Oddbean new post about | logout
▲ ▼
melvincarvalho | 1 months ago (raw) | root | parent | reply | flag +1 
 "OpenAI just released Whisper Turbo, a lightning-fast speech-to-text model with 809M parameters—8x faster than the large version and needing only 6GB of VRAM. Awni Hannun reports it can transcribe 12 minutes of audio in just 14 seconds on an M2 Ultra. "
▲ ▼
ToxicFemininity | 1 months ago (raw) | root | parent | reply | flag +1 
 So, this is local?
▲ ▼
TheGuySwann | 1 months ago (raw) | root | parent | reply | flag +1 
 Yes. Whisper is completely local run.
▲ ▼
ToxicFemininity | 1 months ago (raw) | root | parent | reply | flag 
 That’s awesome

How do I download and setup?