I have used it with decent success on the IMDB App for Android and also searching on the Google Play Store, for this basic type stuff as long as you speak clearly with a concise break between words it seems to work just fine.
Not used it though for anything else really.
I think (ignoring possible accent/background noise issues in certain situations) it can work reasonably well as a keyboard replacement in certain situations where a relatively narrow range of commands or common language can be expected. However, any application that requires specialist language (business, scientific etc.) or where there will be frequent input of non-standard words (e.g. names), it is currently not even close to useful.
I use it on my Windows Phone, it's great for sending text messages when I'm driving and for setting appointments/reminders too. Plus Cortana works well with it as well
I only use it on android when I know it will understand what I'm asking for… it either doesn't like my accent (it's not heavy) on certain words. Other than that it's usually quicker to just ‘type it’ because I normally need to fix what it thinks I'm saying.
All the time. But mostly on my phone. Great for notes and not having to fish my phone out of a pocket. I do have to check any notes and documents latter though as even when it has learnt my speech patterns it still makes mistakes.
On my first android phone when it seemed all new and exciting? All the time. Now? I can honestly say never. If there was a competent assistant who can talk back to me on Android I would use it all the time. I'm not willing to switch to iOS or Windows.
Normally when my hands are full/messy. Okay Google from any page is actually quite handy.
I haven't even tried, I will wait till they can recognize my voice regardless of accent and in the language I prefer.
It could be cool however, I imagine one day to sit at my couch and shoot commands in my phone, which will be wireless connected to my TV and act like full desktop computer and use hand gestures to control from a distance while all the other input is done with voice.
Only really ever worked with cortana and I live up north!
“holy grail”? I don't think so - can you imagine trying to work in an office with more than one person and the input method is voice recognition?
Cortana is fantastic for it. I found siri/android to be not so good!
Use it a lot. Have for years. If I'm inputting a lot of text, I find it invaluable. I can sit, eyes closed, thinking about what I want to say, and with a surprising degree of accuracy, there it is on my screen.
BUT …. to get to that, requires (IMHO) :-
- a suitable working environment (i.e. minimum background noise)
- a decent headset (or, I guess, microphone)
- some time invested in training the software to your voice, and
- more time-consuming, training yourself to use clear diction, and the punctuation foibles and command structure of the software.
It is not, at least in my experience, a totally pain-free learning curve, but if you are prepared to stick to it, it can be VERY effective.
I do tend to use it for larger blocks of text, though, not things like this post.
Oh, and I use it regularly in my SatNav too.
using to voice-control elite;dangerous- rock and roll
I use it every time someone talks to me.
Really? Not now as most of the speech software are not fully compatible to our accent or say language.. I think it will take some more time to develop but I can bet that in 2020 we will be using speech recognition just as we have now adopted capacitive touch panels. We will find it use in most of the applications. I'm working on similar thing(for better future :P).
My only real experience of this is Kinect tbh. I found it incredibly frustrating as something as simple as Xbox On would take 3 or 4 tries to work, sometimes this was affected by the xbox itself refusing to respond rather than the voice command though. My accent is fairly strong I guess, somewhere between a glaswegian and a lanarkshire accent, neither of which will get you reading the news for the BBC. However I do work in a call centre and I know how to modulate my voice to be understood (took a while, I once had a german girl translate to an american guy as he couldn't understand me).
When with friends I tend to slip into colloquialisms (sp?) and my accent is a bit broader, but when dealing with customers and when abroad, I attempt to pronounce words more clearly and speak more slowly. Kinect however, refuses to recognise me even now. I have it turned off most of the time. (Have a look at Voice Actiivated Lift on Youtube if you want a visual :))
My brothers american mother in law once walked into a room where he and my mother and other brother were having a conversation. She excused herself and said “ I didnt realise you all spoke gaelic when you were together”…
VR has a long way to go for the Scots i think :)
I use voice attack with Elite Dangerous to raise and lower my landing gear. Guess my next command will be “Lights ON/OFF”
Oh and it works great for sending text on my Windows phone through my car audio while I'm driving.
Never used it, well I did about 10 years ago so may as well have been never. :)
I use it now and then, for a quick text message or when using the maps app on the android.
