Trainable windows for SincNet architecture

SincNet architecture has shown significant benefits over traditional Convolutional Neural Networks (CNN), especially for speaker recognition applications. SincNet comprises parameterized Sinc functions as filters in the first layer followed by convolutional layers. Although SincNet is compact in nat...

Full description

Saved in:
Bibliographic Details
Published in:EURASIP journal on audio, speech, and music processing speech, and music processing, 2023-01, Vol.2023 (1), p.3-9, Article 3
Main Authors: H C, Prashanth, Rao, Madhav, Eledath, Dhanya, V, Ramasubramanian
Format: Article
Language:English
Subjects:
Acoustics
Artificial neural networks
Empirical Research
Engineering
Engineering Acoustics
Feature extraction
Filter banks
Investigations
Leakage
Mathematics in Music
Musical performances
Neural networks
Signal,Image and Speech Processing
SincNet architecture
Speaker recognition
Speech
Speech recognition
Trainable windows
Trigonometric functions
Window functions
Windows (intervals)
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:SincNet architecture has shown significant benefits over traditional Convolutional Neural Networks (CNN), especially for speaker recognition applications. SincNet comprises parameterized Sinc functions as filters in the first layer followed by convolutional layers. Although SincNet is compact in nature and offers top-level understanding of the features extracted, the effect of window function used in SincNet is not thoroughly addressed yet. Hamming and Hann are popularly used as the default time-localized windows to reduce spectral leakage. Hence, a comprehensive investigation of 28 different windowing functions on SincNet architecture towards speaker recognition task using TIMIT dataset was performed in this work. Additionally, “trainable” window functions were configured with tunable parameters to characterize the performance. The paper benchmarks the effect of the time-localized windowing function in terms of the bandwidth, side-lobe suppression, and spectral leakage for the filter banks employed in the first layer of the SincNet architecture. Trainable Gaussian and Cosine-Sum functions exhibited relative improvement of 41.46% and 82.11% in the sentence level classification error rate over Hamming window when employed on SincNet architecture.
ISSN:1687-4722
1687-4714
1687-4722
DOI:10.1186/s13636-023-00271-0