QUADS: Quantized Distillation Framework for Efficient Speech Language Understanding Article Swipe

View

Subrata Biswas , Mohammad Nur Hossain Khan , Bashima Islam ·

YOU? · · 2025 · Open Access · · DOI: https://doi.org/10.21437/interspeech.2025-532

Spoken Language Understanding (SLU) systems must balance performance and efficiency, particularly in resource-constrained environments. Existing methods apply distillation and quantization separately, leading to suboptimal compression as distillation ignores quantization constraints. We propose QUADS, a unified framework that optimizes both through multi-stage training with a pre-tuned model, enhancing adaptability to low-bit regimes while maintaining accuracy. QUADS achieves 71.13\% accuracy on SLURP and 99.20\% on FSC, with only minor degradations of up to 5.56\% compared to state-of-the-art models. Additionally, it reduces computational complexity by 60--73$\times$ (GMACs) and model size by 83--700$\times$, demonstrating strong robustness under extreme quantization. These results establish QUADS as a highly efficient solution for real-world, resource-constrained SLU applications.

Related Topics

Truth And Reconciliation Commission Of Canada

2025 Nba Draft

28 Years Later

Mahmood Mamdani

Reich Ministry Of Public Enlightenment And Propaganda

Rick Hurst

Concepts

No concepts available.

Metadata

Type: article
Language: en
Landing Page: https://doi.org/10.21437/interspeech.2025-532
OA Status: green
OpenAlex ID: https://openalex.org/W4415320245

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4415320245

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.21437/interspeech.2025-532

Digital Object Identifier
Title: QUADS: Quantized Distillation Framework for Efficient Speech Language Understanding

Work title
Type: article

OpenAlex work type
Language: en

Primary language
Publication year: 2025

Year of publication
Publication date: 2025-08-17

Full publication date if available
Authors: Subrata Biswas, Mohammad Nur Hossain Khan, Bashima Islam

List of authors in order
Landing page: https://doi.org/10.21437/interspeech.2025-532

