scholarly journals Racial disparities in automated speech recognition

2020 ◽  
Vol 117 (14) ◽  
pp. 7684-7689 ◽  
Author(s):  
Allison Koenecke ◽  
Andrew Nam ◽  
Emily Lake ◽  
Joe Nudell ◽  
Minnie Quartey ◽  
...  
Keyword(s):  
Speech Recognition ◽  
Racial Disparities ◽  
Large Scale ◽  
Machine Learning Algorithms ◽  
Structured Interviews ◽  
Black And White ◽  
Closed Captioning ◽  
Vernacular English ◽  
Automated Speech Recognition ◽  
And Gender

Automated speech recognition (ASR) systems, which use sophisticated machine-learning algorithms to convert spoken language to text, have become increasingly widespread, powering popular virtual assistants, facilitating automated closed captioning, and enabling digital dictation platforms for health care. Over the last several years, the quality of these systems has dramatically improved, due both to advances in deep learning and to the collection of large-scale datasets used to train the systems. There is concern, however, that these tools do not work equally well for all subgroups of the population. Here, we examine the ability of five state-of-the-art ASR systems—developed by Amazon, Apple, Google, IBM, and Microsoft—to transcribe structured interviews conducted with 42 white speakers and 73 black speakers. In total, this corpus spans five US cities and consists of 19.8 h of audio matched on the age and gender of the speaker. We found that all five ASR systems exhibited substantial racial disparities, with an average word error rate (WER) of 0.35 for black speakers compared with 0.19 for white speakers. We trace these disparities to the underlying acoustic models used by the ASR systems as the race gap was equally large on a subset of identical phrases spoken by black and white individuals in our corpus. We conclude by proposing strategies—such as using more diverse training datasets that include African American Vernacular English—to reduce these performance differences and ensure speech recognition technology is inclusive.

Barriers and Mitigating Strategies of Entrepreneurial Business Growth: The Role of Entrepreneur Race and Gender

2021 ◽  
Vol 0 (0) ◽  
Author(s):  
Natasha Conley ◽  
Diana Bilimoria
Keyword(s):  
Business Growth ◽  
Structured Interviews ◽  
Female Entrepreneurs ◽  
Race And Gender ◽  
High Performing ◽  
Black And White ◽  
Corporate Diversity ◽  
And Gender ◽  
Barriers To Growth ◽  
Racial Ethnic

Abstract In this study, we investigate the obstacles to growth and the mitigating strategies of high-performing (over $1 million in revenues) entrepreneurial businesses, and how these differ between businesses owned by Black and White entrepreneurs and between female and male entrepreneurs. Data were collected through semi-structured interviews and qualitatively analyzed using grounded theory and thematic analysis methods. Findings reveal that the lack of access to capital is faced by all groups of entrepreneurs, but that Black and female entrepreneurs additionally faced racial discrimination and gender bias obstacles to their business growth. While all entrepreneurs used social capital strategies to mitigate the barriers to growth that their businesses faced, Black and female entrepreneurs additionally employed faith and prayer as well as business engagement in governmental and corporate diversity initiatives as strategies to overcome the obstacles. Implications of the findings for the entrepreneurial business growth of racial/ethnic minority and female-owned firms are discussed.

scholarly journals Speech Recognition Across the Life Span: Longitudinal Changes From Middle-Age to Older Adults

2015 ◽  
Vol 24 (2) ◽  
pp. 84-87 ◽  
Author(s):  
Judy R. Dubno
Keyword(s):  
Speech Recognition ◽  
Word Recognition ◽  
Life Span ◽  
Pure Tone ◽  
Large Scale ◽  
Mixed Model ◽  
Middle Age ◽  
Age Related ◽  
And Gender ◽  
Age Related Hearing Loss

Purpose The purpose of this article is to provide an overview of evidence of age-related declines in speech recognition in middle age to older adulthood; to review contributions of pure-tone thresholds, age, and gender; and to report preliminary results from a longitudinal study. Method Pure-tone thresholds and word recognition in quiet and babble are being measured in a large sample of adults yearly or every 2 to 3 years. Analyses included >16,000 audiograms and speech recognition scores from >1,200 adults whose ages ranged from the 40s to the 90s. A multivariable generalized linear repeated mixed model assessed changes in thresholds and speech recognition over time. Results Word recognition in quiet declined significantly while controlling for threshold increases, and declines appeared to accelerate near ages 65 to 70 years. Scores for men were poorer than those for women even after controlling for gender differences in thresholds, but rates of decline did not differ by gender. Smaller declines in key word recognition in babble were observed, and declines appeared to accelerate near ages 75 to 80 years. Conclusions Additional evidence is needed from large-scale longitudinal cohort studies to determine rates of change of auditory function across the life span. These studies can identify associations with modifiable risk factors and potential mechanisms to reduce, to prevent, or to delay the onset of age-related hearing loss.

