Qwen vs. Gemma Integration with Whisper: A Comparative Study in Multilingual SpeechLLM Systems
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_ | 1866913929301590016 |
|---|---|
| author | Nguyen, Tuan Hoang, Long-Vu Tran, Huy-Dat |
| author_facet | Nguyen, Tuan Hoang, Long-Vu Tran, Huy-Dat |
| contents | This paper presents our system for the MLC-SLM Challenge 2025, focusing on multilingual speech recognition and language modeling with large language models (LLMs). Our approach combines a fine-tuned Whisper-large-v3 encoder with efficient projector architectures and various decoder configurations. We employ a three-stage training methodology that progressively optimizes the encoder, projector, and LLM components. Our system achieves competitive performance with a private test average WER/CER result of 16.63% using the Gemma3-12B and 18.6% using the Qwen2.5-7B as decoder-only language model. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_13596 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Qwen vs. Gemma Integration with Whisper: A Comparative Study in Multilingual SpeechLLM Systems Nguyen, Tuan Hoang, Long-Vu Tran, Huy-Dat Computation and Language Sound Audio and Speech Processing This paper presents our system for the MLC-SLM Challenge 2025, focusing on multilingual speech recognition and language modeling with large language models (LLMs). Our approach combines a fine-tuned Whisper-large-v3 encoder with efficient projector architectures and various decoder configurations. We employ a three-stage training methodology that progressively optimizes the encoder, projector, and LLM components. Our system achieves competitive performance with a private test average WER/CER result of 16.63% using the Gemma3-12B and 18.6% using the Qwen2.5-7B as decoder-only language model. |
| title | Qwen vs. Gemma Integration with Whisper: A Comparative Study in Multilingual SpeechLLM Systems |
| topic | Computation and Language Sound Audio and Speech Processing |
| url | https://arxiv.org/abs/2506.13596 |