Publisher landing page
Open access: Yes

Whether a free full text is available
OA status: green

Open access status per OpenAlex
OA URL: https://arxiv.org/pdf/2505.14723

Direct OA link when available
Cited by: 0

Total citation count in OpenAlex

Full payload

id	https://openalex.org/W4415320245
doi	https://doi.org/10.21437/interspeech.2025-532
ids.doi	https://doi.org/10.21437/interspeech.2025-532
ids.openalex	https://openalex.org/W4415320245
fwci	0.0
type	article
title	QUADS: Quantized Distillation Framework for Efficient Speech Language Understanding
biblio.issue
biblio.volume
biblio.last_page	4102
biblio.first_page	4098
topics[0].id	https://openalex.org/T10201
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.9952999949455261
topics[0].domain.id	https://openalex.org/domains/3
topics[0].domain.display_name	Physical Sciences
topics[0].subfield.id	https://openalex.org/subfields/1702
topics[0].subfield.display_name	Artificial Intelligence
topics[0].display_name	Speech Recognition and Synthesis
topics[1].id	https://openalex.org/T11714
topics[1].field.id	https://openalex.org/fields/17
topics[1].field.display_name	Computer Science
topics[1].score	0.991599977016449
topics[1].domain.id	https://openalex.org/domains/3
topics[1].domain.display_name	Physical Sciences
topics[1].subfield.id	https://openalex.org/subfields/1707
topics[1].subfield.display_name	Computer Vision and Pattern Recognition
topics[1].display_name	Multimodal Machine Learning Applications
topics[2].id	https://openalex.org/T12031
topics[2].field.id	https://openalex.org/fields/17
topics[2].field.display_name	Computer Science
topics[2].score	0.9907000064849854
topics[2].domain.id	https://openalex.org/domains/3
topics[2].domain.display_name	Physical Sciences
topics[2].subfield.id	https://openalex.org/subfields/1702
topics[2].subfield.display_name	Artificial Intelligence
topics[2].display_name	Speech and dialogue systems
is_xpac	False
apc_list
apc_paid
language	en
locations[0].id	doi:10.21437/interspeech.2025-532
locations[0].is_oa	False
locations[0].source
locations[0].license
locations[0].pdf_url
locations[0].version	publishedVersion
locations[0].raw_type	proceedings-article
locations[0].license_id
locations[0].is_accepted	True
locations[0].is_published	True
locations[0].raw_source_name	Interspeech 2025
locations[0].landing_page_url	https://doi.org/10.21437/interspeech.2025-532
locations[1].id	pmh:oai:arXiv.org:2505.14723
locations[1].is_oa	True
locations[1].source.id	https://openalex.org/S4306400194
locations[1].source.issn
locations[1].source.type	repository
locations[1].source.is_oa	True
locations[1].source.issn_l
locations[1].source.is_core	False
locations[1].source.is_in_doaj	False
locations[1].source.display_name	arXiv (Cornell University)
locations[1].source.host_organization	https://openalex.org/I205783295
locations[1].source.host_organization_name	Cornell University
locations[1].source.host_organization_lineage	https://openalex.org/I205783295
locations[1].license
locations[1].pdf_url	https://arxiv.org/pdf/2505.14723
locations[1].version	submittedVersion
locations[1].raw_type	text
locations[1].license_id
locations[1].is_accepted	False
locations[1].is_published	False
locations[1].raw_source_name
locations[1].landing_page_url	http://arxiv.org/abs/2505.14723
indexed_in	arxiv, crossref
authorships[0].author.id	https://openalex.org/A5061646605
authorships[0].author.orcid	https://orcid.org/0000-0001-5959-8666
authorships[0].author.display_name	Subrata Biswas
authorships[0].author_position	first
authorships[0].raw_author_name	Subrata Biswas
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5101384237
authorships[1].author.orcid	https://orcid.org/0000-0002-8312-7001
authorships[1].author.display_name	Mohammad Nur Hossain Khan
authorships[1].author_position	middle
authorships[1].raw_author_name	Mohammad Nur Hossain Khan
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5035909313
authorships[2].author.orcid	https://orcid.org/0000-0002-1917-054X
authorships[2].author.display_name	Bashima Islam
authorships[2].author_position	last
authorships[2].raw_author_name	Islam, Bashima
authorships[2].is_corresponding	False
has_content.pdf	False
has_content.grobid_xml	False
is_paratext	False
open_access.is_oa	True
open_access.oa_url	https://arxiv.org/pdf/2505.14723
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-19T00:00:00
display_name	QUADS: Quantized Distillation Framework for Efficient Speech Language Understanding
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T03:46:38.306776
primary_topic.id	https://openalex.org/T10201
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.9952999949455261
primary_topic.domain.id	https://openalex.org/domains/3
primary_topic.domain.display_name	Physical Sciences
primary_topic.subfield.id	https://openalex.org/subfields/1702
primary_topic.subfield.display_name	Artificial Intelligence
primary_topic.display_name	Speech Recognition and Synthesis
cited_by_count	0
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2505.14723
best_oa_location.is_oa	True
