Uncovering bias in ASR systems: Evaluating the performance of Wav2vec2 and Whisper for Dutch speakers

Research output: Contribution to conferencePosterAcademic

21 Downloads (Pure)

Abstract

The study evaluated two speech recognition systems, Wav2vec2 and Whisper, for potential biases for Dutch speakers.Results obtained by evaluating on the JASMIN corpus revealed biases against non-native speakers, children, and the elderly,with (slightly) better performance for women. The study emphasizes the need for ASR systems to handle variations in speakingin order to reach equal performance among all users.
Original languageEnglish
Publication statusPublished - 19 Feb 2024
Event2nd Dutch Speech Tech Day : Towards inclusive speech technology in research and applications - Netherlands Institute for Sound & Vision, Hilversum, Netherlands
Duration: 19 Feb 202419 Feb 2024
https://sites.google.com/view/dutchspeechtechday/home

Conference

Conference2nd Dutch Speech Tech Day
Abbreviated titleDutch Speech Tech Day
Country/TerritoryNetherlands
CityHilversum
Period19/02/2419/02/24
Internet address

Fingerprint

Dive into the research topics of 'Uncovering bias in ASR systems: Evaluating the performance of Wav2vec2 and Whisper for Dutch speakers'. Together they form a unique fingerprint.

Cite this