Never used it and not interested in it.
Dareos
My only real experience of this is Kinect tbh. ….
VR has a long way to go for the Scots i think :)
Is the limitation with Scots, or Glaswegian, accents with VR, or with Kinect VR?
Having no experience of Kinect, and not having a Scot/Glaswegian accent, I personally couldn't comment. What I would comment on, though, is that having tested several, dedicated VR systems, from the major competitors like Phillips, Lernout and Hauspie, etc, over about 20 years, they vary a LOT.
I settled on Dragon Dictate. It works for me, given the criteria I mentioned above. Also, I can dictate into a pocket voice recorder (Olympus, in my case) and then simply run the file through Dragon later on to transcribe.
My SatNav (Garmin) works pretty well, but it's with pretty simple commands. Dragon is FAR more sophisticated in what it does, but also costs about the same as the entire SatNav did.
Which makes me wonder …. if basing the effectiveness of VR on that found in a SatNav, or phone, or Kinect, are we short-selling what
'proper' voice recog can do? Is that the equivalent of assessing a car's performance or handling on a sub-£10k “reasonably priced car”, while ignoring what a £30k BMW or Merc, let alone a Ferrari, Lotus or Aston Martin, can do?
It's a genuine question, too. I don't have a smartphone or Kinect to compare to, but I do have a properly set up
and trained PC-based dedicated package, and it works remarkably well. I wonder how well it'd cope with your accent,
once properly trained?
Voice recog is not (yet, at least) something that (IMHO) is yet
quite mass-market, in the sense of open the box, plug it in, and expect 100% results. It's getting there, but still has a way to go for that level of convenience, yet alone a Star-Trekian implementation. It's probably still about at the stage cellphones were just before going digital from analog …. I.e. past the initial ‘brick’ stage and more like Sony ‘Mars bar’ CMH-333 stage, that is, fairly small, but still village idiots compared to today's models.
But give it a few years …. ;)
I dare say a decent mic and Dragon Naturally Speaking or todays equivalent would be better than Kinect and the relatively few commands its programmed to understand. I know someone who works for STV typing live subtitles and they alternate between Dragon and manually entering on a pc, so while it works well for some people it can take too long to train to a voice/accent for other situations such as a live broadcast of a political speech.
Im not sure if I would take to using VR software when it comes to long dissertations or reports as theres a lot of to-ing and fro-ing in order to get it right. I would like to see it implemented in phones for sending text messages and so on while driving, its not always ideal to call or leave a voicemail.
never. dont like repeating myself.
g8ina
I use it every time someone talks to me.
What a brilliant comment! Spot on! :laugh:
Tried using Dragon Dictate for when I am too weak to use the keyboard, but with two parrots in the room I was saying “scratch that” way too much :)
I get on ok with voice recognition in the car or using Cortana on my phone, but my wife can never seem to be understood by software, despite sounding clear enough to me.
g8ina
I use it every time someone talks to me.
Quoted For Truth :)
I have found myself using the Android one recently to set reminders or alarms. Nothing more than that yet, great to see Saracen's input too .
I only really use it in two situations:
1) Google Glass/Android Wear (LG R) - works brilliantly here, even if its mostly just a (hiking) sat nav for me these days, its great for replying to SMS on the go.
2) Oculus Rift/Elite Dangerous. Voice control really adds to the already impressive immersion, and with ED it completely changes the game. I have a great Cortana style pack for voice attack that I use, and it means you can leave you hands on the HOTAS all the time, and just talk to the ship as if it were a person.
Otherwise I find that typing or some other input method is usually much more reliable and efficient. We're not quite there yet, and even if we can crack the final bits of the technology issue, its the social one that holds us back. I still feel odd using a hands free kit and walking/talking let alone talking to my watch!
I always use it on my mobile. Don't have patience to write messages.
“How often do you use speech recognition?”
Never if at all possible.
I'm a tech enthusiast, but i've always hated voice recognition :/ My worst nightmare was back when the occasional company attempted to replace telephone numpad-tone automated systems with voice-activated ones.
Probably related to social anxiety though, i dislike using phones in general. But somehow talking to machines was worse than people, and that seems to have stuck.
