USED: Universal Speaker Extraction and Diarization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ao, Junyi, Yıldırım, Mehmet Sinan, Tao, Ruijie, Ge, Meng, Wang, Shuai, Qian, Yanmin, Li, Haizhou
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909457269653504
author Ao, Junyi
Yıldırım, Mehmet Sinan
Tao, Ruijie
Ge, Meng
Wang, Shuai
Qian, Yanmin
Li, Haizhou
author_facet Ao, Junyi
Yıldırım, Mehmet Sinan
Tao, Ruijie
Ge, Meng
Wang, Shuai
Qian, Yanmin
Li, Haizhou
contents Speaker extraction and diarization are two enabling techniques for real-world speech applications. Speaker extraction aims to extract a target speaker's voice from a speech mixture, while speaker diarization demarcates speech segments by speaker, annotating `who spoke when'. Previous studies have typically treated the two tasks independently. In practical applications, it is more meaningful to have knowledge about `who spoke what and when', which is captured by the two tasks. The two tasks share a similar objective of disentangling speakers. Speaker extraction operates in the frequency domain, whereas diarization is in the temporal domain. It is logical to believe that speaker activities obtained from speaker diarization can benefit speaker extraction, while the extracted speech offers more accurate speaker activity detection than the speech mixture. In this paper, we propose a unified model called Universal Speaker Extraction and Diarization (USED) to address output inconsistency and scenario mismatch issues. It is designed to manage speech mixtures with varying overlap ratios and variable number of speakers. We show that the USED model significantly outperforms the competitive baselines for speaker extraction and diarization tasks on LibriMix and SparseLibriMix datasets. We further validate the diarization performance on CALLHOME, a dataset based on real recordings, and experimental results indicate that our model surpasses recently proposed approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2309_10674
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle USED: Universal Speaker Extraction and Diarization
Ao, Junyi
Yıldırım, Mehmet Sinan
Tao, Ruijie
Ge, Meng
Wang, Shuai
Qian, Yanmin
Li, Haizhou
Sound
Audio and Speech Processing
Speaker extraction and diarization are two enabling techniques for real-world speech applications. Speaker extraction aims to extract a target speaker's voice from a speech mixture, while speaker diarization demarcates speech segments by speaker, annotating `who spoke when'. Previous studies have typically treated the two tasks independently. In practical applications, it is more meaningful to have knowledge about `who spoke what and when', which is captured by the two tasks. The two tasks share a similar objective of disentangling speakers. Speaker extraction operates in the frequency domain, whereas diarization is in the temporal domain. It is logical to believe that speaker activities obtained from speaker diarization can benefit speaker extraction, while the extracted speech offers more accurate speaker activity detection than the speech mixture. In this paper, we propose a unified model called Universal Speaker Extraction and Diarization (USED) to address output inconsistency and scenario mismatch issues. It is designed to manage speech mixtures with varying overlap ratios and variable number of speakers. We show that the USED model significantly outperforms the competitive baselines for speaker extraction and diarization tasks on LibriMix and SparseLibriMix datasets. We further validate the diarization performance on CALLHOME, a dataset based on real recordings, and experimental results indicate that our model surpasses recently proposed approaches.
title USED: Universal Speaker Extraction and Diarization
topic Sound
Audio and Speech Processing
url https://arxiv.org/abs/2309.10674