Apple's Dictation System Transcribes the Word 'Racist' as 'Trump' | Hacker News Reader

Apple's Dictation System Transcribes the Word 'Racist' as 'Trump'

nytimes.com

53 pointsksec32 comments

> An Apple spokeswoman blamed the issue on phonetic overlap between the two words, and said the company was working on a fix.

I don't envy the spokeswoman her job, but this is hilariously off base from a linguistics perspective:

* /tɹʌmp/

* /ˈɹeɪ.sɪst/

One word is two syllables, the other is one. One word has a bunch of sibilance and the other has none. One word has close front vowels (including a diphthong!) and the other has a single open back vowel. The only thing they have in common is the /ɹ/, but even that has a slightly different character when in a cluster with /t/.

There's absolutely no way this was an accident of phonetics.

Can audio of someone speaking be reliably turned into something like IPA by a computer though? (Very interested if that exists) My impression is that they're working with fewer elements and a lot more fuzziness about what they are.

I doubt we'll ever get a good writeup of exactly how this actually happens, but I'd really like to.

I guess it depends how much of this is running on the device, and how much is running off the device. If its all running on the device then it might be possible for someone to inspect what exactly is going on. If important parts are running off the device then those parts will be a black box.

You're correct, I'm using the IPA as a proxy for the audio data they actually use, but it's a decent proxy.

I'd have linked spectrograms if I could find them, but in general sibilants come out very heavily in the audio data, so the complete absence of sibilance in "Trump" alone would make it extremely unlikely that the words would get mixed up. The fact that the words are substantially different in length would also come out in the audio data, as would all of the other features that I mentioned.

If you were to create spectrograms for each word they would be completely distinct from each other to even a human eye across multiple dimensions. There's no good reason to claim that phonetics are the reason why their advanced statistical model connected the two words.

That's a programmer's understanding of linguistics. You're highlighting descriptive features which are approximations of averages of how we speak English under perfect recording conditions in a specific dialect. There is way more variance and alternation to how people pronounce words than what you're suggesting. How those sounds get interpreted by STT engines is likewise subject to this same variance. IPA isn't code. We don't perfectly reproduce sounds when we're speaking and certainly not to the letter of an approximated IPA transcription.

You didn't correctly transcribe Trump, btw. It's /ˈtɹəmp/ for American English. But it does exemplify my point: there are absolutely people who pronounce Trump with an unrounded vowel and STT engines have to account for this (and they get it wrong for people who are further from the average, outliers).

There are a lot of people who use a triphthong when saying racist as in [ɹɘeɪ]. With a short recording window, it's totally plausible to have these sounds be misinterpreted.

I can imagine a vector embedding where those two words have a large dot product.

Hahaha. This is the funniest thing I read this morning. Well, now we expect to see Tim Cook donating another 500 million or something to appease the dictator.

At Trump's inauguration, Tim Cook was pretty much the most prominently visible person. Right on the isle, right in the center of the camera for much of it.

This is 100% what is happening :D

I use Apple dictation a lot, and it frequently goes back and changes previous words that it's "heard" wrong after you give it more context.

And why was this flagged? It is real news about technology.

I've heard there is auto-flag present when there are a lot of arguments happening in comments.

Edit: see here.

https://news.ycombinator.com/item?id=43170850

Dang said there is "flame war detector" that flags posts.

The flame war detector doesn't flag the post (users do), but it does downrank them. You can see it in "hot" discussions. Say they go from 20 to 100 comments in minutes, mostly with deeply nested chats consisting of 2-3 people in a back and forth. The submission will fall off the front page quickly, but not be marked flagged unless users flagged it.

I don't see a lot of arguments happening here.

A lot of the stories talking about current administration gets flagged rapidly even those purely technical.

Apple's AI is probably learning you can be racist but you just can't state it.

Seems to be working well to me.

https://archive.is/cxVd1

Also - 23 points and 4 comments here: https://news.ycombinator.com/item?id=43183435

[dupe] https://news.ycombinator.com/item?id=43177857

"The best way to stop racism is to stop talking about it."

Glad Apple agrees with Morgan Freeman.

It also promotes friendly dialog between the political camps:

"Deport all illegal aliens." - "You Trump!" - "Thank you!" - "Who will pick our avocados?" - "Former federal DEI hires." - "That's extremely Trump!" - "I know, I love this guy, too!"

Someone getting fired...

Reply on news.ycombinator.com