Post

Typing is for suckers. Voice ftw. What are you using for dictating? I've been going back an forth between superwhisper and the built in MacOS one Also testing a few others like Tuna

2 reposts29 likes

The transcribing speed seems to be all over the place even with the same models. I wonder if some of them are transcribing it as it's streamed in versus waiting for the audio to finish.

Yeah, or they’re batching/chunking frames, if I had to guess. I’m designing mine to work locally with the lowest latency I’ve ever measured.

let me know how it goes! I'm also curious about streaming from hosted models since I'd only use cents a day. the new xai Speech to text model is very accurate and cheap

Doesn't matter what I use, none of them can work out my British West Midlands accent. I spend more time correcting the text than it would to type in the first place.

my initial ramblings are such junk tho… I'm rarely on point on my first verbal pass. so then I go in and edit it with a keyboard anyway, and I'm like, why did I even spew this fluffy garbage to begin with. i feel like it's for certain types of people?

It must be because I've been dictating my emails for years. I'm a pretty fast typer, but I feel so much faster when I can just spit out what I want.