Automated Speech Recognition Systems and Older Adults: A Literature Review and Synthesis

Werner, Lauren; Huang, Gaojian; Pitts, Brandon J.

doi:10.1177/1071181319631121

Cited by 12 publications

(5 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, age-related physical changes may alter speech production and limit the effectiveness of ASR systems for older individuals. Evaluation of several automated speech recognition systems confirmed previous research that has suggested that those systems have more difficulty in recognising the speech of older adults (Werner et al, 2019).…”

Section: Age Bias In Aisupporting

confidence: 74%

“…Recent analysis shows also that age-biased samples and biased tools used for constructing algorithms tend to exclude the habits, interests and values of older people what contributes to strengthening already existing structural ageism (Rosales and Fernández-Ardèvol, 2019 ). Studies of age bias in machine learning are still rare, but persistently show that age bias exists in sentiment analysis models (Díaz et al, 2018 ), face recognition systems using advanced deep learning techniques (Meade et al, 2021 ), in emotion recognition systems (Kim et al, 2021 ), as well as in speech recognition systems (Werner et al, 2019 ).…”

Section: Age Bias In Aimentioning

confidence: 99%

See 1 more Smart Citation

AI revolution in healthcare and medicine and the (re-)emergence of inequalities and disadvantages for ageing population

Stypińska

Franke

2023

Front. Sociol.

View full text Add to dashboard Cite

AI systems in medicine and healthcare are being extensively explored in prevention, diagnosis, novel drug designs and after-care. The application of AI technology in healthcare systems promises impressive outcomes such as equalising healthcare, reducing mortality rate and human error, reducing medical costs, as well as reducing reliance on social services. In the light of the WHO “Decade of Healthy Ageing”, AI applications are designed as digital innovations to support the quality of life for older persons. However, the emergence of evidence of different types of algorithmic bias in AI applications, ageism in the use of digital devices and platforms, as well as age bias in digital data suggests that the use of AI might have discriminatory effects on older population or even cause harm. This paper addresses the issue of age biases and age discrimination in AI applications in medicine and healthcare systems and try to identify main challenges in this area. It will reflect on the potential of AI applications to amplify the already existing health inequalities by discussing two levels where potential negative impact of AI on age inequalities might be observed. Firstly, we will address the technical level of age bias in algorithms and digital datasets (especially health data). Secondly, we will discuss the potential disparate outcomes of automatic decision-making systems (ADMs) used in healthcare on the older population. These examples will demonstrate, although only partially, how AI systems may create new structures of age inequalities and novel dimensions of exclusion in healthcare and medicine.

show abstract

Section: Age Bias In Aisupporting

confidence: 74%

Section: Age Bias In Aimentioning

confidence: 99%

AI revolution in healthcare and medicine and the (re-)emergence of inequalities and disadvantages for ageing population

Stypińska

Franke

2023

Front. Sociol.

View full text Add to dashboard Cite

show abstract

“…Several implications for speech assistants, like longer pauses between words, are mentioned, which should be considered in future systems. Werner et al [34] also indicate, that there might be agerelated differences in the word error rates of ASR systems, which would need to be considered when building ASR models.…”

Section: Resultsmentioning

confidence: 99%

Development of a Privacy-By-Design Speech Assistant Providing Nutrient Information for German Seniors

Seiderer

Ritschel

André

2020

Proceedings of the 6th EAI International Conference on Smart Objects and Technologies for Social Good

View full text Add to dashboard Cite

Elderly people could take benefit of speech assistants since they provide the most natural way to interact with assistive technology. Nevertheless, current speech assistants are mainly based on cloud-systems which are non-functional without a stable internet connection or introduce severe privacy-concerns. In this work we provide an overview of state-of-the-art available components for developing a privacy-by-design, open source speech assistant for seniors, as well as a fully functional implementation. We chose the health related task of getting nutrition information of packaged foodstuff for which we rely on an open data database. Our prototype was used in a workshop with two German seniors to gather further insights into the special requirements of elderly people, which can be taken into account for future speech assistants. CCS CONCEPTS • Human-centered computing → Natural language interfaces; • Applied computing → Consumer health; • Social and professional topics → Seniors.

show abstract

“…This could be related to the age of the persons in this group; since residents are the group with the highest average age, they are more likely to suffer from speech problems. 24 …”

Section: Discussionmentioning

confidence: 99%

The development of an automatic speech recognition model using interview data from long-term care for older adults

Hacking

Verbeek

Hamers

et al. 2022

Journal of the American Medical Informatics Association

View full text Add to dashboard Cite

Objective In long-term care (LTC) for older adults, interviews are used to collect client perspectives that are often recorded and transcribed verbatim, which is a time-consuming, tedious task. Automatic speech recognition (ASR) could provide a solution; however, current ASR systems are not effective for certain demographic groups. This study aims to show how data from specific groups, such as older adults or people with accents, can be used to develop an effective ASR. Materials and methods An initial ASR model was developed using the Mozilla Common Voice dataset. Audio and transcript data (34 h) from interviews with residents, family, and care professionals on quality of care were used. Interview data were continuously processed to reduce the word error rate (WER). Results Due to background noise and mispronunciations, an initial ASR model had a WER of 48.3% on interview data. After finetuning using interview data, the average WER was reduced to 24.3%. When tested on speech data from the interviews, a median WER of 22.1% was achieved, with residents displaying the highest WER (22.7%). The resulting ASR model was at least 6 times faster than manual transcription. Discussion The current method decreased the WER substantially, verifying its efficacy. Moreover, using local transcription of audio can be beneficial to the privacy of participants. Conclusions The current study shows that interview data from LTC for older adults can be effectively used to improve an ASR model. While the model output does still contain some errors, researchers reported that it saved much time during transcription.

show abstract

Automated Speech Recognition Systems and Older Adults: A Literature Review and Synthesis

Cited by 12 publications

References 15 publications

AI revolution in healthcare and medicine and the (re-)emergence of inequalities and disadvantages for ageing population

AI revolution in healthcare and medicine and the (re-)emergence of inequalities and disadvantages for ageing population

Development of a Privacy-By-Design Speech Assistant Providing Nutrient Information for German Seniors

The development of an automatic speech recognition model using interview data from long-term care for older adults

Contact Info

Product

Resources

About