Never, I don't trust it and what's the point if I have to push a button to activate it.
A utopian dream that seems more trouble than it's worth, especially with the lack of privacy these days.
I use it all the time use it on phone and I have kniect and xbox one set up as my main media box with virgin media telling xbox to watch a channel is great fun and the main reason I got my xbox
I often try to use speech recognition with my wife and kids, but they rarely do what I tell them. So my conclusion is that the technology still has a long way to go.
This might surprise some, but I actually use it LESS than I used to. At the present it's limited to purely the odd “Ok Google” when the mood takes me. By the way Dareos, I'm the only person in the household that can consistently use Google Now, it seems to have a real problem with the “educated-Fifer” brogue that my wife and kids use, but has no problem with my Sue Pollard style Nottingham twang.
Anyway, in the past I've been a sometimes-avid trier of these systems - L&H, IBM, Dragon, etc. And I've got to say that the Dragon system has consistently impressed. However, I tend to do a lot of technical documents for which, being honest, it's not exactly satisfactory. One aspect I found though, is that a good recog system actually produced better docs, maybe because having to speak it out loud made it less likely to end up with poorly structured text.
What's more use to me is not voice dictation, but voice command. In the past these systems have been touted as a “throw away your keyboard”, which they never were - but used as an adjoint to a keyboard/mouse they're actually capable of increasing your “fluidity”.
Actually, there's something I'd quite like to see in Windows 11 - some standard voice recog API so that applications could easily/quickly could be voice-command-enabled. With even basic PC's being so powerful these days, being able to multitask like that (voice command for some apps while keyboarding for another at the same time ) would be useful.
What would make me a “definite” for Windows10 though would be if I could do (while sitting in Word):
Cortana … launch calculator, 7.682 squared times 3.14 equals, copy, switch to word, paste text, close calculator without having to do any keystrokes.
All the time - I drive a lot (90 mile per day commute), but I use it more and more. Ok Google… I find it generally quite accurate.
I use voice activated bluetooth in my car but not for anything else.
Played with it in general use - Found it to work well, but far slower than quickly pressing a button for most things.
Have used it with Voice Attack for Elite Dangerous and a few other things - It's very cool, but still gimmicky.
I don't see myself using it on any serious basis until the computer can properly talk back to me and have a conversation, as per Star Trek.
When I try to look like one of the cool kids in the street but usually end up looking like some weirdo shouting at their phone.
I use it to call my wife from the car, via bluetooth. If I press the button on the bluetooth thingy to activate the voice command, then talk, it seems to have trouble picking up what I say. If I just shout my Motox X opening command, it does it via the phone and picks out the right number straight away, then transfers audio to the bluetooth. Strange.
I use it more than I thought I would, really need to get in the habit of using it more though now that the accuracy has increased a lot.
I tried using Siri when I had an iphone 5 for work, it NEVER worked for me. It normally went along the lines of:
Q “What's the capital of Argentina?”
A “Calling Steve…”
It put me off for a long time but I recently tried the google voice and it's been completely correct so far….
I use Cortana daily via voice commands for setting reminders.
Several times a week for searches and queries and calendar events and calling and reading SMS's.
What drives me up the f'ing wall is Cortana only works in portrait mode so when browsing the web in portrait I now go to google webpage for searches rather then use Cortana so I don't have to keep flipping the phone from landscape to portrait.
From many of the posts in this thread, I wonder where, and indeed if, people are distinguishing between what I'd call “voice recognition” and "voice command"?
Perhaps a better question would narrow down what the QOTW was referring to, differentiating between voice command and voice dictation, both of which use voice recognition, but are very different environments.
I do use voice command on my SatNav, not least because it doesn't require pushing a button to activate … the ‘button’ is me saying “voice command” then telling it what to do, like “go home” or “navigate, favourites, Janet”. But I don't really regard that as voice recognition, any more than commands to Windows to save or print, or Excel to select, copy, switch apps and paste.
Yet, I guess it is voice recognition, or at least, that's what underpins it.
