Qwen vs. Gemma Integration with Whisper: A Comparative Study in Multilingual SpeechLLM Systems
Not on the map. Why?
Abstract (arXiv)
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.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 12 of its 14 references that are in the map. First posted on arXiv 16 Jun 2025.
Builds on
12 of 14 references.
References · 12
- 2023Visual Instruction Tuningcited by 7,869
- 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
- 2022Robust Speech Recognition via Large-Scale Weak Supervisioncited by 4,057
- 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,598
- 2024Qwen2 Technical Reportcited by 2,120
- 2025Gemma 3 Technical Reportcited by 1,824
- 2020GLU Variants Improve Transformercited by 1,804
- 2019SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognitioncited by 1,380
- 2024Phi-4 Technical Reportcited by 711
- 2023LLaMA-Adapter V2: Parameter-Efficient Visual Instruction Modelcited by 602
- 2023AudioPaLM: A Large Language Model That Can Speak and Listencited by 343
- 2024An Embarrassingly Simple Approach for LLM with Strong ASR Capacitycited by 88
Built on by
Citing papers per year: 2025: 1.
1 paper.