Medical Speech Recognition

[spiral warren]

I am building a medical speech recognition system. I have experience with NLP but I have no experience whatsoever with speech. Might anyone here guide me on how can I start and if there are some resoureces that would be great too? TIA

[timber needle]

Do you need an ASR (automatic speech recognition) model or is there something more specific you're looking for?

[eternal apex]

Gold standard cloud resource for medical transcription is currently Amazon medical transcribe: https://aws.amazon.com/transcribe/medical/

Other than that, a good starting point would be trying to use Whisper or fine tuning it if you have extra data.

[spiral warren]

I am looking for datasets/papers/models specific to medical speech recognition.

[spiral warren]

I can not use commercial tools. For fine-tuning whisper, I need medical speech recognition data. I found some links to public datasets but either they require me to apply for access or were removed.

[eternal apex]

Datasets for medical voice recognition are hard to come by without being a researcher. Here’s something that might be useful though: https://github.com/babylonhealth/primock57

[spiral warren]

Thanks for sharing.