Skip to content

Instantly share code, notes, and snippets.

@jshbrntt
Last active April 20, 2017 12:17
Show Gist options
  • Save jshbrntt/19d0e3541e40cdd9a7c1c0823b9a44ea to your computer and use it in GitHub Desktop.
Save jshbrntt/19d0e3541e40cdd9a7c1c0823b9a44ea to your computer and use it in GitHub Desktop.
An automatically transcribed discussion on ATP about automatic transcription.
Your transcription for atp217_clips_app.mp3 is now available for download from http://app.speechmatics.com/dashboard
Transcriptions can be downloaded as plain text or in a structured JSON format.
=======================================
SPEAKER: M1
But clips came out today I played with it very very very briefly and it's pretty cool. I like the I don't know what the official term for it is but the transcribe what you're saying as you were saying it feature so you can turn on a like text mode or a little speech bubble. And as you record speech will or your speech will appear on screen. So you know you can hold the record and then as you talk you can make it show the words that you're speaking. I just did that as we were recording as I was saying those words. And it says Honda Accord instead of hold while you record.
SPEAKER: M3
So already going well I mean you know automated transcription is still a very very long way away from being good reliably. Like even you know we you know we can criticize Apple for like not being good at some of these things sometimes but nobody can do this very well yet. You know even like the fancy Google algorithms that do youtube transcription and everything like even they are pretty bad at this so it's going to be a fun little thing finding like you know funny mistranslations and people's video sometimes and otherwise it's a cool idea. It's a cool use of technology but yeah it's going to be pretty imperfect for possibly ever if not a long time.
SPEAKER: M2
It's going to make people enunciate as well it's going to do. I tried that feature as well because I was excited to see if I could do the thing that I talked about when we first talked about clips which is make the words appear over the video with with a cadence with a purpose like say things in a certain way and emphasize certain things to have the words appear with that rhythm and the lag seem to be so big I'm not quite sure why whether it's the you know rendering it on top of the video versus understanding the speech or whatever but the lag was so much worse than for example when you talk to one of you know Google now or how they call it where it where it shows your words in real time or even Siri these days but Google Now is the first one I remember being startled by exactly how fast the words were appearing as I spoke them that there wasn't a delay that it wasn't like. Let me just think about what you just said and make it appear that it seemed to follow my cadence and clips cli
ps lags. I found it difficult even when trying to emphasize in a very exaggerated rhythm to get that rhythm to manifest in the video. So. Oh well the people using real video editors can continue to dominate in The Perfect timing of their word appearances and transitions. But in the meantime Yeah it is a cool app and the effects are really neat and it makes it so easy to do something you don't need any expertise in any kind of video editing or production. You just need to be able to tap things on the screen and hold your finger on a giant red button and you will make cool videos.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment