<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="cho-etal-2026-stride">
    <titleInfo>
        <title>Stride-k Subsampling: Train-Free Audio Token Reduction for Whisper</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Chanhee</namePart>
        <namePart type="family">Cho</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Junhyuk</namePart>
        <namePart type="family">Choi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Bugeun</namePart>
        <namePart type="family">Kim</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>Whisper exposes speech through a fixed 1500-token encoder interface, now a default representation for ASR decoders and Whisper-based speech language models (SpeechLMs), yet its redundancy remains largely unexamined. We propose stride-k subsampling, a deterministic indexing operation that retains every k-th token after the convolutional stem or encoder transformer. Across five Whisper scales, k=2 preserves baseline WER at both positions, with CKA attributing this stability to acoustic overlap at the stem and attention-induced redistribution at the encoder output. Applying stride-2 at both positions cuts audio tokens by 75% and total GFLOPs by 52–58%, with small WER costs on most ASR benchmarks and larger costs on harder ones. The same configuration extends to three Whisper-based SpeechLMs, yielding modest accuracy drops on stronger baselines and larger drops on weaker ones, while reducing end-to-end latency by 19.6–27.4%. Requiring no training or auxiliary computation, stride-k subsampling suggests that the standard Whisper audio-token interface is over-provisioned for many downstream uses.</abstract>
    <identifier type="citekey">cho-etal-2026-stride</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.1608/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>36572</start>
            <end>36599</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Stride-k Subsampling: Train-Free Audio Token Reduction for Whisper
%A Cho, Chanhee
%A Choi, Junhyuk
%A Kim, Bugeun
%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 cho-etal-2026-stride
%X Whisper exposes speech through a fixed 1500-token encoder interface, now a default representation for ASR decoders and Whisper-based speech language models (SpeechLMs), yet its redundancy remains largely unexamined. We propose stride-k subsampling, a deterministic indexing operation that retains every k-th token after the convolutional stem or encoder transformer. Across five Whisper scales, k=2 preserves baseline WER at both positions, with CKA attributing this stability to acoustic overlap at the stem and attention-induced redistribution at the encoder output. Applying stride-2 at both positions cuts audio tokens by 75% and total GFLOPs by 52–58%, with small WER costs on most ASR benchmarks and larger costs on harder ones. The same configuration extends to three Whisper-based SpeechLMs, yielding modest accuracy drops on stronger baselines and larger drops on weaker ones, while reducing end-to-end latency by 19.6–27.4%. Requiring no training or auxiliary computation, stride-k subsampling suggests that the standard Whisper audio-token interface is over-provisioned for many downstream uses.
%U https://aclanthology.org/2026.emnlp-main.1608/
%P 36572-36599

Markdown (Informal)

[Stride-k Subsampling: Train-Free Audio Token Reduction for Whisper](https://aclanthology.org/2026.emnlp-main.1608/) (Cho et al., EMNLP 2026)

• Stride-k Subsampling: Train-Free Audio Token Reduction for Whisper (Cho et al., EMNLP 2026)

ACL

• Chanhee Cho, Junhyuk Choi, and Bugeun Kim. 2026. Stride-k Subsampling: Train-Free Audio Token Reduction for Whisper. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 36572–36599, Budapest, Hungary. Association for Computational Linguistics.