Mind the Shift: Using Delta SSL Embeddings to Enhance Child ASR
Saved in:
| Title: | Mind the Shift: Using Delta SSL Embeddings to Enhance Child ASR |
|---|---|
| Language: | English |
| Authors: | Zilai Wang, Natarajan Balaji Shankar, Kaiyuan Zhang, Zihan Wang, Abeer Alwan |
| Source: | Grantee Submission. 2026Paper presented at the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2026). |
| Peer Reviewed: | Y |
| Page Count: | 5 |
| Publication Date: | 2026 |
| Sponsoring Agency: | National Science Foundation (NSF) Institute of Education Sciences (ED) |
| Contract Number: | R305C240046 |
| Document Type: | Speeches/Meeting Papers Reports - Research |
| Descriptors: | Independent Study, Speech Communication, Speech Evaluation, Word Recognition, Children |
| DOI: | 10.1109/icassp55912.2026.11462835 |
| Abstract: | Self-supervised learning (SSL) models have achieved impressive results across many speech tasks, yet child automatic speech recognition (ASR) remains challenging due to limited data and pretraining domain mismatch. Fine-tuning SSL models on child speech induces shifts in the representation space. We hypothesize that delta SSL embeddings, the differences between embeddings from a fine-tuned model and those from its pre-trained counterpart, encode task-specific information that complements fine-tuned features from another SSL model. We evaluate multiple fusion strategies on the MyST children's corpus with different models. Results show that delta embedding fusion with WavLM yields up to a 10% relative WER reduction for HuBERT and a 4.4% reduction for W2V2, compared to fine-tuned embedding fusion. Notably, fusing WavLM with delta W2V2 embeddings achieves a WER of 9.64, setting a new state-of-the-art among SSL models on the MyST corpus. These findings demonstrate the effectiveness of delta embeddings and highlight feature fusion as a promising direction for advancing child ASR. [This paper was published in: IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) Proceedings, 2026, pp. 18417-18421.] |
| Abstractor: | As Provided |
| IES Funded: | Yes |
| Entry Date: | 20260101 |
| Accession Number: | ED681686 |
| Database: | ERIC |
| Abstract: | Self-supervised learning (SSL) models have achieved impressive results across many speech tasks, yet child automatic speech recognition (ASR) remains challenging due to limited data and pretraining domain mismatch. Fine-tuning SSL models on child speech induces shifts in the representation space. We hypothesize that delta SSL embeddings, the differences between embeddings from a fine-tuned model and those from its pre-trained counterpart, encode task-specific information that complements fine-tuned features from another SSL model. We evaluate multiple fusion strategies on the MyST children's corpus with different models. Results show that delta embedding fusion with WavLM yields up to a 10% relative WER reduction for HuBERT and a 4.4% reduction for W2V2, compared to fine-tuned embedding fusion. Notably, fusing WavLM with delta W2V2 embeddings achieves a WER of 9.64, setting a new state-of-the-art among SSL models on the MyST corpus. These findings demonstrate the effectiveness of delta embeddings and highlight feature fusion as a promising direction for advancing child ASR. [This paper was published in: IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) Proceedings, 2026, pp. 18417-18421.] |
|---|---|
| DOI: | 10.1109/icassp55912.2026.11462835 |
