Performance Evaluation of SLAM-ASR: The Good, the Bad, the Ugly, and the Way Forward

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F26%3A0201439" target="_blank" >RIV/00216305:26230/26:0201439 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.fit.vut.cz/research/group/speech/public/publi/2025/kumar_interspeech2025_co-author_Motlicek.pdf" target="_blank" >https://www.fit.vut.cz/research/group/speech/public/publi/2025/kumar_interspeech2025_co-author_Motlicek.pdf</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/ICASSPW65056.2025.11010998" target="_blank" >10.1109/ICASSPW65056.2025.11010998</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Performance Evaluation of SLAM-ASR: The Good, the Bad, the Ugly, and the Way Forward

Popis výsledku v původním jazyce

Recent research has demonstrated that training a linear connector between speech foundation encoders and large language models (LLMs) enables this architecture to achieve strong ASR capabilities. Despite the impressive results, it remains unclear whether these simple approaches are robust enough across different scenarios and speech conditions, such as domain shifts and speech perturbations. In this paper, we address these questions by conducting various ablation experiments using a recent and widely adopted approach called SLAM-ASR. We present novel empirical findings that offer insights on how to effectively utilize the SLAM-ASR architecture across a wide range of settings. Our main findings indicate that SLAM-ASR exhibits poor performance in cross-domain evaluation settings. Additionally, speech perturbations on in-domain data, such as changes in speech rate or additive noise, can significantly degrade performance. Our findings offer critical insights for fine-tuning and configuring robust LLM-based ASR models, tailored to different data characteristics and computational resources.

Název v anglickém jazyce

Performance Evaluation of SLAM-ASR: The Good, the Bad, the Ugly, and the Way Forward

Popis výsledku anglicky

Recent research has demonstrated that training a linear connector between speech foundation encoders and large language models (LLMs) enables this architecture to achieve strong ASR capabilities. Despite the impressive results, it remains unclear whether these simple approaches are robust enough across different scenarios and speech conditions, such as domain shifts and speech perturbations. In this paper, we address these questions by conducting various ablation experiments using a recent and widely adopted approach called SLAM-ASR. We present novel empirical findings that offer insights on how to effectively utilize the SLAM-ASR architecture across a wide range of settings. Our main findings indicate that SLAM-ASR exhibits poor performance in cross-domain evaluation settings. Additionally, speech perturbations on in-domain data, such as changes in speech rate or additive noise, can significantly degrade performance. Our findings offer critical insights for fine-tuning and configuring robust LLM-based ASR models, tailored to different data characteristics and computational resources.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

R - Projekt Ramcoveho programu EK

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

2025 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING WORKSHOPS, ICASSPW

ISBN

979-8-3315-1932-2

ISSN

—

e-ISSN

—

Počet stran výsledku

5

Strana od-do

1-5

Název nakladatele

IEEE

Místo vydání

Hyderabad, Indická republika

Místo konání akce

Hyderabad, Indická republika

Datum konání akce

6. 4. 2025

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

001547041200001