search

bnosac / audio.whisper

Transcribe audio files using the "Whisper" Automatic Speech Recognition model from R
Other
118 stars 13 forks source link

alternatives to provide binaries #54

Open jwijffels opened 8 months ago

jwijffels commented 8 months ago

https://github.com/r-hub/repos

shikokuchuo commented 8 months ago

As this is not a CRAN package, R-releases would seem like a good way to share with the community (see https://github.com/r-releases/help)

R-universe should then build at least the default Mac and Windows binaries, but more importantly allow source install using standard functions.

jwijffels commented 8 months ago

Given that the compilation settings are not working cross-platform and the optimal settings for doing a speedy transcription are not something CRAN allows, and given that if you use the package which was compiled with non-supported compilation flags onnyour hardware you get a boom, it´s pretty hard to distribute the package other then in source variant. Nevertheless, if you have specific ideas, let me know.

shikokuchuo commented 8 months ago

Right, as CRAN doesn't allow those non-portable flags. Sorry I was probably not clear enough before - I am participating in a new initiative called 'R-releases'. It has just emerged from the R Consortium Repositories Working Group and one of its aims is: https://r-releases.github.io/users.html#a-home-for-non-cran-packages I would welcome a contribution of audio.whisper, and hope it might be a way for you to reach a wider audience long term.

jwijffels commented 8 months ago

Sure, here you go https://github.com/r-releases/contributions/pull/138 I'd love to see for which architectures this is building things.

shikokuchuo commented 8 months ago

Thanks! Seems to have built fine on all platforms supported by R-universe. https://r-releases.r-universe.dev/audio.whisper