WEBVTT

00:00:00.000 --> 00:00:02.000
Thanks for making the time.

00:00:02.680 --> 00:00:05.680
I want to start with the part you said was
messy.

00:00:06.320 --> 00:00:11.520
Right, so the messy part is that the file
never arrives the way you expect it.

00:00:12.400 --> 00:00:14.153
Someone records on a phone,

00:00:14.233 --> 00:00:18.240
someone else on a laptop, and the levels
are nothing alike.

00:00:19.200 --> 00:00:21.600
We stopped sending the audio anywhere.

00:00:22.400 --> 00:00:25.501
Once the whole thing ran on the machine
that

00:00:25.581 --> 00:00:28.826
recorded it, half the problems were just
gone.

00:00:28.906 --> 00:00:32.224
What surprised me was how much of the time
went

00:00:32.304 --> 00:00:35.187
into the download, not the transcription.

00:00:35.267 --> 00:00:40.400
The model is about 200 megabytes, and
after the first run it is cached.

00:00:41.280 --> 00:00:44.800
On the laptop an hour of audio takes about
three minutes.

00:00:44.880 --> 00:00:49.911
On the desktop with a proper graphics card
well under one

00:00:49.991 --> 00:00:55.022
that was the moment it stopped feeling
like a demo let me

00:00:55.102 --> 00:01:00.402
add a pause here and then continue with
the second paragraph

00:01:00.482 --> 00:01:05.513
about subtitle formats SRT works on every
video platform.

00:01:05.593 --> 00:01:08.712
VTT is what browsers use.

00:01:08.792 --> 00:01:10.393
Play text is for notes.