Speaker Recognition Benchmark Using the CHiME-5 Corpus

Abstract

In this paper, we introduce a speaker recognition benchmark derived from the publicly-available CHiME-5 corpus. Our goal is to foster research that tackles the challenging artifacts introduced by far-field multi-speaker recordings of naturally occurring spoken interactions. The benchmark comprises four tasks that involve enrollment and test conditions with single-speaker and/or multi-speaker recordings. Additionally, it supports performance comparisons between close-talking vs distant/far-field microphone recordings, and single-microphone vs microphone-array approaches. We validate the evaluation design with a single-microphone state-of-the-art DNN speaker recognition and diarization system (that we are making publicly available). The results show that the proposed tasks are very challenging, and can be used to quantify the performance gap due to the degradations present in far-field multi-speaker recordings.

Dialect classification based on the speed and the pause of speech utterances*

“…MFCC (Davis & Mermelstein, 1980)는 저주파 대역에서의 오디 오 신호를 세밀하게 표현하고 고주파 대역에서는 상대적으로 소 략하게 표현하는 특성을 가진다. MFCC는 방언 분류 (Chowdhury et al, 2020;Khurana et al, 2017;Mukherjee et al, 2020;Tawaqal & Suyanto, 2021;Wan et al, 2022;Wang et al, 2021;Zhang & Hansen, 2018)뿐만 아니라 음성인식 (Shahnawazuddin et al, 2016;Tüske et al, 2014;Wallington et al, 2021), 화자 인식 (Fenu et al, 2020;Garcia-Romero et al, 2019;Lin & Mak, 2020;Pappagari et al, 2020) 그리고 감정 인식 (Keesing et al, 2021;Likitha et al, 2017;Sarma et al, 2018;Saste & Jagdale, 2017;Seo & Lee, 2022)과 같은 음성 분석의 다양한 분야에서도 활용된다.…”

AutoSpeech 2020: The Second Automated Machine Learning Challenge for Speech Classification

“…In the past few decades, machine learning, especially deep learning, has achieved remarkable breakthroughs in a wide range of speech tasks, e.g., speech recognition [1,2], speaker verification [3,4,5], language identification [6,7] and emotion classification [8,9]. Each speech task has its own specific techniques in achieving the state-of-the-art results [3,6,8,10,11,12], which require efforts of a large number of experts. Thus, it is very difficult to switch between different speech tasks without human efforts.…”