The NTNU System at the S&I Challenge 2025 SLA Open Track
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912582558810112 |
|---|---|
| author | Lin, Hong-Yun Lo, Tien-Hong Fang, Yu-Hsuan Lin, Jhen-Ke Wang, Chung-Chun Lu, Hao-Chien Chen, Berlin |
| author_facet | Lin, Hong-Yun Lo, Tien-Hong Fang, Yu-Hsuan Lin, Jhen-Ke Wang, Chung-Chun Lu, Hao-Chien Chen, Berlin |
| contents | A recent line of research on spoken language assessment (SLA) employs neural models such as BERT and wav2vec 2.0 (W2V) to evaluate speaking proficiency across linguistic and acoustic modalities. Although both models effectively capture features relevant to oral competence, each exhibits modality-specific limitations. BERT-based methods rely on ASR transcripts, which often fail to capture prosodic and phonetic cues for SLA. In contrast, W2V-based methods excel at modeling acoustic features but lack semantic interpretability. To overcome these limitations, we propose a system that integrates W2V with Phi-4 multimodal large language model (MLLM) through a score fusion strategy. The proposed system achieves a root mean square error (RMSE) of 0.375 on the official test set of the Speak & Improve Challenge 2025, securing second place in the competition. For comparison, the RMSEs of the top-ranked, third-ranked, and official baseline systems are 0.364, 0.384, and 0.444, respectively. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_05121 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | The NTNU System at the S&I Challenge 2025 SLA Open Track Lin, Hong-Yun Lo, Tien-Hong Fang, Yu-Hsuan Lin, Jhen-Ke Wang, Chung-Chun Lu, Hao-Chien Chen, Berlin Computation and Language Sound Audio and Speech Processing A recent line of research on spoken language assessment (SLA) employs neural models such as BERT and wav2vec 2.0 (W2V) to evaluate speaking proficiency across linguistic and acoustic modalities. Although both models effectively capture features relevant to oral competence, each exhibits modality-specific limitations. BERT-based methods rely on ASR transcripts, which often fail to capture prosodic and phonetic cues for SLA. In contrast, W2V-based methods excel at modeling acoustic features but lack semantic interpretability. To overcome these limitations, we propose a system that integrates W2V with Phi-4 multimodal large language model (MLLM) through a score fusion strategy. The proposed system achieves a root mean square error (RMSE) of 0.375 on the official test set of the Speak & Improve Challenge 2025, securing second place in the competition. For comparison, the RMSEs of the top-ranked, third-ranked, and official baseline systems are 0.364, 0.384, and 0.444, respectively. |
| title | The NTNU System at the S&I Challenge 2025 SLA Open Track |
| topic | Computation and Language Sound Audio and Speech Processing |
| url | https://arxiv.org/abs/2506.05121 |