TimeScaleNet : a Multiresolution Approach for Raw Audio Recognition using Learnable Biquadratic IIR Filters and Residual Networks of Depthwise-Separable One-Dimensional Atrous Convolutions
Résumé
In the present paper, we show the benefit of a multi-resolution approach that allows to encode the relevant information contained in unprocessed time domain acoustic signals. TimeScaleNet aims at learning an efficient representation of a sound, by learning time dependencies both at the sample level and at the frame level. The proposed approach allows to improve the interpretability of the learning scheme, by unifying advanced deep learning and signal processing techniques. In particular, TimeScaleNet's architecture introduces a new form of recurrent neural layer, which is directly inspired from digital IIR signal processing. This layer acts as a learnable passband biquadratic digital IIR filterbank. The learnable filterbank allows to build a time-frequency-like feature map that self-adapts to the specific recognition task and dataset, with a large receptive field and very few learnable parameters. The obtained frame-level feature map is then processed using a residual network of depthwise separable atrous convolutions. This second scale of analysis aims at efficiently encoding relationships between the time fluctuations at the frame timescale, in different learnt pooled frequency bands, in the range of [20 ms ; 200 ms]. TimeScaleNet is tested both using the Speech Commands Dataset and the ESC-10 Dataset. We report a very high mean accuracy of 94.87 ± 0.24% (macro averaged F1-score : 94.9 ± 0.24%) for speech recognition, and a rather moderate accuracy of 69.71 ± 1.91% (macro averaged F1-score : 70.14 ± 1.57%) for the environmental sound classification task.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...