VECTOR-QUANTIZED SPEECH SEPARATION

Jiang, Xilin

Permalink

https://hdl.handle.net/2142/115012 Copy

Description

Title

VECTOR-QUANTIZED SPEECH SEPARATION

Author(s)

Jiang, Xilin

Issue Date

2021-12

Keyword(s)

• source separation
• vector quantization
• signal representation
• neural networks

Language

en

Abstract

In this paper, we present a novel method to address the single-channel speech separation problem. We propose a two-step training procedure for speech separation in a discrete latent space. In the first step, we learn multiple vector-quantized codebooks to optimize reconstruction and entropy and functions to transform between discrete codes and waveform. In the second step, we train multiple classifiers to select codes from codebooks to synthesize speech sources. The proposed method reaches comparable speech separation performance and is general enough to be applicable to other regression problems.

Graduation Semester

2022-10-15T16:08:38-05:00

Type of Resource

text

Owning Collections