@inproceedings{dsouza-etal-2026-aip,
title = "{AIP}-Speech: Auditing Speech {LLM} Robustness to Background Interference",
author = "Dsouza, Aiden and
Mishra, Shreyansh and
Suneja, Yashraj and
Swain, Rohan and
Borkar, Samyak and
Jaishankar, Arjun and
Urade, Tanmay",
editor = "Shekhar, Ravi and
Swain, Monorama and
Mishra, Jagabandhu and
Dhar, Sandipan and
Mouratidis, Haralambos and
Purver, Matthew",
booktitle = "Proceedings of the {IMPACT}-{SPEECH} Workshop: Identifying, Measuring, Preventing, and Assessing Consequences of Bias in Speech {LLM}s",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.impactspeech-1.2/",
pages = "10--19",
ISBN = "979-8-89176-479-8",
abstract = "Speech Large Language Models (Speech LLMs) are increasingly deployed in real-world acoustic environments, yet it remains unclear which measurable properties of background sounds best explain performance degradation. We present AIP-Speech, an inference-only benchmark that evaluates Speech LLM robustness under controlled mixtures of realworld background recordings. Rather than comparing robustness across background categories alone, we characterize each recording using interpretable acoustic descriptors and examine their associations with degradation. Across five open Speech LLMs, we find that simple acoustic descriptors{---}including Zero Crossing Rate, Spectral Centroid, and Spectral Rolloff{---}consistently exhibit strong associations with degradation. We further quantify robustness disparities across accents and evaluate whether inference-time prompting improves resilience under adverse acoustic conditions. AIP-Speech provides a reproducible benchmark and descriptor-annotated background corpus for analyzing Speech LLM robustness."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="dsouza-etal-2026-aip">
<titleInfo>
<title>AIP-Speech: Auditing Speech LLM Robustness to Background Interference</title>
</titleInfo>
<name type="personal">
<namePart type="given">Aiden</namePart>
<namePart type="family">Dsouza</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shreyansh</namePart>
<namePart type="family">Mishra</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yashraj</namePart>
<namePart type="family">Suneja</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rohan</namePart>
<namePart type="family">Swain</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Samyak</namePart>
<namePart type="family">Borkar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Arjun</namePart>
<namePart type="family">Jaishankar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmay</namePart>
<namePart type="family">Urade</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 IMPACT-SPEECH Workshop: Identifying, Measuring, Preventing, and Assessing Consequences of Bias in Speech LLMs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ravi</namePart>
<namePart type="family">Shekhar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Monorama</namePart>
<namePart type="family">Swain</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jagabandhu</namePart>
<namePart type="family">Mishra</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sandipan</namePart>
<namePart type="family">Dhar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haralambos</namePart>
<namePart type="family">Mouratidis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Matthew</namePart>
<namePart type="family">Purver</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">979-8-89176-479-8</identifier>
</relatedItem>
<abstract>Speech Large Language Models (Speech LLMs) are increasingly deployed in real-world acoustic environments, yet it remains unclear which measurable properties of background sounds best explain performance degradation. We present AIP-Speech, an inference-only benchmark that evaluates Speech LLM robustness under controlled mixtures of realworld background recordings. Rather than comparing robustness across background categories alone, we characterize each recording using interpretable acoustic descriptors and examine their associations with degradation. Across five open Speech LLMs, we find that simple acoustic descriptors—including Zero Crossing Rate, Spectral Centroid, and Spectral Rolloff—consistently exhibit strong associations with degradation. We further quantify robustness disparities across accents and evaluate whether inference-time prompting improves resilience under adverse acoustic conditions. AIP-Speech provides a reproducible benchmark and descriptor-annotated background corpus for analyzing Speech LLM robustness.</abstract>
<identifier type="citekey">dsouza-etal-2026-aip</identifier>
<location>
<url>https://aclanthology.org/2026.impactspeech-1.2/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>10</start>
<end>19</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T AIP-Speech: Auditing Speech LLM Robustness to Background Interference
%A Dsouza, Aiden
%A Mishra, Shreyansh
%A Suneja, Yashraj
%A Swain, Rohan
%A Borkar, Samyak
%A Jaishankar, Arjun
%A Urade, Tanmay
%Y Shekhar, Ravi
%Y Swain, Monorama
%Y Mishra, Jagabandhu
%Y Dhar, Sandipan
%Y Mouratidis, Haralambos
%Y Purver, Matthew
%S Proceedings of the IMPACT-SPEECH Workshop: Identifying, Measuring, Preventing, and Assessing Consequences of Bias in Speech LLMs
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-479-8
%F dsouza-etal-2026-aip
%X Speech Large Language Models (Speech LLMs) are increasingly deployed in real-world acoustic environments, yet it remains unclear which measurable properties of background sounds best explain performance degradation. We present AIP-Speech, an inference-only benchmark that evaluates Speech LLM robustness under controlled mixtures of realworld background recordings. Rather than comparing robustness across background categories alone, we characterize each recording using interpretable acoustic descriptors and examine their associations with degradation. Across five open Speech LLMs, we find that simple acoustic descriptors—including Zero Crossing Rate, Spectral Centroid, and Spectral Rolloff—consistently exhibit strong associations with degradation. We further quantify robustness disparities across accents and evaluate whether inference-time prompting improves resilience under adverse acoustic conditions. AIP-Speech provides a reproducible benchmark and descriptor-annotated background corpus for analyzing Speech LLM robustness.
%U https://aclanthology.org/2026.impactspeech-1.2/
%P 10-19
Markdown (Informal)
[AIP-Speech: Auditing Speech LLM Robustness to Background Interference](https://aclanthology.org/2026.impactspeech-1.2/) (Dsouza et al., IMPACT-SPEECH 2026)
ACL
- Aiden Dsouza, Shreyansh Mishra, Yashraj Suneja, Rohan Swain, Samyak Borkar, Arjun Jaishankar, and Tanmay Urade. 2026. AIP-Speech: Auditing Speech LLM Robustness to Background Interference. In Proceedings of the IMPACT-SPEECH Workshop: Identifying, Measuring, Preventing, and Assessing Consequences of Bias in Speech LLMs, pages 10–19, Budapest, Hungary. Association for Computational Linguistics.