{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,3]],"date-time":"2026-04-03T06:59:12Z","timestamp":1775199552395,"version":"3.50.1"},"reference-count":76,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,12,6]],"date-time":"2025-12-06T00:00:00Z","timestamp":1764979200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,12,6]],"date-time":"2025-12-06T00:00:00Z","timestamp":1764979200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,12,6]]},"DOI":"10.1109\/asru65441.2025.11434591","type":"proceedings-article","created":{"date-parts":[[2026,4,2]],"date-time":"2026-04-02T19:48:04Z","timestamp":1775159284000},"page":"1-8","source":"Crossref","is-referenced-by-count":0,"title":["Continual Pre-training for Codec-Based Speech LLMs: Balancing Understanding and Generation"],"prefix":"10.1109","author":[{"given":"Jiatong","family":"Shi","sequence":"first","affiliation":[{"name":"CMU"}]},{"given":"Chunlei","family":"Zhang","sequence":"additional","affiliation":[{"name":"Tencent AI Lab"}]},{"given":"Jinchuan","family":"Tian","sequence":"additional","affiliation":[{"name":"CMU"}]},{"given":"Junrui","family":"Ni","sequence":"additional","affiliation":[{"name":"UIUC"}]},{"given":"Hao","family":"Zhang","sequence":"additional","affiliation":[{"name":"Tencent AI Lab"}]},{"given":"Shinji","family":"Watanabe","sequence":"additional","affiliation":[{"name":"CMU"}]},{"given":"Dong","family":"Yu","sequence":"additional","affiliation":[{"name":"Tencent AI Lab"}]}],"member":"263","reference":[{"key":"ref1","article-title":"A survey of large language models","author":"Zhao","year":"2023","journal-title":"arXiv preprint arXiv:2303.18223"},{"key":"ref2","article-title":"GPT-4 technical report","volume-title":"arXiv preprint arXiv:2303.08774","author":"Achiam","year":"2023"},{"key":"ref3","article-title":"The Llama 3 herd of models","author":"Dubey","year":"2024","journal-title":"arXiv preprint arXiv:2407.21783"},{"key":"ref4","article-title":"Qwen2 technical report","volume-title":"arXiv preprint arXiv:2407.10671","author":"Yang","year":"2024"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.593"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/ASRU57964.2023.10389742"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1109\/TASLP.2024.3436618"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP48485.2024.10447112"},{"key":"ref9","article-title":"Neural codec language models are zero-shot text to speech synthesizers","author":"Wang","year":"2023","journal-title":"arXiv preprint arXiv:2301.02111"},{"key":"ref10","article-title":"Moshi: a speech-text foundation model for real-time dialogue","author":"D\u00e9fossez","year":"2024","journal-title":"arXiv preprint arXiv:2410.00037"},{"key":"ref11","article-title":"CLamTTS: Improving neural codec language model for zero-shot text-tospeech","volume-title":"The Twelfth International Conference on Learning Representations","author":"Kim"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.naacl-srw.6"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP49660.2025.10888202"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.21437\/interspeech.2025-1106"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP49660.2025.10889794"},{"key":"ref16","article-title":"UniAudio: Towards universal audio generation with large language models","volume-title":"Forty-first International Conference on Machine Learning","author":"Yang"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/SLT61566.2024.10832289"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v39i24.34761"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.findings-emnlp.1055"},{"key":"ref20","article-title":"AudioPaLM: A large language model that can speak and listen","author":"Rubenstein","year":"2023","journal-title":"arXiv preprint arXiv:2306.12925"},{"key":"ref21","article-title":"SpeechComposer: Unifying multiple speech tasks with prompt composition","author":"Wu","year":"2024"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.235"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP48485.2024.10447929"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2024-330"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.findings-emnlp.1055"},{"key":"ref26","article-title":"Scaling speech-text pre-training with synthetic interleaved data","volume-title":"The Thirteenth International Conference on Learning Representations","author":"Zeng"},{"key":"ref27","article-title":"Mini-omni: Language models can hear, talk while thinking in streaming","author":"Xie","year":"2024","journal-title":"arXiv preprint arXiv:2408.16725"},{"key":"ref28","article-title":"Continual learning for large language models: A survey","author":"Wu","year":"2024"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.521"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP48485.2024.10448257"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.findings-acl.616"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/msp.2024.3444318"},{"key":"ref33","article-title":"VALL-E 2: Neural codec language models are human parity zero-shot text to speech synthesizers","author":"Chen","year":"2024","journal-title":"arXiv preprint arXiv:2406.05370"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.52202\/075280-2066"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.naacl-demo.12"},{"key":"ref36","article-title":"Qwen2. 