Emphasized Non-Target Speaker Knowledge in Knowledge Distillation for Automatic Speaker Verification
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929400872697856 |
|---|---|
| author | Truong, Duc-Tuan Tao, Ruijie Yip, Jia Qi Lee, Kong Aik Chng, Eng Siong |
| author_facet | Truong, Duc-Tuan Tao, Ruijie Yip, Jia Qi Lee, Kong Aik Chng, Eng Siong |
| contents | Knowledge distillation (KD) is used to enhance automatic speaker verification performance by ensuring consistency between large teacher networks and lightweight student networks at the embedding level or label level. However, the conventional label-level KD overlooks the significant knowledge from non-target speakers, particularly their classification probabilities, which can be crucial for automatic speaker verification. In this paper, we first demonstrate that leveraging a larger number of training non-target speakers improves the performance of automatic speaker verification models. Inspired by this finding about the importance of non-target speakers' knowledge, we modified the conventional label-level KD by disentangling and emphasizing the classification probabilities of non-target speakers during knowledge distillation. The proposed method is applied to three different student model architectures and achieves an average of 13.67% improvement in EER on the VoxCeleb dataset compared to embedding-level and conventional label-level KD methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2309_14838 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Emphasized Non-Target Speaker Knowledge in Knowledge Distillation for Automatic Speaker Verification Truong, Duc-Tuan Tao, Ruijie Yip, Jia Qi Lee, Kong Aik Chng, Eng Siong Sound Audio and Speech Processing Knowledge distillation (KD) is used to enhance automatic speaker verification performance by ensuring consistency between large teacher networks and lightweight student networks at the embedding level or label level. However, the conventional label-level KD overlooks the significant knowledge from non-target speakers, particularly their classification probabilities, which can be crucial for automatic speaker verification. In this paper, we first demonstrate that leveraging a larger number of training non-target speakers improves the performance of automatic speaker verification models. Inspired by this finding about the importance of non-target speakers' knowledge, we modified the conventional label-level KD by disentangling and emphasizing the classification probabilities of non-target speakers during knowledge distillation. The proposed method is applied to three different student model architectures and achieves an average of 13.67% improvement in EER on the VoxCeleb dataset compared to embedding-level and conventional label-level KD methods. |
| title | Emphasized Non-Target Speaker Knowledge in Knowledge Distillation for Automatic Speaker Verification |
| topic | Sound Audio and Speech Processing |
| url | https://arxiv.org/abs/2309.14838 |