CoChat

Making Emergency Calls More Accessible to Older Adults Through a Hands-free Speech Interface in the House

Open access
Vacher, Michel, Aman, Frédéric, Rossato, Solange, Portet, François, Lecouteux, Benjamin · 2019 · ACM Transactions on Accessible Computing

The DOI matches its Crossref record; 74 of the 74 references it cites were found in a registry or an index; 9 of the 46 sources its claims cite were checked for support.

Abstract

Wearable personable emergency response (PER) systems are the mainstream solution for allowing frail and isolated individuals to call for help in an emergency. However, these devices are not well adapted to all users and are often not worn all the time, meaning they are not available when needed. This article presents a Voice User Interface system for emergency-call recognition. The interface is designed to permit hands-free interaction using natural language. Crucially, this allows a call for help to be registered without necessitating physical proximity to the system. The system is based on an ASR engine and is tested on a corpus collected to simulate realistic situations. The corpus contains French speech from 4 older adults and 13 younger people wearing an old-age simulator to hamper their mobility, vision, and hearing. On-line evaluation of the preliminary system showed an emergency-call error rate of 27%. Subsequent off-line experimentation improved the results (call error rate 24%), demonstrating that emergency-call recognition in the home is achievable. Another contribution of this work is the corpus, which is made available for research with the hope that it will facilitate related research and quicker development of robust methods for automatic emergency-call recognition in the home.

Overview

CoChat Summary

TL;DR

This paper presents CirdoX, a hands-free voice user interface using ASR to recognize emergency calls in the home for older adults. In a Living Lab simulation with 4 older adults and 13 younger participants wearing an old-age simulator, online call error rate was 27%, improving to 24% off-line with adapted SGMM models.

Key findings

  • Online preliminary system showed an emergency call error rate of 27%; subsequent off-line experimentation improved this to 24%.
  • With the adapted acoustic model BREF120_SWEET-HOME_G, emergency-call WER fell from 80.46% to 49.32%, and overall CER fell from 32% to 27%; filtering-stage sensitivity was 78.0%, specificity 72.5%, and FAR 6.5%.
  • The off-line SGMM system achieved average emergency-call WER of 32.0% (young: 30%, older: 37%), all-speech WER of 36% (young: 34%, older: 40%), and average CER of 24% (young: 23%, older: 28%).
  • Speech/non-speech discrimination had 95.3% accuracy but low sensitivity (Se = 0.736); 73 of 277 emergency calls (26%) were classified as non-speech and missed. Noisy speech was classified as speech only 43.2% of the time versus 86.3% for clean speech.

Methods

  • CirdoX pipeline: sound activity detection, speech/non-speech discrimination using GMMs, ASR, and emergency-call filtering by aligning ASR output to predefined calls via phoneme Levenshtein distance.
  • Data: Cirdo-set corpus from Living Lab DOMUS; 17 participants (4 older adults and 13 younger people wearing an old-age simulator) enacted risky scenarios (slip, stumble, fall, unable to rise, true-negative). It contained 1,950 audio events: 277 emergency calls, 45 colloquial sentences, and 1,628 non-speech events.
  • Analyses: online experiment used Sphinx3 with BREF120 and BREF120_SWEET-HOME_G adapted via MLLR; off-line experiment used Kaldi SGMM adapted with ESTER/REPERE, SWEET-HOME, and Voix-détresse. Metrics included WER, sensitivity, specificity, FAR, CER, and F-measure.

Limitations

  • The paper states that no in situ spontaneous emergency data from real events was collected; scenarios were constrained, the test corpus was small, and the main limitation was noise produced by a fall when the person speaks while falling. It also notes limited older-adult-specific training data and that performance remains insufficient for real-world use (e.g., recall 57%, CER 27% online/24% off-line, filtering FAR 6.5% too large).
  • Evident from the text: only 4 older adults participated; younger participants simulated aging and their voices were not affected; no real falls or real emergency calls were recorded, so generalization to actual emergency situations is uncertain.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1145/3310132

Cites

74 of 74 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

7 Valid · 5 Not valid · 23 Not checked · 84 Own results and interpretations

45 more claims

Create a free account to see all 119 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 47
  • Probably not a substantial claim. 35