To be clear, voice ‘command’ is, IMHO, kinda ho-hum-whatever. I use it on the SatNav because it's marginally more convenient, and faster. But I wouldn't pay extra to buy a SatNav that had it over one that didn't.
Voice dictation, on the other hand, is one hell of a productivity booster for me. Bear in mind, I'm a writer, usually working at home, few distractions and little ambient noise, unless I have music on. But I can sit down and dictate, more or less non-stop, for an hour or more in a block, and with voice recog/dictation, I get a lot more words down in that hour. Often, double or more. Triple, even. Why? Just a guess, but because I'm thinking about what to say, not typing, reading, correcting my dodgy typing AND thinking about what to say.
In THAT context, all I can say is that a GOOD voice recog package (Dragon, for me) AND a decent headset (got several, all old now, but Plantronic noice-cancelling USB unit is my normal choice) is certainly one of, if not the best investment I ever made. It's that useful to me.
Of course, a good system, used in a quiet environment, to dictate several hours of text a day, is a very different scenario from trying to use a mobile phone voice command to locate the nearest Starbucks while walking down an unknown high street. I doubt much of a proportion of posters here have the need, or use, that I do, for VR.
Oh, and in case anyone wonders, most posts here are on an Android tablet with touch screen, not a PC with Dragon …. which explains the typos, and/or time spent editing them out.
Is there really much difference between the two, with advanced systems at least?
If you put the very basic one-two word commands to one side, and look at something like Voice Attack - i'd suggest it can be just as complex as something like Dragon. The profile i'm using for Elite for example can understand long sentences/phrases and interpret correctly, and makes use of a rudimentary “AI” to respond. E.g. i've got it configured to require an initiation phrase before each command, and to check the current status. These can be combined too..Very much like the star trek computer. (note this is all based on Hazey's “S.A.R.A” profile, which taught me all the basics! All credit to Hazey)
https://www.youtube.com/watch?v=NdpsAucSIgwe.g.
Me: “Sarah?”
Ship: “<reply noise>”
Me: “Deploy landing gear”
Ship: either: “Landing Gear Deployed” or “The landing gear is already deployed, commander”
This can be also be combined as “Sarah, Deploy landing gear” (all one phrase), or even more complex commands like “Sarah, deploy hardpoints and target the first enemy”
Yes its still just picking up phrases and then doing a quick comparison - but it does provide for context, which I think is one of the more important/difficult aspects of voice recognition. Context is also the big selling point of dragon (and back in the day, ViaVoice) as once you get past the basic “what word was that” the context of a given word can have a huge impact on what the software interprets.
Whether its quicker/more efficient etc is a very personal thing. I only use it in situations where I have no alternative - I can touch type much faster than I can dictate and get much better accuracy, so I tend to use voice control only where I don't have access to a keyboard..and thats typically in something like Elite, using the Rift or with a totally hands free tech like a watch/glass.
Saracen
From many of the posts in this thread, I wonder where, and indeed if, people are distinguishing between what I'd call “voice recognition” and "voice command"?
I did.
Saracen
To be clear, voice ‘command’ is, IMHO, kinda ho-hum-whatever. I use it on the SatNav because it's marginally more convenient, and faster. But I wouldn't pay extra to buy a SatNav that had it over one that didn't.
Voice dictation, on the other hand, is one hell of a productivity booster for me. Bear in mind, I'm a writer, usually working at home, few distractions and little ambient noise, unless I have music on. But I can sit down and dictate, more or less non-stop, for an hour or more in a block, and with voice recog/dictation, I get a lot more words down in that hour.
Agree with what you're saying about voice dictate but, like I said, it really depends what you're writing. If it's docs with a lot of technical content (or in my case user manuals) then you're on a hiding to nothing … you spend more time correcting prediction errors than the recognition system saves. On the other hand, I'm having to multitask, so anything that assists that would be useful.
I really don't like voice command for satnavs, but that's maybe me being cheap and refusing to buy a standalone unit. I'd prefer to pull to the side of the road and press-screen. And tbh a smartphone satnav app is good enough for the few times I need a digital map reader.
Here's a clarification question - how enthusiastic would people be about using speech recognition at home? For a (large?) proportion of the Hexus readers I suspect SR would be more suitable for domestic use than @work.
