Voice recognition and activation!

[dmale7]

I posted this on the NX forum as well but...

I have lived through Wondersilk, Photoskins, launcher wars, Mythomania, dvd ripping to clie veiwing, WiFi, bluetooth, tappads, jackflash, and cf drivers...I want more!

Am I the only person that thinks that Voice activation and recognition would make the NX's and NZ sickeningly awesome?&nbsp; I'd like to know how many of us agree with this and what do some of you think could be done to bring this about.&nbsp; <IMG alt="" src="http://www.cliesource.com/forums/images/smilies/cool.gif" border=0>





Report (http://www.cliesource.com/forums/report.php?s=&amp;postid=197354)

[Cymril]

Agree, this would be a very nice addition. But I don't know if it has enough power for it.

[megazone]

I've never seen the big appeal of voice control. I have voice dial on my cell phone, and that's cool, but I wouldn't want to constantly speak my commands. Even the best software I've seen, on fast PCs, has some lag and isn't accurate enough for me. Plus you'd have to wear a headset (I have a Jabra FreeSpeak BlueTooth headset for my phone) to work it - background noise would interfere too much with a mic on the unit out in public, unless you want to keep holding it up to your face.

I can tap, hit a button, whatever, MUCH faster than I can speak commands that a unit will recognize.

And no way will it do voice recognition text entry without an order of magnitude more power - and even then voice recognition software isn't that reliable yet.

[charlie4n6]

I think text to speech would be even better :)

[dmale7]

Originally posted by megazone
I've never seen the big appeal of voice control. I have voice dial on my cell phone, and that's cool, but I wouldn't want to constantly speak my commands. Even the best software I've seen, on fast PCs, has some lag and isn't accurate enough for me. Plus you'd have to wear a headset (I have a Jabra FreeSpeak BlueTooth headset for my phone) to work it - background noise would interfere too much with a mic on the unit out in public, unless you want to keep holding it up to your face.

I can tap, hit a button, whatever, MUCH faster than I can speak commands that a unit will recognize.

And no way will it do voice recognition text entry without an order of magnitude more power - and even then voice recognition software isn't that reliable yet.

&nbsp;

I see your point Megazone, however, if the voice commands were very specific I do not believe background noise would be a problem.&nbsp; The voice recorder is very sensitive, so I do not believe a headset would be needed.&nbsp; Along with speech-to-text capability, I believe opening apps via voice command could be extremely useful.&nbsp; It occurs to me everytime I'm using my cell phone and have to pull an address or phone # from my NX...or if I need to jot something down while driving.&nbsp; (This happens often for some reason)

Alternatively, since the thing cost a small fortune, I believe that we should wring the thing dry of it's potential.&nbsp; I would like to reach the point where we can all say that we can do no more with this PDA.

[dmale7]

Megazone or anyone,

Please explain why more processing power is need for voice commands. If a cell phone can do simple voice recognition why can't my NX?

[Unregistered]

Originally posted by dmale7
Please explain why more processing power is need for voice commands. If a cell phone can do simple voice recognition why can't my NX? There is a huge processing leap from voice commands to voice to text.

For voice commands you 'teach' the unit - for my cell phone it only recognizes the commands that I've preset. I have to program it by speaking the command. Then to use the command I have to tell it to listen (on my headset I press a button, or I can press and hold the 'Yes' button on my phone until it beeps). Then I have to speak the command in close to the same inflection and cadence that I used when programming it. All this does is sample my voice and look for a close match between the encoding and the pre-programmmed encodings.

So that's a VERY limited pool to handle - at most a few hundred samples to check on my phone, and right now a couple of dozen in practice.

Text to speech needs to recognize thousands of words, ir not tens of thousands for really useful software, and you don't train it to recognize every word. It needs to do MUCH more processing with a lot more intelligent software to figure out what phonemes you've said, then match them to the correct word. Good voice input software also takes context, grammar, etc into consideration to help determine the right word. And it needs to do this MUCH faster than command input - it needs to keep up with comfortably spoken speech.

For voice commands you usually speak one word, then pause, to give it a chance to act. I've used it on a PC, and the speaking pattern tends to be Shatnerian:
File. <pause> Open.

For my cell phone it is actually:
*push button* *beep* Name <pause> *beep* Home/Work/etc <pause> *dial*

You. Don't. Want. To. Have. To. Speak. Like. That. To. Enter. Text.

There's a reason why voice recognition still hasn't gone mainstream even with 3GHz CPUs and 512MB RAM, etc. It is *vastly* improved to just a few years ago, but it still needs work.

[megazone]

OOps, that was me, apparently not logged in.

[dmale7]

Thanks for the explanation Mega! It seems from what you say, that speech-to-text may be out of the question. I still like the idea of simple voice commands to open and use certain apps.

How about something like this...

An app that allows you to 'train' the NX or NZ to write what you say like with voice COMMANDS. Maybe a 'library' can be compiled
with voice commands that do what we preset it to do, whether it's writing text or performing some other operation. Something like an audio mcphling...or a shortcuts but with recorded voice commands.

[GolfBot]

Hello,

my Siemens Xelibri mobile phone has quite a voice recognition, but I noticed that I am almost never in the situation to talk to my phone. Be it in the subway, on the airport or even at home with my family, it feels very strange to talk to my phone if a few taps on its keyboard bring me as fast to the desired result. And background noise worsens results notably. I felt equally strange to sit in teh subway and tell my TH55 to open Plucker. Really.

What I'd like to see in my Clie OS is one feature that lets me highlight text or scribble it down and it triggers other actions; something the Newton MessagePad had, when you could scribble "Call John" or "Meet Peter tomorrow at 12 am" and it would open the respective application and perform like commanded, scheduling an appointement or whatsoever. I always liked the feature and felt it added to the Newton's Man-Machine interface...

And yes, speech-to-text would be a bummer as well.

Cheers, GB

[Sentient Carbon]

Megazone - That just made my all time best new word list. I love it! :D

[spidey]

Megazone or anyone,

Please explain why more processing power is need for voice commands. If a cell phone can do simple voice recognition why can't my NX?

My understanding is that voice commands on a cell phone are processed on the telco's computer not by the processor onboard the phone?
Can someone confirm this?

Some voice mail and customer service systems (interactive voice response, IVR) now also do this.
It can understand almost anybody's speech and does not need to be trained.

I also would like to see simple voice commands on the Clies, if it's done with a limited trained command set and pattern match recognition it has a much better chance of working.

Speech to text is another story.

Text to speech has been done by PSpeak and SpeakMemo and bookreader. http://www.mapopolis.com/free.jsp?s=1089151485609616&p=&q=6
I'd love to see it done for appointments and alarms.

Anyone know if any Datebook apps that have the ability to play a different WAV file for each alarm/appointment?

[asghar]

Most of the time when you are using your PDA you are in a public place, last thing I would like as a bystander is to hear anyone talk or yell to their PDA. I think your problem with pulling up numbers from the PDA to dial is going to be solved by the phone and PDA combined units.