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

4 of 4

Citations

Cites

Show all 74 references

Research works printed but not linked: 7

  1. Frédéric Aman , Véronique Auberge , and Michel Vacher . 2013 a. How affects can perturbe the automatic speech recognition of domotic interactions . In Proceedings of WASSS’13 , Satellite Workshop of Interspeech 2013. ISCA, 1--5. Frédéric Aman, Véronique Auberge, and Michel Vacher. 2013a. How affects can perturbe the automatic speech recognition of domotic interactions. In Proceedings of WASSS’13, Satellite Workshop of Interspeech 2013. ISCA, 1--5.
  2. Michel Duée and Cyril Rebillard . 2006. La dépendance des personnes âgées : une projection en 2040 . In Données sociales—La société française. INSEE , 613--619. Michel Duée and Cyril Rebillard. 2006. La dépendance des personnes âgées : une projection en 2040. In Données sociales—La société française. INSEE, 613--619.
  3. Institut National pour la Prévention et l’Education à la Santé. 2006. Mieux prévenir les chutes chez les personnes âgées {Preventing falls of the elderly people}. La Santé de l’homme 381 ( 2006 ), 22--44. Institut National pour la Prévention et l’Education à la Santé. 2006. Mieux prévenir les chutes chez les personnes âgées {Preventing falls of the elderly people}. La Santé de l’homme 381 (2006), 22--44.
  4. Lori Lamel , Jean-Luc Gauvain , and Maxine Eskénazi . 1991 . BREF, a large vocabulary spoken corpus for French . In Proceedings of the European Conference on Speech Communication and Technology (EUROSPEECH’91) , Vol. 2 . ISCA, 505--508. Lori Lamel, Jean-Luc Gauvain, and Maxine Eskénazi. 1991. BREF, a large vocabulary spoken corpus for French. In Proceedings of the European Conference on Speech Communication and Technology (EUROSPEECH’91), Vol. 2. ISCA, 505--508.
  5. Aurélie Névéol , Cyril Grouin , Jeremy Leixa , Sophie Rosset , and Pierre Zweigenbaum . 2014 . The QUAERO French medical corpus: A ressource for medical entity recognition and normalization . In Proceedings of the 4th Workshop on Building and Evaluating Ressources for Health and Biomedical Text Processing. 24--30 . Aurélie Névéol, Cyril Grouin, Jeremy Leixa, Sophie Rosset, and Pierre Zweigenbaum. 2014. The QUAERO French medical corpus: A ressource for medical entity recognition and normalization. In Proceedings of the 4th Workshop on Building and Evaluating Ressources for Health and Biomedical Text Processing. 24--30.
  6. Daniel Povey , Arnab Ghoshal , Gilles Boulianne , Lukás Burget , Ondrej Glembek , Nagendra Goel , Mirko Hannemann , Petr Motlicek , Yanmin Qian , Peter Schwarz , et al. 2011 b. The Kaldi speech recognition toolkit . In Proceedings of the IEEE 2011 Workshop on Automatic Speech Recognition and Understanding (ASRU’11) . IEEE, 1--4. Daniel Povey, Arnab Ghoshal, Gilles Boulianne, Lukás Burget, Ondrej Glembek, Nagendra Goel, Mirko Hannemann, Petr Motlicek, Yanmin Qian, Peter Schwarz, et al. 2011b. The Kaldi speech recognition toolkit. In Proceedings of the IEEE 2011 Workshop on Automatic Speech Recognition and Understanding (ASRU’11). IEEE, 1--4.
  7. Michel Vacher , Benjamin Lecouteux , and François Portet . 2012 . Recognition of voice commands by multisource ASR and noise cancellation in a smart home environment . In Proceedings of the European Signal Processing Conference (EUSIPCO’12) . EURASIP, 1663--1667. Michel Vacher, Benjamin Lecouteux, and François Portet. 2012. Recognition of voice commands by multisource ASR and noise cancellation in a smart home environment. In Proceedings of the European Signal Processing Conference (EUSIPCO’12). EURASIP, 1663--1667.

Cited by

Metadata

Authors
Vacher, M., Aman, F., Rossato, S., Portet, F., & Lecouteux, B.
Affiliations
Institut polytechnique de Grenoble (FR)
Venue
ACM Transactions on Accessible Computing
Published
2019-06-11
Volume
12
Issue
2
Pages
1-25
ISSN
  • 1936-7228 (ISSN-L)
  • 1936-7236
OpenAlex
W2951952110
Cited by
10
References
63
Citations by year
2026: 1 · 2024: 1 · 2023: 1 · 2022: 3 · 2021: 3 · 2020: 1
FWCI
0.81
Type
journal article
Language
en
Open access
Locations
  • ACM Transactions on Accessible Computing (publishedVersion)
  • HAL (submittedVersion, other-oa)
Funders
National Agency for Research under the project CIRDO ”Un compagnon Intelligent Réagissant au Doigt et à l'Oeil„ (ANR-2010-TECS-012)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Speech Recognition and Synthesis
  • Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
  • Health Sciences › Health Professions › Occupational Therapy › Assistive Technology in Communication and Mobility
Keywords
voice user interface, touchless interaction, older adults, speech recognition, natural language commands, speech corpus, aging simulation, assistive technology

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.