TL;DR
This work demonstrates Project ResonAIte, an end-to-end ASR pipeline fine-tuned on Marathi deaf speech, and reduces word error rates (WER) from a baseline of 97% to 64% with just one hour of training data.
Speech therapists often face difficulties diagnosing impairments due to the lack of efficient tools for transcribing speech into the International Phonetic Alphabet (IPA). This work addresses this challenge with Broca, a Conformer-based deep learning system pretrained on 8 days of adult speech and fine-tuned on a 165-m...
This paper presents a scalable method that measures consonant contribution using acoustic masking, and relates MMR to two linguistic factors previously reported to correlate with consonant contribution, namely phoneme frequency and functional load.
For people who are deaf or hard-of-hearing (HoH), everyday conversations can be difficult to navigate, even with modern hearing aids. Augmented reality (AR) smart glasses offer a practical way to provide live captions directly in the user’s field of view. However, current models are often too expensive, rely exclusivel...
Abstract In the real world, listeners often find speech in a non-native (L2) accent harder to understand than speech in their own native (L1) accent. In ideal laboratory listening settings, these struggles can become smaller. For this reason, most research on L2 accent processing uses speech-in-noise paradigms to avoid...
This work demonstrates TAPA on the 2016 U.S. presidential debate, and suggests that TAPA can be used to increase access to naturalistic speech data and speed up the processing timeline with experts' supervision.
The present results demonstrate the feasibility of objective, parallel measurement across this hierarchy and point to impaired language processing as a possible mechanism to LiD.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
