Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

25 Years In Speech Technology and I still don’t talk to my computer

matthewkaras.medium.com

Post reply on HN
Print view View original

251–260 of 296 posts

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#251

Post by Kye »

dheera wrote: post #5

A big problem with the assistants is that as soon as they fail at a query they seem stupid, I feel stupid, and I stop using them for a long time. Last weekend I had the following failed queries: "OK Google, what is the air quality like at Mt. Shasta today?" "OK Google, add a waypoint for the last gas station before the mountain pass" "OK Google, what percentage of people can you detect to be wearing masks on recent I…

"Hey Siri, remind me to add [x] to the grocery list at [y]."

"There is no 'Grocery' list. Would you like to make one?"

An example of how Apple's ecosystem breaks down if you want to use something outside it. Saying no should offer to add a reminder as requested, but it doesn't.

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#252

Post by aaron695 »

It's the classic IT scam that is the entire industry.

When people demo Speech Technology they have the computer doing general AI in response.

Things that are impossible with text or any computer input or anything that can happen for decades.

Reality is, no one wants Speech Technology, we want a computer that can order a pizza with the simple typed words, get me a pizza.

Heck, you pull that off people would even do the way more annoying 'talking' to the computer to access that tech.

What people really want is text to speech. But that's to hard to do the AI scam on.

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#253

Post by p1necone »

frankosaurus wrote: post #144

For me, the thing that killed voice commands had nothing to do with speech technology. It was latency and error handling. At the start of my morning commute, I would say, "Ok Google, navigate to work". Often, this would fail because I was in the network limbo area outside my house, where my phone struggles to transition from home WiFi to data. Worst of all: The failure would be horribly slow. I would have to drive fo…

It took me months to work out that "navigate to X" was the magic phrase to get google maps to do what you would expect in car navigation to do. The phrase that came more naturally to me was "give me directions to X", but that only gets you to the screen with the route and you still have to manually press the "start" button with your finger. And then it would randomly pick other modes of transport unless I remember to say "by car". Systems that are inherently unrecoverable like voice commands need actual documentation.

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#254

Post by 205guy »

dheera wrote: post #232
mikestew wrote: post #143

Earlier quoted context omitted.

"...set an alarm for tomorrow @ eighteen hundred." works just fine. At least in U. S. English, I don't think I've ever heard anyone refer to 6 p. m. as "eighteen".

Except it isn't a hundred, it's sixty, and I protest that usage. >.<. I prefer eighteen o'clock.

It's standard US military terminology and part of US culture and language.

Just imagine all the idioms they have to account for in these voice algorithms. I know in German they also handle "halb neun" correctly as 8:30.

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#255

Post by selestify »

overthemoon wrote: post #107

I suck at talking. I speak in halting, quiet sentence fragments. My mind wanders and I lose my point. I get in my head a lot. I'm much better at writing and reading as a communication method. I'm open to speech stuff, but currently no speech recognition solution meets any needs I have. Just my 2c, I'm sure other people have uses for it. The most interesting one (to me) has popped up a few times on HN, which is voice-…

I do the same when talking. Never realized others have this condition as well :)

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#256

Post by GirkovArpa »

PeterisP wrote: post #137
ketzo wrote: post #48

Earlier quoted context omitted.

Well, to extend the GUI/console metaphor, it means that at some point soon, we'll all be using NLP because it's dramatically more user-friendly for the vast majority of people.

GUIs won over console workflows because GUIs have better discoverability and the "recall vs recognize" difference; it's mentally much easier to recognize the option you want when presented it than to recall the existence or the naming of that option. In those aspects of UX, voice interfaces have the same drawbacks as console apps when compared to a good GUI. Also, they have to work within the "bandwidth bottleneck" o…

This "recall vs recognize" point should be raised in every console vs GUI debate. It's pretty much the final word.

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#257

Post by 205guy »

crazygringo wrote: post #111

For most short interactions, the mouse/trackpad/finger is simply faster. Now for long-form typing, I'd love to use dictation and sometimes do for taking down short thoughts I e-mail myself from my iPhone. But the problem is not just that it still makes tons of mistakes. (Probably a quarter of my notes-to-self involve errors so big it's even impossible for me to later figure out what I even meant by trying to sound it…

Picks up the mouse and speaks into it: "Computer?"

Classic scene: https://youtu.be/xaVgRj2e5_s?t=171

Star Trek isn't my favorite sci-fi show/movie, but they really nailed some futurism and in this case, lack therof.

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#258

Post by 0xffff2 »

YeGoblynQueenne wrote: post #211
fennecfoxen wrote: post #153

Earlier quoted context omitted.

> They seem to actually use some sort of formal query language that fits English a bit closer, but is still distinct from when the characters speak to each other. "Computer, begin auto-destruct sequence, authorization Picard 4-7 Alpha Tango." - Wake word. Command. Authorization stanza. (I bet the computer would prompt for authorization if missing.) "Computer, Commander Beverly Crusher. Confirm auto-destruct sequence,…

To be fair, that's just how Picard speaks (e.g. "engage"). I haven't noticed anyone else saying "Tea. Earl Grey. Hot". In any case I think this kind of speech is formulaic for the benefit of the audience, most of all, who are made aware through the formality that the speaker is addressing a machine. Additionally, we're watching navy men and women in space, so we expect them to speak to each other and to their compute…

Maybe I'm misremembering, but I distinctly remember that basically everyone ever shown interacting with a replicator follows the generic->specific parameter hierarchy.

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#259

Post by remir »

I often use Siri when I'm driving to send text messages and it's OK. The speech recognition in Google Maps also works great.

But beyond that, I don't really use these "digital assistant" because you can't teach them when they're wrong, so after 2-3 failed attempt at a request, you lose interest because you can't trust the system.

Re: 25 Years In Speech Technology and I still don’t talk to my computer

#260

Post by whimsicalism »

sida wrote: post #71

One thing I haven't seen discussed is the poor affordance / discoverability of speech technology. Google Home can do some clever things however, it also does not have the ability to do some very basic stuff. As a user, how do you know what Google Home can do and cannot do? It is just trial and error. And if Google Home introduces a new feature to be able to complete new types of queries. What now? How does a user kno…

Once they start hooking them up to conversational language models that can also submit queries, I think it is going to get a lot better. The conversational model results are starting to look very good.
Post reply on HN