[skykooler]

Interesting. The DeepSpeech github indicates that it's not capable of realtime transcription, even using a GPU. How is that fast enough for this use case?