5-omni technical report","volume-title":"arXiv preprint arXiv:2503.20215","author":"Xu","year":"2025"},{"key":"ref37","article-title":"Kimi-audio technical report","volume-title":"arXiv preprint arXiv:2504.18425","author":"Ding","year":"2025"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.iwslt-1.1"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.iwslt-1.1"},{"key":"ref40","article-title":"Seamless: Multilingual expressive and streaming speech translation","author":"Barrault","year":"2023","journal-title":"arXiv preprint arXiv:2312.05187"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1109\/TASLP.2021.3129994"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.52202\/075280-1214"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.1109\/icsda.2017.8384449"},{"key":"ref44","article-title":"Aishell-2: Transforming mandarin asr research into industrial scale","author":"Du","year":"2018","journal-title":"arXiv preprint arXiv:1808.10583"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2021-755"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP43922.2022.9746682"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2021-1965"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.2015.7178964"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2020-2826"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-99579-3_21"},{"key":"ref51","article-title":"The WSJ speech database","author":"Rottland","year":"1997"},{"key":"ref52","first-page":"4218","article-title":"Common voice: A massively-multilingual speech corpus","volume-title":"Proc. LREC","author":"Ardila"},{"key":"ref53","article-title":"The pile: An 800 gb dataset of diverse text for language modeling","author":"Gao","year":"2020","journal-title":"arXiv preprint arXiv:2101.00027"},{"key":"ref54","article-title":"Megatron-lm: Training multibillion parameter language models using model parallelism","author":"Shoeybi","year":"2019","journal-title":"arXiv preprint arXiv:1909.08053"},{"key":"ref55","article-title":"Investigating the catastrophic forgetting in multimodal large language model fine-tuning","volume-title":"Conference on Parsimony and Learning (Proceedings Track)","author":"Zhai"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v34i05.6428"},{"key":"ref57","first-page":"341","article-title":"Reducing activation recomputation in large transformer models","volume-title":"Proceedings of Machine Learning and Systems","volume":"5","author":"Korthikanti"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2023-2051"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2019-2441"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1109\/icassp49660.2025.10890510"},{"key":"ref61","first-page":"28492","article-title":"Robust speech recognition via largescale weak supervision","volume-title":"International conference on machine learning.","author":"Radford"},{"key":"ref62","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2024-1345"},{"key":"ref63","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2022-439"},{"key":"ref64","doi-asserted-by":"publisher","DOI":"10.1109\/SLT61566.2024.10832365"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2021-2027"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2023-1233"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/W18-6319"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2022-970"},{"key":"ref69","article-title":"Vec-Sok speech: Speech vectorization and tokenization for neural speech generation","author":"Zhu","year":"2023","journal-title":"arXiv preprint arXiv:2310.07246"},{"key":"ref70","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.iwslt-1.7"},{"key":"ref71","article-title":"SALMONN: Towards generic hearing abilities for large language models","volume-title":"The Twelfth International Conference on Learning Representations","author":"Tang"},{"key":"ref72","article-title":"Qwen2-audio technical report","author":"Chu","year":"2024"},{"key":"ref73","doi-asserted-by":"publisher","DOI":"10.52202\/075280-2066"},{"key":"ref74","doi-asserted-by":"publisher","DOI":"10.1109\/ASRU57964.2023.10389676"},{"key":"ref75","article-title":"Speech translation with large language models: An industrial practice","author":"Huang","year":"2023","journal-title":"arXiv preprint arXiv:2312.13585"},{"key":"ref76","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P16-1144"}],"event":{"name":"2025 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)","location":"Honolulu, HI, USA","start":{"date-parts":[[2025,12,6]]},"end":{"date-parts":[[2025,12,10]]}},"container-title":["2025 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11434577\/11433836\/11434591.pdf?arnumber=11434591","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,3]],"date-time":"2026-04-03T04:55:46Z","timestamp":1775192146000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11434591\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,6]]},"references-count":76,"URL":"https:\/\/doi.org\/10.1109\/asru65441.2025.11434591","relation":{},"subject":[],"published":{"date-parts":[[2025,12,6]]}}}