Acoustic Model Adaptation from Raw Waveforms with Sincnet

Fainberg, Joachim; Klejch, Ondrej; Loweimi, Erfan; Bell, Peter; Renals, Steve

Abstract

Raw waveform acoustic modelling has recently gained interest due to neural networks' ability to learn feature extraction, and the potential for finding better representations for a given scenario than hand-crafted features. SincNet has been proposed to reduce the number of parameters required in raw-waveform modelling, by restricting the filter functions, rather than having to learn every tap of each filter. We study the adaptation of the SincNet filter parameters from adults' to children's speech, and show that the parameterisation of the SincNet layer is well suited for adaptation in practice: we can efficiently adapt with a very small number of parameters, producing error rates comparable to techniques using orders of magnitude more parameters.

Citation

Fainberg, J., Klejch, O., Loweimi, E., Bell, P., & Renals, S. (2019, December). Acoustic Model Adaptation from Raw Waveforms with Sincnet. Presented at 2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), Singapore

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)
• Start Date: Dec 14, 2019
• End Date: Dec 18, 2019
• Online Publication Date: Feb 20, 2020
• Publication Date: 2019-12
• Deposit Date: Apr 3, 2024
• Publisher: Institute of Electrical and Electronics Engineers
• Book Title: 2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)
• DOI: https://doi.org/10.1109/asru46091.2019.9003974
• Public URL: http://researchrepository.napier.ac.uk/Output/3585879