective and more accurate model training, where the contribution of hard and easy acoustic events to the total cost function can be automatically adjusted. Furthermore, an event-specific post processing is designed to improve the prediction of target event time-stamps. Our experiments are performed on the public DCASE 2019 Task 4 dataset, results show that our approach achieves competitive performances in both AT (81.2\% F1-score) and AED (49.8\% F1-score) tasks.
Multi-label audio tagging consists of assigning sets of tags to audio recordings. At inference time, thresholds are applied on the confidence scores outputted by a probabilistic classifier, in order to decide which classes are detected active. In this work, we consider having at disposal a trained classifier and we seek to automatically optimize the decision thresholds according to a performance metric of interest, in our case F-measure (micro-F1). We propose a new method, called SGL-Thresh for Surrogate Gradient Learning of Thresholds, that makes use of gradient descent. Since F1 is not differentiable, we propose to approximate the thresholding operation gradients with the gradients of a sigmoid function. We report experiments on three datasets, using state-of-the-art pre-trained deep neural networks. In all cases, SGL-Thresh outperformed three other approaches: a default threshold value (defThresh), an heuristic search algorithm and a method estimating F1 gradients numerically. It reached 54.9\% F1 on AudioSet eval, compared to 50.7% with defThresh. SGL-Thresh is very fast and scalable to a large number of tags. To facilitate reproducibility, data and source code in Pytorch are available online: https://github.com/topel/SGL-Thresh
In this article, we adapted five recent SSL methods to the task of audio classification. The first two methods, namely Deep Co-Training (DCT) and Mean Teacher (MT), involve two collaborative neural networks. The three other algorithms, called MixMatch (MM), ReMixMatch (RMM), and FixMatch (FM), are single-model methods that rely primarily on data augmentation strategies. Using the Wide-ResNet-28-2 architecture in all our experiments, 10% of labeled data and the remaining 90% as unlabeled data for training, we first compare the error rates of the five methods on three standard benchmark audio datasets: Environmental Sound Classification (ESC-10), UrbanSound8K (UBS8K), and Google Speech Commands (GSC). In all but one cases, MM, RMM, and FM outperformed MT and DCT significantly, MM and RMM being the best methods in most experiments. On UBS8K and GSC, MM achieved 18.02% and 3.25% error rate (ER), respectively, outperforming models trained with 100% of the available labeled data, which reached 23.29% and 4.94%, respectively. RMM achieved the best results on ESC-10 (12.00% ER), followed by FM which reached 13.33%. Second, we explored adding the mixup augmentation, used in MM and RMM, to DCT, MT, and FM. In almost all cases, mixup brought consistent gains. For instance, on GSC, FM reached 4.44% and 3.31% ER without and with mixup. Our PyTorch code will be made available upon paper acceptance at https:// github. com/ Labbe ti/ SSLH.
Data augmentation is an inexpensive way to increase training data diversity and is commonly achieved via transformations of existing data. For tasks such as classification, there is a good case for learning representations of the data that are invariant to such transformations, yet this is not explicitly enforced by classification losses such as the cross-entropy loss. This paper investigates the use of training objectives that explicitly impose this consistency constraint and how it can impact downstream audio classification tasks. In the context of deep convolutional neural networks in the supervised setting, we show empirically that certain measures of consistency are not implicitly captured by the cross-entropy loss and that incorporating such measures into the loss function can improve the performance of audio classification systems. Put another way, we demonstrate how existing augmentation methods can further improve learning by enforcing consistency.
Audio tagging is an active research area and has a wide range of applications. Since the release of AudioSet, great progress has been made in advancing model performance, which mostly comes from the development of novel model architectures and attention modules. However, we find that appropriate training techniques are equally important for building audio tagging models with AudioSet, but have not received the attention they deserve. To fill the gap, in this work, we present PSLA, a collection of training techniques that can noticeably boost the model accuracy including ImageNet pretraining, balanced sampling, data augmentation, label enhancement, model aggregation and their design choices. By training an EfficientNet with these techniques, we obtain a single model (with 13.6M parameters) and an ensemble model that achieve mean average precision (mAP) scores of 0.444 and 0.474 on AudioSet, respectively, outperforming the previous best system of 0.439 with 81M parameters. In addition, our model also achieves a new state-of-the-art mAP of 0.567 on FSD50K.
Acoustic scene classification systems using deep neural networks classify given recordings into pre-defined classes. In this study, we propose a novel scheme for acoustic scene classification which adopts an audio tagging system inspired by the human perception mechanism. When humans identify an acoustic scene, the existence of different sound events provides discriminative information which affects the judgement. The proposed framework mimics this mechanism using various approaches. Firstly, we employ three methods to concatenate tag vectors extracted using an audio tagging system with an intermediate hidden layer of an acoustic scene classification system. We also explore the multi-head attention on the feature map of an acoustic scene classification system using tag vectors. Experiments conducted on the detection and classification of acoustic scenes and events 2019 task 1-a dataset demonstrate the effectiveness of the proposed scheme. Concatenation and multi-head attention show a classification accuracy of 75.66 % and 75.58 %, respectively, compared to 73.63 % accuracy of the baseline. The system with the proposed two approaches combined demonstrates an accuracy of 76.75 %.