Examining black and white racial disparities in emergency department consultations by age and gender

2021 ◽  
Vol 45 ◽  
pp. 65-70
Author(s):  
Dylan Rose Balter ◽  
Amanda Bertram ◽  
C. Matthew Stewart ◽  
Rosalyn W. Stewart
Keyword(s):  
Emergency Department ◽  
Racial Disparities ◽  
Age And Gender ◽  
Black And White ◽  
And Gender

scholarly journals “I don’t Think These Devices are Very Culturally Sensitive.”—Impact of Automated Speech Recognition Errors on African Americans

2021 ◽  
Vol 4 ◽  
Author(s):  
Zion Mengesha ◽  
Courtney Heldreth ◽  
Michal Lahav ◽  
Juliana Sublewski ◽  
Elyse Tuennerman
Keyword(s):  
African American ◽  
Speech Recognition ◽  
African Americans ◽  
Geographic Location ◽  
Detailed Examination ◽  
Lessons Learned ◽  
Psychological Consequences ◽  
Vernacular English ◽  
Automated Speech Recognition ◽  
Experiential Effects

Automated speech recognition (ASR) converts language into text and is used across a variety of applications to assist us in everyday life, from powering virtual assistants, natural language conversations, to enabling dictation services. While recent work suggests that there are racial disparities in the performance of ASR systems for speakers of African American Vernacular English, little is known about the psychological and experiential effects of these failures paper provides a detailed examination of the behavioral and psychological consequences of ASR voice errors and the difficulty African American users have with getting their intents recognized. The results demonstrate that ASR failures have a negative, detrimental impact on African American users. Specifically, African Americans feel othered when using technology powered by ASR—errors surface thoughts about identity, namely about race and geographic location—leaving them feeling that the technology was not made for them. As a result, African Americans accommodate their speech to have better success with the technology. We incorporate the insights and lessons learned from sociolinguistics in our suggestions for linguistically responsive ways to build more inclusive voice systems that consider African American users’ needs, attitudes, and speech patterns. Our findings suggest that the use of a diary study can enable researchers to best understand the experiences and needs of communities who are often misunderstood by ASR. We argue this methodological framework could enable researchers who are concerned with fairness in AI to better capture the needs of all speakers who are traditionally misheard by voice-activated, artificially intelligent (voice-AI) digital systems.

Implementing a Community Policing Model for Work with Juveniles

1987 ◽  
Vol 14 (2) ◽  
pp. 211-245 ◽  
Author(s):  
JOANNE BELKNAP ◽  
MERRY MORASH ◽  
ROBERT TROJANOWICZ
Keyword(s):  
Police Officers ◽  
Community Policing ◽  
Role Identity ◽  
Future Research ◽  
Police Behavior ◽  
Structured Interviews ◽  
Black And White ◽  
Identity Orientation ◽  
And Gender ◽  
And Behavior

Based on extensive, structured interviews with 59 police officers, an effort was made to explain police behavior and to analyze issues relevant to the conceptualization and implementation of a community policing model. Theories of role identity were used to determine officers' ideal, actual, and behavioral identities in interactions with teenagers, complainants, and supervisors when dealing with a case of “rowdy teens.” A purposive sample was drawn representing foot and motor patrol, as well as race (black and white) and gender. The findings support the implementation of the community police model as was intended by the staff, and suggests the validity of the foot patrol concept. The results did not find gender, race, education, and years on the force predictive of police identity orientation or behavior. However, differences between foot and motor patrol in terms of role identity orientation and behavior were found. The research also suggests the importance of role identity orientation as a useful concept in explaining police behavior, and the discussion pinpoints areas and methods that would be fruitful in future research.

Cross-border Migration and Gender Boundaries in Central Eastern Europe – Female Perspectives

2020 ◽  
Vol 17 (4) ◽  
pp. 499-509
Author(s):  
Ágnes Erőss ◽  
Monika Mária Váradi ◽  
Doris Wastl-Walter
Keyword(s):  
Gender Roles ◽  
Eastern Europe ◽  
Labour Migration ◽  
Structured Interviews ◽  
Livelihood Strategy ◽  
Central Eastern Europe ◽  
Cross Border ◽  
Dual Earner ◽  
And Gender ◽  
Gender Boundaries

In post-Socialist countries, cross-border labour migration has become a common individual and family livelihood strategy. The paper is based on the analysis of semi-structured interviews conducted with two ethnic Hungarian women whose lives have been significantly reshaped by cross-border migration. Focusing on the interplay of gender and cross-border migration, our aim is to reveal how gender roles and boundaries are reinforced and repositioned by labour migration in the post-socialist context where both the socialist dual-earner model and conventional ideas of family and gender roles simultaneously prevail. We found that cross-border migration challenged these women to pursue diverse strategies to balance their roles of breadwinner, wife, and mother responsible for reproductive work. Nevertheless, the boundaries between female and male work or status were neither discursively nor in practice transgressed. Thus, the effect of cross-border migration on altering gender boundaries in post-socialist peripheries is limited.

Racial Disparities in Allergic Outcomes Persist to Age 10 Years in Black and White Children

PEDIATRICS ◽  
2020 ◽  
Vol 146 (Supplement 4) ◽  
pp. S330.2-S331
Author(s):  
Timothy Chow ◽  
Jeffrey Chambliss
Keyword(s):  
Racial Disparities ◽  
White Children ◽  
Black And White

Large-Scale Demonstration Experiment of Mental Imagery Ability

1993 ◽  
Vol 76 (3_suppl) ◽  
pp. 1089-1090 ◽  
Author(s):  
F. Richard Ferraro
Keyword(s):  
Gender Differences ◽  
Individual Differences ◽  
Present Article ◽  
Mental Imagery ◽  
Large Scale ◽  
Introductory Psychology Class ◽  
Imagery Ability ◽  
Psychology Class ◽  
Demonstration Experiment ◽  
And Gender

The present article describes a demonstration experiment used in a large introductory psychology class pertaining to mental imagery ability. The experiment is effective in providing a concrete instance of mental imagery as well as an effective discussion regarding individual differences and gender differences in imagery ability.

Sign in / Sign up

Export Citation Format

Share Document