2021/07/30 by Debottam Dutta, Dutta, Debottam, Purvi Agrawal +3
Computer Science · Engineering · Mathematics · #Acoustics #Artificial intelligence #Audio and Speech Processing (eess.AS) #Computer science #Computer vision #FOS: Computer and information sciences #FOS: Electrical engineering #Filter (signal processing) #Filter bank #Mathematics #Music and Audio Processing #Parametric statistics #Pattern recognition (psychology) #Relevance (law) #Signal Processing (eess.SP) #Sound (cs.SD) #Spectrogram #Speech Recognition and Synthesis #Speech and Audio Processing #Speech recognition #Waveform #Weighting #cs.SD #eess.AS #eess.SP #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2107.14793
Submitted to 2021 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics(WASPAA 2021)
arxiv created 2021/07/30 · openalex publication_date 2021/07/30 · arxiv updated 2021/08/02 · openalex created_date 2022/07/25 · openalex updated_date 2026/08/06
In this work, we propose a multi-head relevance weighting framework to learn audio representations from raw waveforms. The audio waveform, split into windows of short duration, are processed with a 1-D convolutional layer of cosine modulated Gaussian filters acting as a learnable filterbank. The key novelty of the proposed framework is the introduction of multi-head relevance on the learnt filterbank representations. Each head of the relevance network is modelled as a separate sub-network. These heads perform representation enhancement by generating weight masks for different parts of the time-frequency representation learnt by the parametric acoustic filterbank layer. The relevance weighted representations are fed to a neural classifier and the whole system is trained jointly for the audio classification objective. Experiments are performed on the DCASE2020 Task 1A challenge as well as the Urban Sound Classification (USC) tasks. In these experiments, the proposed approach yields relative improvements of 10% and 23% respectively for the DCASE2020 and USC datasets over the mel-spectrogram baseline. Also, the analysis of multi-head relevance weights provides insights on the learned representations.