notice that you too has some captioning software that the Justice all that allows for voice recognition I think is very interesting how far behind police recognition software is recognition software constituency you think...
I contest your statement that voice recognition software is really behind visual... visual data are more dense and have more ways to analyze it. Vocal data is simpler by comparison and in some ways vocal software is more advanced since it does well with much less information to work from.
I tried it the other day with hilarious results. I guess it depends on the clarity of the speaker, but the captions looked like a really bad Japanese translation by someone that didn't even know English.
The irony is: their are entire communities dedicated to the transcribing and translating of video for the purposes of subtitling. Instead of attempting to harness that, Google chose to relegate it to a less capable machine. What happens when our scribes become dumb (in every sense of the word)? Then again, these communities are usually associated with supposed copyright violation and file sharing.
OK, so it looks like this voice recognition business hasn't advanced much since I last checked it out years ago --to quote from your video transcription:
%uh least democrat that we may not agree with or you may find a and few years you may find up the sir words have more populated means and it is still the surgeons are clearer the claims are clear
Such a translation could make a person trying to understand you via such text feel really obtuse.
Older voice recognition software was useful for generating humorous content which could end up creating unexpected good discussions via exchanging emails converted from rather rational speech to curiously warped phrases worthy of a semi-troll. Thanks for mentioning this new feature, as I had not noticed it and a friend could make good use of it, I suspect.
I have a hard time understanding how you think sound is an easy matter compared to video, video is photos after one another, not that hard.
Phonemes and morphemes get disturbed by accents age and all kinds of stuff. To this you add different kinds of pronunciation of same letters..
You have most likely tried listning to a non-native english speaker and not understanding him due to this, and having a program doing something so abstract is just amazing if even slightly accurate
Apparently it was trained on transcribed news reports so similar material is best trained. I've been watching some lectures on literary theory on Yale's channel and the transcription seems perfect.
◄FUCK THE NEW YOUTUBE LAYOUT►
dangletsbanglol35435 1 year ago
◄FUCK THE NEW YOUTUBE LAYOUT►
SignedSealedAndLost 1 year ago
◄FUCK THE NEW YOUTUBE LAYOUT►
smoovmuvs 1 year ago
I pretty much lost it when the software translated your speech into, "the fall of saigon."
spazfox 1 year ago
With feedback, it will evolve.
prhughes0 1 year ago
notice that you too has some captioning software that the Justice all that allows for voice recognition I think is very interesting how far behind police recognition software is recognition software constituency you think...
mnwalke 1 year ago
I contest your statement that voice recognition software is really behind visual... visual data are more dense and have more ways to analyze it. Vocal data is simpler by comparison and in some ways vocal software is more advanced since it does well with much less information to work from.
stefanlittle 1 year ago
Agent Smith: Never send a human to do a machine's job.
matrixcmitech 1 year ago
I laugh at the strange interpretations which the transcribing program thinks people are saying. It's really amazing.
TheSkepticalAtheist 1 year ago
I tried it the other day with hilarious results. I guess it depends on the clarity of the speaker, but the captions looked like a really bad Japanese translation by someone that didn't even know English.
ScottishAtheist 1 year ago
Howziit gawn. It dizny even innerstawn scoattish
kensho123456 1 year ago
aye a ken.
ScottishAtheist 1 year ago
Uhm no very share o' this commentin' system yet but its gid tae hear frae ye.
Lang may yer lum reek on ither foks coal - an' here's tae independence et the next election if ye ken whit's gid for auld Scotia
awra best freen
kensho123456 1 year ago
The audio capture told me you ' saw your grandads boat '
LimpLoser 1 year ago
The irony is: their are entire communities dedicated to the transcribing and translating of video for the purposes of subtitling. Instead of attempting to harness that, Google chose to relegate it to a less capable machine. What happens when our scribes become dumb (in every sense of the word)? Then again, these communities are usually associated with supposed copyright violation and file sharing.
TheOuroborosWyrm 1 year ago
O I C U. U R OK 2 A T.
I C Y U B. I C Y U B.
notonewhit 1 year ago
I often put it on if someone's a bit boring. It can be hilarious.
TWITfromURANUS 1 year ago 2
OK, so it looks like this voice recognition business hasn't advanced much since I last checked it out years ago --to quote from your video transcription:
%uh least democrat that we may not agree with or you may find a and few years you may find up the sir words have more populated means and it is still the surgeons are clearer the claims are clear
Such a translation could make a person trying to understand you via such text feel really obtuse.
gedgetips 1 year ago
Older voice recognition software was useful for generating humorous content which could end up creating unexpected good discussions via exchanging emails converted from rather rational speech to curiously warped phrases worthy of a semi-troll. Thanks for mentioning this new feature, as I had not noticed it and a friend could make good use of it, I suspect.
gedgetips 1 year ago
"It's a it's pretty hard blow" - Actual CC from this video, lol
FistfulofDicks 1 year ago
I have a hard time understanding how you think sound is an easy matter compared to video, video is photos after one another, not that hard.
Phonemes and morphemes get disturbed by accents age and all kinds of stuff. To this you add different kinds of pronunciation of same letters..
You have most likely tried listning to a non-native english speaker and not understanding him due to this, and having a program doing something so abstract is just amazing if even slightly accurate
futuramani 1 year ago
another really good vid....thanks!
kensho123456 1 year ago
Millions of years evolving. the computer isn't doing that bad
swifter358 1 year ago
ROLF it's so funny to watch this captions, they are totally messed up. Very interesting point, prof.
vls174 1 year ago
You have a really soothing voice Prof. Anton -- it's really starting to make me uncomfortable in the pants.
karausu 1 year ago
Pretty impressed actually with this caption and especially with the translation of it in various languages.
I had ignored it until this video.
A few years back I've tried some voice recognition software that was completely useless.
adorianvlad 1 year ago
Yea it is jacked up, they can add (CC). But Google cant change / rename account names.
Backspace was a simple thing they must have forgot.
acternasoul 1 year ago
LOL. I switched it on watching this video. I have to say, I'm impressed. Yes, it fucks up a lot of the words but it gets most of them right.
Orygyn 1 year ago
appearantty police's have some problem with their recognition software
pkingo1 1 year ago
This is by far my favorite channel. Every video is interesting and you're very consistent.
Indeed to you my friend
5*
Pussymcfats 1 year ago 3
Apparently it was trained on transcribed news reports so similar material is best trained. I've been watching some lectures on literary theory on Yale's channel and the transcription seems perfect.
GirlyVoice 1 year ago