Organisation Name

Happy Prime, Inc

Industry

Health services

Organisation Website

https://happyprime.io

Country

Canada

Sustainable Development Goals (SDGs)

General Description of the AI tool

AllSpeak is a cutting-edge AI-powered app designed specifically to provide accurate automatic speech recognition (ASR) for atypical speakers—unlocking independence in everyday conversations, healthcare settings, and beyond. Our combination of simple design and state of the art machine learning techniques, allows the user to easy speak to application that then outputs text, allowing the users to more actively participate in their communities.

Github, open data repository

https://speechaccessibilityproject.beckman.illinois.edu/
https://github.com/happy-prime-inc

Relevant Research and Publications

Mohan, A., Chakraborti, M., Eng, K., Kushaeva, N., Prpa, M., Lewis, J., Zhang, T., Geisler, V. and Geisler, C. (2024) 'A powerful and modern AAC composition tool for impaired speakers', in Interspeech 2024, pp. 991-992. doi: 10.21437/Interspeech.2024-588.
Xu, Y., Chakraborti, M., Zhang, T., Eng, K., Mohan, A. and Prpa, M. (2025) 'Your voice is your voice: Supporting Self-expression through Speech Generation and LLMs in Augmented and Alternative Communication', arXiv preprint, arXiv:2503.17479.
M. Hui, J. Zhang and A. Mohan, "Enhancing AAC Software for Dysarthric Speakers in E-Health Settings: An Evaluation Using TORGO," ICC 2025 – IEEE International Conference on Communications, Montreal, QC, Canada, 2025, pp. 3673-3679, doi: 10.1109/ICC52391.2025.11161747.
Zhang, J., Mohan, A. (2025) Towards atypical speech transcription using LLM-based ASR. Proc. Interspeech 2025, 574-578, doi: 10.21437/Interspeech.2025-2214
Alderete, J., Hui, M.K.F., Mohan, A. (2025) Evaluating ASR Robustness to Spontaneous Speech Errors: A Study of WhisperX Using a Speech Error Database. Proc. Interspeech 2025, 1803-1807, doi: 10.21437/Interspeech.2025-2164

The references are our own work which look at iterations of our user interface and modelling efforts to build an atypical speech communication tool.

Needs