Replace OpenAI Whisper with Whisper Turbo

[lalalune]

npm install whisper-turbo

GPU accelerated Whisper. Should be faster on local than sending a request to OpenAI. We can use the 'tiny' model.

https://github.com/FL33TW00D/whisper-turbo