The TEA-ASLP System for Multilingual Conversational Speech Recognition and Speech Diarization in MLC-SLM 2025 Challenge
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913958816907264 |
|---|---|
| author | Xue, Hongfei Huang, Kaixun Zhou, Zhikai Huang, Shen Shang, Shidong |
| author_facet | Xue, Hongfei Huang, Kaixun Zhou, Zhikai Huang, Shen Shang, Shidong |
| contents | This paper presents the TEA-ASLP's system submitted to the MLC-SLM 2025 Challenge, addressing multilingual conversational automatic speech recognition (ASR) in Task I and speech diarization ASR in Task II. For Task I, we enhance Ideal-LLM model by integrating known language identification and a multilingual MOE LoRA structure, along with using CTC-predicted tokens as prompts to improve autoregressive generation. The model is trained on approximately 180k hours of multilingual ASR data. In Task II, we replace the baseline English-Chinese speaker diarization model with a more suitable English-only version. Our approach achieves a 30.8% reduction in word error rate (WER) compared to the baseline speech language model, resulting in a final WER of 9.60% in Task I and a time-constrained minimum-permutation WER of 17.49% in Task II, earning first and second place in the respective challenge tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_18051 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | The TEA-ASLP System for Multilingual Conversational Speech Recognition and Speech Diarization in MLC-SLM 2025 Challenge Xue, Hongfei Huang, Kaixun Zhou, Zhikai Huang, Shen Shang, Shidong Sound Audio and Speech Processing This paper presents the TEA-ASLP's system submitted to the MLC-SLM 2025 Challenge, addressing multilingual conversational automatic speech recognition (ASR) in Task I and speech diarization ASR in Task II. For Task I, we enhance Ideal-LLM model by integrating known language identification and a multilingual MOE LoRA structure, along with using CTC-predicted tokens as prompts to improve autoregressive generation. The model is trained on approximately 180k hours of multilingual ASR data. In Task II, we replace the baseline English-Chinese speaker diarization model with a more suitable English-only version. Our approach achieves a 30.8% reduction in word error rate (WER) compared to the baseline speech language model, resulting in a final WER of 9.60% in Task I and a time-constrained minimum-permutation WER of 17.49% in Task II, earning first and second place in the respective challenge tasks. |
| title | The TEA-ASLP System for Multilingual Conversational Speech Recognition and Speech Diarization in MLC-SLM 2025 Challenge |
| topic | Sound Audio and Speech Processing |
| url | https://arxiv.org/abs/2507.18051 |