PMID- 33997106 OWN - NLM STAT- PubMed-not-MEDLINE LR - 20240402 IS - 2329-9290 (Print) IS - 2329-9304 (Electronic) VI - 29 DP - 2021 TI - Meta-learning with Latent Space Clustering in Generative Adversarial Network for Speaker Diarization. PG - 1204-1219 LID - 10.1109/taslp.2021.3061885 [doi] AB - The performance of most speaker diarization systems with x-vector embeddings is both vulnerable to noisy environments and lacks domain robustness. Earlier work on speaker diarization using generative adversarial network (GAN) with an encoder network (ClusterGAN) to project input x-vectors into a latent space has shown promising performance on meeting data. In this paper, we extend the ClusterGAN network to improve diarization robustness and enable rapid generalization across various challenging domains. To this end, we fetch the pre-trained encoder from the ClusterGAN and fine tune it by using prototypical loss (meta-ClusterGAN or MCGAN) under the meta-learning paradigm. Experiments are conducted on CALLHOME telephonic conversations, AMI meeting data, DIHARD-II (dev set) which includes challenging multi-domain corpus, and two child-clinician interaction corpora (ADOS, BOSCC) related to the autism spectrum disorder domain. Extensive analyses of the experimental data are done to investigate the effectiveness of the proposed ClusterGAN and MCGAN embeddings over x-vectors. The results show that the proposed embeddings with normalized maximum eigengap spectral clustering (NME-SC) back-end consistently outperform the Kaldi state-of-the-art x-vector diarization system. Finally, we employ embedding fusion with x-vectors to provide further improvement in diarization performance. We achieve a relative diarization error rate (DER) improvement of 6.67% to 53.93% on the aforementioned datasets using the proposed fused embeddings over x-vectors. Besides, the MCGAN embeddings provide better performance in the number of speakers estimation and short speech segment diarization compared to x-vectors and ClusterGAN on telephonic conversations. FAU - Pal, Monisankha AU - Pal M AD - Signal Analysis and Interpretation Laboratory, University of Southern California, Los Angeles, USA. FAU - Kumar, Manoj AU - Kumar M AD - Signal Analysis and Interpretation Laboratory, University of Southern California, Los Angeles, USA. FAU - Peri, Raghuveer AU - Peri R AD - Signal Analysis and Interpretation Laboratory, University of Southern California, Los Angeles, USA. FAU - Park, Tae Jin AU - Park TJ AD - Signal Analysis and Interpretation Laboratory, University of Southern California, Los Angeles, USA. FAU - Kim, So Hyun AU - Kim SH AD - Center for Autism and the Developing Brain, Weill Cornell Medicine, USA. FAU - Lord, Catherine AU - Lord C AD - Semel Institute of Neuroscience and Human Behavior, University of California Los Angeles, USA. FAU - Bishop, Somer AU - Bishop S AD - Department of Psychiatry, University of California, San Francisco, USA. FAU - Narayanan, Shrikanth AU - Narayanan S AD - Signal Analysis and Interpretation Laboratory, University of Southern California, Los Angeles, USA. LA - eng GR - R01 HD093012/HD/NICHD NIH HHS/United States GR - R01 MH114925/MH/NIMH NIH HHS/United States PT - Journal Article DEP - 20210226 PL - United States TA - IEEE/ACM Trans Audio Speech Lang Process JT - IEEE/ACM transactions on audio, speech, and language processing JID - 101646714 PMC - PMC8118028 MID - NIHMS1688622 OTO - NOTNLM OT - ClusterGAN OT - MCGAN OT - NME-SC OT - speaker diarization OT - speaker embeddings OT - x-vector EDAT- 2021/05/18 06:00 MHDA- 2021/05/18 06:01 PMCR- 2022/01/01 CRDT- 2021/05/17 06:18 PHST- 2021/05/17 06:18 [entrez] PHST- 2021/05/18 06:00 [pubmed] PHST- 2021/05/18 06:01 [medline] PHST- 2022/01/01 00:00 [pmc-release] AID - 10.1109/taslp.2021.3061885 [doi] PST - ppublish SO - IEEE/ACM Trans Audio Speech Lang Process. 2021;29:1204-1219. doi: 10.1109/taslp.2021.3061885. Epub 2021 Feb 26.