Const-me / Whisper

High-performance GPGPU inference of OpenAI's Whisper automatic speech recognition (ASR) model
Mozilla Public License 2.0
7.65k stars 664 forks source link

It's this software abandonware? #214

Open klonte opened 2 months ago

klonte commented 2 months ago

I guess people used this just because AMD GPUs back in the days, but even average CPUs are almost the same speed using other solutions by now.

Yanngo commented 5 hours ago

I wouldn’t say it is abandoned. The program is feature complete.

It could use enhancements like the ability to support newer Whisper models (v3) or the ability to start transcription from a certain timecode of the audio or the video file in order to fix hallucinations. But it works. I am certainly grateful I don't have to use command line interface or compile the code myself.