Skip to Main content Skip to Navigation
Conference papers

CGCNN: COMPLEX GABOR CONVOLUTIONAL NEURAL NETWORK ON RAW SPEECH

Abstract : Convolutional Neural Networks (CNN) have been used in Automatic Speech Recognition (ASR) to learn representations directly from the raw signal instead of hand-crafted acoustic features, providing a richer and lossless input signal. Recent researches propose to inject prior acoustic knowledge to the first convolutional layer by integrating the shape of the impulse responses in order to increase both the interpretabil-ity of the learnt acoustic model, and its performances. We propose to combine the complex Gabor filter with complex-valued deep neural networks to replace usual CNN weights kernels, to fully take advantage of its optimal time-frequency resolution and of the complex domain. The conducted experiments on the TIMIT phoneme recognition task shows that the proposed approach reaches top-of-the-line performances while remaining interpretable.
Complete list of metadatas

Cited literature [27 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02474746
Contributor : Paul-Gauthier Noé <>
Submitted on : Tuesday, February 11, 2020 - 3:46:50 PM
Last modification on : Wednesday, February 19, 2020 - 1:21:34 AM

File

gabor_complex_cnn_final.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02474746, version 1

Collections

Citation

Paul-Gauthier Noé, Titouan Parcollet, Mohamed Morchid. CGCNN: COMPLEX GABOR CONVOLUTIONAL NEURAL NETWORK ON RAW SPEECH. ICASSP 2020, May 2020, Barcelone, Spain. ⟨hal-02474746⟩

Share

Metrics

Record views

31

Files downloads

7