Paper deep dive

An Omnilingual-ASR-Based Speech-LLM System for the 2nd MLC-SLM Challenge

Shuming Fang, Shuifei Zeng

Year: 2026Venue: arXiv preprintArea: cs.SDType: PreprintEmbeddings: 6

Intelligence

Status: succeeded | Model: Gemma-4-26B-A4B | Prompt: intel-v1 | Confidence: 94%

Last extracted: 7/18/2026, 11:05:08 AM

Summary

This paper presents a submission to the 2nd MLC-SLM Challenge Task 1, detailing a cascased speech processing system that combines DiariZen-Large-s80 segmentation, CAM++ embedding-based clustering, and a LoRA-adapted omniASR LLM 7B v2 recognizer. The system achieves a macro tcpMER of 29.27% on the Development set and 50.23% on the Evaluation set, significantly outperforming the official baseline. The authors analyze engineering choices, noting that embedding-based clustering outperforms end-to-end alternatives and that overlap-aware segmentation negatively impacts performance due to double transcription of overlapped speech.

Entities (10)

MLC-SLM Challenge · event · 98%Shuifei Zeng · person · 98%Shuming Fang · person · 98%DiariZen-Large-s80 · model · 95%INTER SPEECH 2026 · venue · 95%omniASR LLM 7B v2 · model · 95%tcpMER · metric · 95%CAM++ · algorithm · 92%LoRA · technique · 90%WavLM-Large · model · 90%

Relation Signals (7)

Shuifei Zeng → authored → An Omnilingual-ASR-Based Speech-LLM System for the 2nd MLC-SLM Challenge

confidence 99% · Authors: Shuming Fang, Shuifei Zeng

Shuming Fang → authored → An Omnilingual-ASR-Based Speech-LLM System for the 2nd MLC-SLM Challenge

confidence 99% · Authors: Shuming Fang, Shuifei Zeng

An Omnilingual-ASR-Based Speech-LLM System for the 2nd MLC-SLM Challenge → acceptedto → INTER SPEECH 2026

confidence 95% · Accepted to INTERSPEECH 2026

DiariZen-Large-s80 → usedfor → Segmentation

confidence 95% · DiariZen-Large-s80 ... segmentation

CAM++ → usedfor → speaker_clustering

confidence 92% · CAM++ embedding-based two-speaker clustering

omniASR LLM 7B v2 → adaptedwith → LoRA

confidence 90% · LoRA-adapted omniASR LLM 7B v2 recognizer

DiariZen-Large-s80 → usesbackbone → WavLM-Large

confidence 90% · DiariZen-Large-s80 (WavLM-Large) segmentation

Cypher Suggestions (0)

No Cypher suggestions yet.

Abstract

Abstract:We describe our submission to Task 1 of the 2nd MLCSLM Challenge: a cascaded diarization-then-recognition system that combines DiariZen-Large-s80 (WavLM-Large) segmentation, CAM++ embedding-based two-speaker clustering, and a LoRA-adapted omniASR LLM 7B v2 recognizer, with no oracle segmentation or speaker labels at test time. On the official Development set (150 conversations, 21 language/accent categories) the system attains a macro tcpMER of 29.27%, versus 79.15% for the official baseline; on the Evaluation set it scores 50.23%. We also analyze two engineering choices that substantially affect tcpMER. First, embedding-based speaker clustering outperforms an end-to-end-style alternative that assigns speakers from ASR <sc> turn markers alone. Second, overlap-aware segmentation, although intended to raise diarization recall, increases tcpMER because overlapped speech is transcribed twice.

PDF

Open source PDF →

PDF not stored locally. Use the link above to view on the source site.

Full Text

4,215 characters extracted from source content.

Expand or collapse full text

Skip to main content Search Submit Donate Log in Search arXiv Press Enter to search · Advanced search Computer Science > Sound arXiv:2607.12468v1 (cs) [Submitted on 14 Jul 2026] Title:An Omnilingual-ASR-Based Speech-LLM System for the 2nd MLC-SLM Challenge Authors:Shuming Fang, Shuifei Zeng View a PDF of the paper titled An Omnilingual-ASR-Based Speech-LLM System for the 2nd MLC-SLM Challenge, by Shuming Fang and Shuifei Zeng View PDF Abstract:We describe our submission to Task 1 of the 2nd MLCSLM Challenge: a cascaded diarization-then-recognition system that combines DiariZen-Large-s80 (WavLM-Large) segmentation, CAM++ embedding-based two-speaker clustering, and a LoRA-adapted omniASR LLM 7B v2 recognizer, with no oracle segmentation or speaker labels at test time. On the official Development set (150 conversations, 21 language/accent categories) the system attains a macro tcpMER of 29.27%, versus 79.15% for the official baseline; on the Evaluation set it scores 50.23%. We also analyze two engineering choices that substantially affect tcpMER. First, embedding-based speaker clustering outperforms an end-to-end-style alternative that assigns speakers from ASR <sc> turn markers alone. Second, overlap-aware segmentation, although intended to raise diarization recall, increases tcpMER because overlapped speech is transcribed twice. Comments: Accepted to INTERSPEECH 2026. 4 pages + references. Technical description of our 2nd MLC-SLM Challenge Task 1 submission Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI) Cite as: arXiv:2607.12468 [cs.SD] (or arXiv:2607.12468v1 [cs.SD] for this version) https://doi.org/10.48550/arXiv.2607.12468 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Fang Shuming [view email] [v1] Tue, 14 Jul 2026 07:52:39 UTC (29 KB) Full-text links: Access Paper: View a PDF of the paper titled An Omnilingual-ASR-Based Speech-LLM System for the 2nd MLC-SLM Challenge, by Shuming Fang and Shuifei ZengView PDFTeX Source view license Current browse context: cs.SD < prev | next > new | recent | 2026-07 Change to browse by: cs cs.AI References & Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About · Help · Contact · Subscribe · Copyright · Privacy · Accessibility · Operational Status (opens in new tab) Major funding support from

Instant research discovery