PS totally, totally agree about the use of a good headset - from what I did in the past it can make the difference between “usable” and “take this thing away”. Trouble is that “management” invariably want you to use whatever CorpIT can find at the bottom of the stores cupboard.
I have Windows phone and whilst it is great that Cortana interfaces with my car's Bluetooth to read out text messages for me, her voice recognition is not great and seldom writes anything comprehensible in terms of a reply; for instance, this is a text I sent earlier:
“F*** knows cortana is worse in the. ASX 6 Dan Doran. The van it's just fantasy what she comes up with.”
…what I actually said was; “**** knows! Cortana is worse in the ASX than the van. It's just fun to see what she comes up with”
Firstly, I see no reason why I should be censored by my mobile phone and secondly even though I try to speak clearly, it never ends up like what I say!
I've found google pretty good for voice but i never use it on my phone. Far easier to just type…
crossy
…
Agree with what you're saying about voice dictate but, like I said, it really depends what you're writing. If it's docs with a lot of technical content (or in my case user manuals) then you're on a hiding to nothing … you spend more time correcting prediction errors than the recognition system saves. On the other hand, I'm having to multitask, so anything that assists that would be useful.
I really don't like voice command for satnavs, but that's maybe me being cheap and refusing to buy a standalone unit. I'd prefer to pull to the side of the road and press-screen. And tbh a smartphone satnav app is good enough for the few times I need a digital map reader.
Here's a clarification question - how enthusiastic would people be about using speech recognition at home? For a (large?) proportion of the Hexus readers I suspect SR would be more suitable for domestic use than @work.
PS totally, totally agree about the use of a good headset - from what I did in the past it can make the difference between “usable” and “take this thing away”. Trouble is that “management” invariably want you to use whatever CorpIT can find at the bottom of the stores cupboard.
Agreed on technical uses but there are (or used to be, though I haven't checked recently) some specialist dictionaries for things like legal, medical, maybe others. But not being in those fields, I've never tested them.
As for standalone v. smartphone, personal choice, I guess. I don't have a smartphone, mainly because I don't need one, and for privacy reasons.
On the headset issue, that's not the fault of VRecog, though, but of cheap management. I guess I'm lucky in that - it's been several decades since I lost an argument with management - I always get my way in the end. And yes, I work for myself. ;)
I use Dragon 13 almost every day on my PC.Its as good as my speech input is accurate.
The worst thing about Dragon is installation which can be a nightmare.
I experienced very poor support from Nuance for versions 12 and 13 and had to resort to a clean install of Windows 7 then install Dragon to get it to work.
Worth the hassle though.
uksnapper
…. Its as good as my speech input is accurate.
….
Good point. I said earlier that effectiveness is partly about training Dragon, but also partly about training me. If I speak clearly, with defined words, it's far better.
Compare
1) Canureadthiswhentryingtounderstandme?
with
2) Can. you. read. this. when. trying. to. understand. me.
The full stops are just to indicate using a clear word diction, not a rather more sloppy and lazy ‘conversational’ way of speaking.
Using 2) feels a bit unnatural at first, but no doubt mainly because my normal diction is …. lazy. Speaking almost in a 1950's BBC style (but without the plummy accent) is kinda similar.
Dragon does, IMHO, work
extremely well, but like most complex software, you have to work out the best way to drive it effectively. There is definitely a learning curve. If you input a lot of text, then it's my experience it's well worth it.
I've used it enough to know that it's rubbish and you look like a plonker talking to yourself.
Speech recognition on Google Now using my phone seems pretty accurate, but I only regularly use it to kick off the satnav function (and then almost entirely with “navigate to work” and “navigate to home”). So I'm using it like a glorified hotkey.
I don't think I'd really trust it for much more than that.
aidanjt
I've used it enough to know that it's rubbish and you look like a plonker talking to yourself.
Look like a plonker to who(m)?
Sitting in my home office, with nobody else in the house, there's nobody to see.
Besides, it's perception. Not that long ago, you looked like a plonker walking down the street holding a phone to your ear. Then, the masses got used to it and it's seen as socially acceptable. But you looked like a plonker walking down the street with a barely visible bluetooth headset plugged into your ear, then people got used to it.