best_oa_location.source.id	https://openalex.org/S4306400194
best_oa_location.source.issn
best_oa_location.source.type	repository
best_oa_location.source.is_oa	True
best_oa_location.source.issn_l
best_oa_location.source.is_core	False
best_oa_location.source.is_in_doaj	False
best_oa_location.source.display_name	arXiv (Cornell University)
best_oa_location.source.host_organization	https://openalex.org/I205783295
best_oa_location.source.host_organization_name	Cornell University
best_oa_location.source.host_organization_lineage	https://openalex.org/I205783295
best_oa_location.license
best_oa_location.pdf_url	https://arxiv.org/pdf/2505.14723
best_oa_location.version	submittedVersion
best_oa_location.raw_type	text
best_oa_location.license_id
best_oa_location.is_accepted	False
best_oa_location.is_published	False
best_oa_location.raw_source_name
best_oa_location.landing_page_url	http://arxiv.org/abs/2505.14723
primary_location.id	doi:10.21437/interspeech.2025-532
primary_location.is_oa	False
primary_location.source
primary_location.license
primary_location.pdf_url
primary_location.version	publishedVersion
primary_location.raw_type	proceedings-article
primary_location.license_id
primary_location.is_accepted	True
primary_location.is_published	True
primary_location.raw_source_name	Interspeech 2025
primary_location.landing_page_url	https://doi.org/10.21437/interspeech.2025-532
publication_date	2025-08-17
publication_year	2025
referenced_works_count	0
abstract_inverted_index.a	33, 43, 100
abstract_inverted_index.We	30
abstract_inverted_index.as	25, 99
abstract_inverted_index.by	81, 87
abstract_inverted_index.in	11
abstract_inverted_index.it	77
abstract_inverted_index.of	68
abstract_inverted_index.on	58, 62
abstract_inverted_index.to	22, 48, 70, 73
abstract_inverted_index.up	69
abstract_inverted_index.SLU	107
abstract_inverted_index.and	8, 18, 60, 84
abstract_inverted_index.for	104
abstract_inverted_index.FSC,	63
abstract_inverted_index.both	38
abstract_inverted_index.must	5
abstract_inverted_index.only	65
abstract_inverted_index.size	86
abstract_inverted_index.that	36
abstract_inverted_index.with	42, 64
abstract_inverted_index.(SLU)	3
abstract_inverted_index.QUADS	54, 98
abstract_inverted_index.SLURP	59
abstract_inverted_index.These	95
abstract_inverted_index.apply	16
abstract_inverted_index.minor	66
abstract_inverted_index.model	85
abstract_inverted_index.under	92
abstract_inverted_index.while	51
abstract_inverted_index.5.56\%	71
abstract_inverted_index.QUADS,	32
abstract_inverted_index.Spoken	0
abstract_inverted_index.highly	101
abstract_inverted_index.model,	45
abstract_inverted_index.strong	90
abstract_inverted_index.(GMACs)	83
abstract_inverted_index.71.13\%	56
abstract_inverted_index.99.20\%	61
abstract_inverted_index.balance	6
abstract_inverted_index.extreme	93
abstract_inverted_index.ignores	27
abstract_inverted_index.leading	21
abstract_inverted_index.low-bit	49
abstract_inverted_index.methods	15
abstract_inverted_index.models.	75
abstract_inverted_index.propose	31
abstract_inverted_index.reduces	78
abstract_inverted_index.regimes	50
abstract_inverted_index.results	96
abstract_inverted_index.systems	4
abstract_inverted_index.through	39
abstract_inverted_index.unified	34
abstract_inverted_index.Existing	14
abstract_inverted_index.Language	1
abstract_inverted_index.accuracy	57
abstract_inverted_index.achieves	55
abstract_inverted_index.compared	72
abstract_inverted_index.solution	103
abstract_inverted_index.training	41
abstract_inverted_index.accuracy.	53
abstract_inverted_index.efficient	102
abstract_inverted_index.enhancing	46
abstract_inverted_index.establish	97
abstract_inverted_index.framework	35
abstract_inverted_index.optimizes	37
abstract_inverted_index.pre-tuned	44
abstract_inverted_index.complexity	80
abstract_inverted_index.robustness	91
abstract_inverted_index.suboptimal	23
abstract_inverted_index.compression	24
abstract_inverted_index.efficiency,	9
abstract_inverted_index.maintaining	52
abstract_inverted_index.multi-stage	40
abstract_inverted_index.performance	7
abstract_inverted_index.real-world,	105
abstract_inverted_index.separately,	20
abstract_inverted_index.adaptability	47
abstract_inverted_index.constraints.	29
abstract_inverted_index.degradations	67
abstract_inverted_index.distillation	17, 26
abstract_inverted_index.particularly	10
abstract_inverted_index.quantization	19, 28
abstract_inverted_index.Additionally,	76
abstract_inverted_index.Understanding	2
abstract_inverted_index.applications.	108
abstract_inverted_index.computational	79
abstract_inverted_index.demonstrating	89
abstract_inverted_index.environments.	13
abstract_inverted_index.quantization.	94
abstract_inverted_index.60--73$\times$	82
abstract_inverted_index.83--700$\times$,	88
abstract_inverted_index.state-of-the-art	74
abstract_inverted_index.resource-constrained	12, 106
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	3
citation_normalized_percentile.value	0.22607319
citation_normalized_percentile.is_in_top_1_percent	False
citation_normalized_percentile.is_in_top_10_percent	True