An Effective Transformer-based Contextual Model and Temporal Gate Pooling for Speaker Identification

Harunori Kawano, Sota Shimizu
8/22/2023
cs.SDcs.AIcs.LG

Official code: HarunoriKawano/speaker-identification-with-tgp (7 stars, Apache-2.0, last commit 2023-09-13).

Abstract

Wav2vec2 has achieved success in applying Transformer architecture and self-supervised learning to speech recognition. Recently, these have come to be used not only for speech recognition but also for the entire speech processing. This paper introduces an effective end-to-end speaker identification model applied Transformer-based contextual model. We explored the relationship between the hyper-parameters and the performance in order to discern the structure of an effective model. Furthermore, we propose a pooling method, Temporal Gate Pooling, with powerful learning ability for speaker identification. We applied Conformer as encoder and BEST-RQ for pre-training and conducted an evaluation utilizing the speaker identification of VoxCeleb1. The proposed method has achieved an accuracy of 87.1% with 28.5M parameters, demonstrating comparable precision to wav2vec2 with 317.7M parameters. Code is available at https://github.com/HarunoriKawano/speaker-identification-with-tgp.

View on arXivView PDF

Code Implementations(1)

HarunoriKawano/speaker-identification-with-tgpOfficial100%
Apache-2.0
71PythonAug 22, 20233 years agoApache-2.0

Similar Papers

Speech2Phone: A Novel and Efficient Method for Training Speaker Recognition Models

84% match1 repo92020

ESPnet-SPK: full pipeline speaker embedding toolkit with reproducible recipes, self-supervised front-ends, and off-the-shelf models

84% match1 repo10.0k2024

UniSpeech-SAT: Universal Speech Representation Learning with Speaker Aware Pre-Training

84% match1 repo4912021

Incorporating Talker Identity Aids With Improving Speech Recognition in Adversarial Environments

83% match2024

Evaluating Speaker Identity Coding in Self-supervised Models and Humans

83% match2024

VieSpeaker: A Large-Scale Vietnamese Speaker Recognition Dataset Beyond Visual Dependency

82% match2026

Cite this paper

@article{kawano2023effective,
  title  = {An Effective Transformer-based Contextual Model and Temporal Gate Pooling for Speaker Identification},
  author = {Harunori Kawano and Sota Shimizu},
  year   = {2023},
  eprint = {2308.11241},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2308.11241v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2308.11241.svg)](https://codeofpaper.com/paper/2308.11241)

Discussion