Saved in:
Bibliographic Details
Main Authors: Huo, Mingyue, Shao, Yiwen, Zhang, Yuheng
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.06896
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909986826747904
author Huo, Mingyue
Shao, Yiwen
Zhang, Yuheng
author_facet Huo, Mingyue
Shao, Yiwen
Zhang, Yuheng
contents We present TagSpeech, a unified LLM-based framework that utilizes Temporal Anchor Grounding for joint multi-speaker ASR and diarization. The framework is built on two key designs: (1) decoupled semantic and speaker streams fine-tuned via Serialized Output Training (SOT) to learn turn-taking dynamics; and (2) an interleaved time anchor mechanism that not only supports fine-grained timestamp prediction but also acts as a synchronization signal between semantic understanding and speaker tracking. Compared to previous works that primarily focus on speaker-attributed ASR or implicit diarization, TagSpeech addresses the challenge of fine-grained speaker-content alignment and explicitly models "who spoke what and when" in an end-to-end manner. Experiments on AMI and AliMeeting benchmarks demonstrate that our method achieves consistent improvements in Diarization Error Rate (DER) over strong end-to-end baselines, including Qwen-Omni and Gemini, particularly in handling complex speech overlaps. Moreover, TagSpeech employs a parameter-efficient training paradigm in which the LLM backbone is frozen and only lightweight projectors are trained, resulting in strong performance with low computational cost.
format Preprint
id arxiv_https___arxiv_org_abs_2601_06896
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle TagSpeech: End-to-End Multi-Speaker ASR and Diarization with Fine-Grained Temporal Grounding
Huo, Mingyue
Shao, Yiwen
Zhang, Yuheng
Audio and Speech Processing
Computation and Language
We present TagSpeech, a unified LLM-based framework that utilizes Temporal Anchor Grounding for joint multi-speaker ASR and diarization. The framework is built on two key designs: (1) decoupled semantic and speaker streams fine-tuned via Serialized Output Training (SOT) to learn turn-taking dynamics; and (2) an interleaved time anchor mechanism that not only supports fine-grained timestamp prediction but also acts as a synchronization signal between semantic understanding and speaker tracking. Compared to previous works that primarily focus on speaker-attributed ASR or implicit diarization, TagSpeech addresses the challenge of fine-grained speaker-content alignment and explicitly models "who spoke what and when" in an end-to-end manner. Experiments on AMI and AliMeeting benchmarks demonstrate that our method achieves consistent improvements in Diarization Error Rate (DER) over strong end-to-end baselines, including Qwen-Omni and Gemini, particularly in handling complex speech overlaps. Moreover, TagSpeech employs a parameter-efficient training paradigm in which the LLM backbone is frozen and only lightweight projectors are trained, resulting in strong performance with low computational cost.
title TagSpeech: End-to-End Multi-Speaker ASR and Diarization with Fine-Grained Temporal Grounding
topic Audio and Speech Processing
Computation and Language
url https://arxiv.org/abs/2601.06896