Feasibility of Using Automatic Speech Recognition with Voices of Deaf and Hard-of-Hearing Individuals

Glasser, Abraham; Kushalnagar, Kesavan; Kushalnagar, Raja

doi:10.1145/3132525.3134819

Cited by 22 publications

(8 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, our study suggests that the roots of usability and accessibility challenges for other populations--like older adults, children, people with Alzheimer's, and people with intellectual disabilities--may be traced back to VAPA design guidelines. Prior work finds that speech recognition often breaks down for children and older adults [3,21] as well as people who are deaf or hard-of-hearing [16], because there is wide variation in pitch, pronunciation, patterns of stress, and intersyllabic pauses, leading to higher recognition error rates [16,21,35]. Prior work also finds that the timeout period for speech input is often inadequate for people with Alzheimer's [34] and intellectual disabilities [3].…”

Section: Putting Disability Back Into the Ideal Humanmentioning

confidence: 99%

Reading Between the Guidelines

Branham

Roy

2019

The 21st International ACM SIGACCESS Conference on Computers and Accessibility

View full text Add to dashboard Cite

Voice-Activated Personal Assistants (VAPAs)-like Apple Siri and Amazon Alexa-have rapidly become common features on mobile devices and in homes of millions of people around the world. They have proven to be particularly valuable to people with disabilities, chiefly among people with visual impairments. Yet, we still know relatively little about the fundamental metaphors and guidelines for designing voice assistants, and how they might empower and constrain visually impaired users. To address this need, we conducted a qualitative document review of VAPA design guidelines published by top commercial vendors Amazon, Google, Microsoft, Apple and Alibaba. We found that guidelines have many commonalities that surface an underlying assumption that VAPA interfaces should be modeled after human-human conversation. We draw on prior work about needs of people with visual impairments to critique this taken-for-granted human-human conversation metaphor and offer amendments to prevailing design guidelines that can make this now-pervasive platform more fully achieve its potential to become universally usable.

show abstract

Section: Putting Disability Back Into the Ideal Humanmentioning

confidence: 99%

Reading Between the Guidelines

Branham

Roy

2019

The 21st International ACM SIGACCESS Conference on Computers and Accessibility

View full text Add to dashboard Cite

show abstract

“…In [6], 45 audio files were chosen by a naive listener. 15 samples were rated "good", 16 samples were "fine" and 14 samples were "bad".…”

Section: Methodology Audio Datasetmentioning

confidence: 99%

“…Glasser, Kushalnagar, and Kushalnagar did a preliminary study on using Deaf and Hard-of-Hearing (DHH) speech [6]. However, there have been significant advances in ASR since then.…”

Section: Introduction and Previous Workmentioning

confidence: 99%

See 1 more Smart Citation

Automatic Speech Recognition Services

Glasser¹

2019

Extended Abstracts of the 2019 CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

Nowadays, speech is becoming a more common, if not standard, interface to technology. This can be seen in the trend of technology changes over the years. Increasingly, voice is used to control programs, appliances and personal devices within homes, cars, workplaces, and public spaces through smartphones and home assistant devices using Amazon's Alexa, Google's Assistant and Apple's Siri, and other proliferating technologies. However, most speech interfaces are not accessible for Deaf and Hard-of-Hearing (DHH) people. In this paper, performances of current Automatic Speech Recognition (ASR) with voices of DHH speakers are evaluated. ASR has improved over the years, and is able to reach Word Error Rates (WER) as low as 5-6% [1][2] [3], with the help of cloud-computing and machine learning algorithms that take in custom vocabulary models. In this paper, a custom vocabulary model is used, and the significance of the improvement is evaluated when using DHH speech.

show abstract

“…ASR models trained on a large training dataset have achieved excellent results and are used in personal phones, IoT (Internet of Things) devices, and cloud services, examples of which include Alexa, Siri, and Bixby. However, non-standard speech, such as amyotrophic lateral sclerosis (ALS) speech, Parkinson's speech, and cochlear implant (CI) patients' speech, has a low recognition rate, whereas ASR models are trained using standard speech data sets [4,5]. Therefore, people with non-standard speech cannot use ASR models trained with a standard speech dataset.…”

Section: Introductionmentioning

confidence: 99%

An Effective Learning Method for Automatic Speech Recognition in Korean CI Patients’ Speech

et al. 2021

View full text Add to dashboard Cite

The automatic speech recognition (ASR) model usually requires a large amount of training data to provide better results compared with the ASR models trained with a small amount of training data. It is difficult to apply the ASR model to non-standard speech such as that of cochlear implant (CI) patients, owing to privacy concerns or difficulty of access. In this paper, an effective finetuning and augmentation ASR model is proposed. Experiments compare the character error rate (CER) after training the ASR model with the basic and the proposed method. The proposed method achieved a CER of 36.03% on the CI patient’s speech test dataset using only 2 h and 30 min of training data, which is a 62% improvement over the basic method.

show abstract

Feasibility of Using Automatic Speech Recognition with Voices of Deaf and Hard-of-Hearing Individuals

Cited by 22 publications

References 3 publications

Reading Between the Guidelines

Reading Between the Guidelines

Automatic Speech Recognition Services

An Effective Learning Method for Automatic Speech Recognition in Korean CI Patients’ Speech

Contact Info

Product

Resources

About