We could argue that looking like a plonker using VRecog only makes you look like a plonker, if indeed it does, because it's not (perhaps, not YET) mass market.
That said, in the way I use it, VRec makes a significant difference to my productivity, so I don't give 3/4 of 5/8ths of a flying fig if people think it makes me look liks a plonker. I'm pretty much indifferent to what other people think, just as I am of their view of my fashion (or anti-fashion, or better yet, a-fashion) sense. I wear what I want, because I like it or it's comfortable, and whether it's the height of fashion trends or a ghastly fashion gaff bothers me not one tiny, tichy, minuscule little jot. ;)
Aren't I quite the rebel? :D
“Canureadthiswhentryingtounderstandme?”
Yep, trivially. I knew exactly what it said within two saccades.
But then I'm not some dumb computer with a two-bit, linear processor. ;-)
devBunny
“Canureadthiswhentryingtounderstandme?”
Yep, trivially. I knew exactly what it said within two saccades.
But then I'm not some dumb computer with a two-bit, linear processor. ;-)
I think you're missing the point. Unless you're offering to come round here every day and do my typing for me. Free of charge.
The point was “compare”. The text could have been anything, and the question was rhetorical, to try to illustrate that the way I speak when dictating significantly affects accuracy of results, but IF I speak carefully I get near 100% accuracy, AND typing doesn't interfere with my thought process, so I don't lose track, lose focus or disrupt the text.
I drive VRecog HARD. And it works. It paid for itself in a few days (if not hours) in time saved, and I've been using it for about 15 years.
Saracen
Look like a plonker to who(m)?
Yourself.
Saracen
Sitting in my home office, with nobody else in the house, there's nobody to see.
Well there's every possibility that the photons that bounce off you talking to yourself will run out the window, head into deep space, reach an inhabited planet thousands of light years away and you'll make the poor aliens feel sorry for you.
Saracen
But you looked like a plonker walking down the street with a barely visible bluetooth headset plugged into your ear, then people got used to it.
No, it's just bluetooth users who think they have. Everyone else affectionately refers to them as blue{holes,tards,tools}. ;)
Saracen
We could argue that looking like a plonker using VRecog only makes you look like a plonker, if indeed it does, because it's not (perhaps, not YET) mass market.
After how many decades now? :D
Saracen
That said, in the way I use it, VRec makes a significant difference to my productivity, so I don't give 3/4 of 5/8ths of a flying fig if people think it makes me look liks a plonker.
True. Although in my experience I spend more time correcting the recog algorithm's screw ups than I would just typing it out, and that's not even including the time wasted from talking to myself, which in itself is longer than it would to just type it out. YMMV, I guess.
Saracen
I'm pretty much indifferent to what other people think, just as I am of their view of my fashion (or anti-fashion, or better yet, a-fashion) sense. I wear what I want, because I like it or it's comfortable, and whether it's the height of fashion trends or a ghastly fashion gaff bothers me not one tiny, tichy, minuscule little jot. ;)
Aren't I quite the rebel? :D
Uh huh. :P
I'm much the same regarding clothing, to me it's a tool to insulate my skin to stay warm and not get arrested. So I have a lot of black tshirts and jeans pretty much. But talking to myself is where I draw the line!
Well, I have been known to talk to myself when not using VRecog, and even to swear at myself on occasions, so maybe I can use VRecog as cover for that?
Agreed on the clothing thing, though I'd add “feels comfortable” to insulation and not getting arrested. Especially for shoes.
tried it , doesnt work at all (pc) , main problem is that a lot words are problematic , especially if your native language isnt english or you have a heavy accent
Never tried it before on a mobile or PC, but just got a hold of the latest version of Dragon Naturally speaking to try on my PC. My LG TV's remote has this function and it works fine to change the channels.
If they can get it to work properly on a phone for texting, it would help fix the alphabet soup I get from some people and may make it easy to understand. At times I have had to text back saying I don't know what language they are using, but please use English. If we have trouble understanding each other, how hard must it be to design software to translate it into something that makes sense.
I use it sometimes with OK google, but depends if I have a signal on my phone or not.