# Seewo's Submission to MLC-SLM: Lessons learned from Speech Reasoning Language Models

arXiv: https://arxiv.org/abs/2506.13300 · HTML: https://mlsanity.com/paper/2506.13300/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Audio-language model reasoning, one of 17 directions in LLM reasoning.
- Builds on 24 of its 34 references that are in the map.
- First posted on arXiv 16 Jun 2025.

## Placed in

- LLM reasoning / [Audio-language model reasoning](https://mlsanity.com/lr/audio-language-model-reasoning/)

## Builds on (most cited first, 24 of 24)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://mlsanity.com/paper/2106.07447/) (2021)
- [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://mlsanity.com/paper/2110.13900/) (2021)
- [Understanding R1-Zero-Like Training: A Critical Perspective](https://mlsanity.com/paper/2503.20783/) (2025)
- [Training Language Models to Self-Correct via Reinforcement Learning](https://mlsanity.com/paper/2409.12917/) (2024)
- [MUSAN: A Music, Speech, and Noise Corpus](https://mlsanity.com/paper/1510.08484/) (2015)
- [GLoRe: When, Where, and How to Improve LLM Reasoning via Global and Local Refinements](https://mlsanity.com/paper/2402.10963/) (2024)
- [CAM++: A Fast and Efficient Network for Speaker Verification Using Context-Aware Masking](https://mlsanity.com/paper/2303.00332/) (2023)
- [Wespeaker: A Research and Production oriented Speaker Embedding Learning Toolkit](https://mlsanity.com/paper/2210.17016/) (2022)
- [An Embarrassingly Simple Approach for LLM with Strong ASR Capacity](https://mlsanity.com/paper/2402.08846/) (2024)
- [Unified Language-Vision Pretraining in LLM with Dynamic Discrete Visual Tokenization](https://mlsanity.com/paper/2309.04669/) (2023)
- [An Enhanced Res2Net with Local and Global Feature Fusion for Speaker Verification](https://mlsanity.com/paper/2305.12838/) (2023)
- [Reasoning Language Models: A Blueprint](https://mlsanity.com/paper/2501.11223/) (2025)
- [Instruction Tuning With Loss Over Instructions](https://mlsanity.com/paper/2405.14394/) (2024)
- [Deep-FSMN for Large Vocabulary Continuous Speech Recognition](https://mlsanity.com/paper/1803.05030/) (2018)
- [CN-Celeb: multi-genre speaker recognition](https://mlsanity.com/paper/2012.12468/) (2020)
- [Instruction Fine-Tuning: Does Prompt Loss Matter?](https://mlsanity.com/paper/2401.13586/) (2024)
- [VoxBlink2: A 100K+ Speaker Recognition Corpus and the Open-Set Speaker-Identification Benchmark](https://mlsanity.com/paper/2407.11510/) (2024)
- [3D-Speaker-Toolkit: An Open-Source Toolkit for Multimodal Speaker Verification and Diarization](https://mlsanity.com/paper/2403.19971/) (2024)
- [Babel: Open Multilingual Large Language Models Serving Over 90% of Global Speakers](https://mlsanity.com/paper/2503.00865/) (2025)
- [VoxBlink: A Large Scale Speaker Verification Dataset on Camera](https://mlsanity.com/paper/2308.07056/) (2023)
- [GEC-RAG: Improving Generative Error Correction via Retrieval-Augmented Generation for Automatic Speech Recognition Systems](https://mlsanity.com/paper/2501.10734/) (2025)

## Abstract (arXiv)

This paper presents Seewo's systems for both tracks of the Multilingual Conversational Speech Language Model Challenge (MLC-SLM), addressing automatic speech recognition (ASR) and speaker diarization with ASR (SD-ASR). We introduce a multi-stage training pipeline that explicitly enhances reasoning and self-correction in speech language models for ASR. Our approach combines curriculum learning for progressive capability acquisition, Chain-of-Thought data augmentation to foster intermediate reflection, and Reinforcement Learning with Verifiable Rewards (RLVR) to further refine self-correction through reward-driven optimization. This approach achieves substantial improvements over the official challenge baselines. On the evaluation set, our best system attains a WER/CER of 11.57% for Track 1 and a tcpWER/tcpCER of 17.67% for Track 2. Comprehensive ablation studies demonstrate the effectiveness of each component under challenge constraints.