@inproceedings{li-etal-2026-flexislm,
title = "{F}lexi{SLM}: A Spoken Language Model with Dynamic and Controllable Frame Rates",
author = "Li, Jiaqi and
Wang, Chaoren and
Tian, Xiaohai and
Chen, Mingjie and
Liang, Xinyu and
Li, Xu and
Lin, Yufan and
Qiu, Junwen and
Zhang, Jun and
Lu, Lu and
Wu, Zhizheng",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.497/",
pages = "11417--11440",
ISBN = "979-8-89176-459-0",
abstract = "Spoken language models (SLMs) extend LLMs to speech input and output, but existing systems use fixed frame rates (e.g., 25 or 12.5 Hz), overlooking speech{'}s time-varying information density and limiting inference-time quality{--}speed tradeoffs. Recent dynamic-frame-rate audio tokenizers enable very low average frame rates and controllability, yet had not been applied to SLMs. We introduce FlexiSLM, the first SLM with dynamic, controllable frame rates, using pretrained FlexiCodec for dynamic speech output tokens. It integrates this representation into a multi-task speech-to-speech SLM, extends it with input-side frame compression, and adds direct frame-rate conditioning for accurate control during inference. FlexiSLM outperforms fixed-frame-rate 7B models, including Qwen2.5-Omni and Kimi-Audio, at 12.5 and 6.25 Hz; it can be steered down to 4.0 Hz, and at 6.25 Hz roughly halves inference time relative to 12.5 Hz while retaining strong speech-to-speech quality. Audio samples: https://flexislm.github.io; code and data: https://github.com/AmphionTeam/FlexiSLM."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="li-etal-2026-flexislm">
<titleInfo>
<title>FlexiSLM: A Spoken Language Model with Dynamic and Controllable Frame Rates</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jiaqi</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chaoren</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaohai</namePart>
<namePart type="family">Tian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mingjie</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinyu</namePart>
<namePart type="family">Liang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xu</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yufan</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junwen</namePart>
<namePart type="family">Qiu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhizheng</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Spoken language models (SLMs) extend LLMs to speech input and output, but existing systems use fixed frame rates (e.g., 25 or 12.5 Hz), overlooking speech’s time-varying information density and limiting inference-time quality–speed tradeoffs. Recent dynamic-frame-rate audio tokenizers enable very low average frame rates and controllability, yet had not been applied to SLMs. We introduce FlexiSLM, the first SLM with dynamic, controllable frame rates, using pretrained FlexiCodec for dynamic speech output tokens. It integrates this representation into a multi-task speech-to-speech SLM, extends it with input-side frame compression, and adds direct frame-rate conditioning for accurate control during inference. FlexiSLM outperforms fixed-frame-rate 7B models, including Qwen2.5-Omni and Kimi-Audio, at 12.5 and 6.25 Hz; it can be steered down to 4.0 Hz, and at 6.25 Hz roughly halves inference time relative to 12.5 Hz while retaining strong speech-to-speech quality. Audio samples: https://flexislm.github.io; code and data: https://github.com/AmphionTeam/FlexiSLM.</abstract>
<identifier type="citekey">li-etal-2026-flexislm</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.497/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>11417</start>
<end>11440</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T FlexiSLM: A Spoken Language Model with Dynamic and Controllable Frame Rates
%A Li, Jiaqi
%A Wang, Chaoren
%A Tian, Xiaohai
%A Chen, Mingjie
%A Liang, Xinyu
%A Li, Xu
%A Lin, Yufan
%A Qiu, Junwen
%A Zhang, Jun
%A Lu, Lu
%A Wu, Zhizheng
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F li-etal-2026-flexislm
%X Spoken language models (SLMs) extend LLMs to speech input and output, but existing systems use fixed frame rates (e.g., 25 or 12.5 Hz), overlooking speech’s time-varying information density and limiting inference-time quality–speed tradeoffs. Recent dynamic-frame-rate audio tokenizers enable very low average frame rates and controllability, yet had not been applied to SLMs. We introduce FlexiSLM, the first SLM with dynamic, controllable frame rates, using pretrained FlexiCodec for dynamic speech output tokens. It integrates this representation into a multi-task speech-to-speech SLM, extends it with input-side frame compression, and adds direct frame-rate conditioning for accurate control during inference. FlexiSLM outperforms fixed-frame-rate 7B models, including Qwen2.5-Omni and Kimi-Audio, at 12.5 and 6.25 Hz; it can be steered down to 4.0 Hz, and at 6.25 Hz roughly halves inference time relative to 12.5 Hz while retaining strong speech-to-speech quality. Audio samples: https://flexislm.github.io; code and data: https://github.com/AmphionTeam/FlexiSLM.
%U https://aclanthology.org/2026.emnlp-main.497/
%P 11417-11440
Markdown (Informal)
[FlexiSLM: A Spoken Language Model with Dynamic and Controllable Frame Rates](https://aclanthology.org/2026.emnlp-main.497/) (Li et al., EMNLP 2026)
ACL
- Jiaqi Li, Chaoren Wang, Xiaohai Tian, Mingjie Chen, Xinyu Liang, Xu Li, Yufan Lin, Junwen Qiu, Jun Zhang, Lu Lu, and Zhizheng Wu. 2026. FlexiSLM: A Spoken Language Model with Dynamic and Controllable Frame Rates. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 11417–11440, Budapest, Hungary. Association for Computational Linguistics.