Speech

Lake · 01-21-2005, 11:10 PM

Macs have had speech (both text-to-speech and recognition) capabilities for quite some time, but not alot of people are terribly familiar with what it can do. I was pretty well versed with it back in OS9 and earlier, but the latest revisions of the OS have made some great progress.

Speech Recognition + AppleScript is a very potent combination for even beginners. With a little forethought it should be simple to work into any application.

Here is a good article that should bring you up to speed and start you in the right direction: http://www.macdevcenter.com/pub/a/ma...17/speech.html

~Lake

gork · 01-21-2005, 11:30 PM

Ah yes, I had forgotten about the inbuilt voice recognition. I never use it, but it's a piece of cake to program. The code basically looks like "Listen for the user to say 'Play DVD' then start 'DVD Player'..."

Lake · 01-22-2005, 12:28 AM

Here's a little example I made, playing with it some. I'd almost forgotten how much fun AppleScript was when you get it right the first time (though it's no more pleasant to debug than anything else, heh).

Code:

tell application "SpeechRecognitionServer"
	local choices, question
	set choices to {"Slashdot", "MacSlash", "Google"}
	set question to "Where would you like to go today?"
	try
		set result to listen for choices with prompt question giving up after 10
	end try
end tell

tell application "Safari"
	if result is "Slashdot" then
		open location "http://slashdot.org"
	else if result is "MacSlash" then
		open location "http://macslash.org"
	else if result is "Google" then
		open location "http://google.com"
	end if
end tell

~lake

Lake · 01-22-2005, 06:58 PM

just an FYI: The above script actually works but only by accident. The word "result" is reserved in AppleScript, it's automatically assigned the value of the last expression evaluated. To be 'sane' you'd want to just replace all instances of the word 'result' with some other appropriate variable name (like 'response' or somesuch).

It's been ages since I've really done much with AppleScript, so the above was just thrown together in a vary haphazard way. Re-reading a few things now =)

~Lake

mattdisaster · 01-22-2005, 07:08 PM

gork if what you say is true, could you possibly program it to play all of an artist? For instance say "play the strokes" and have it play all by the strokes? I hope so, that could be really really really cool.

-matt

Bugbyte · 01-22-2005, 09:42 PM

Right, you could do that. The iTunes Applescript dictionary includes ways to search by artist, genre, album, etc. You could do what you're suggesting easily.

mattdisaster · 01-22-2005, 09:46 PM

Oh my lord that has just added gallons of fuel to the fire of me getting a minimac. This will be the best project ever!

-matt

0l33l · 01-22-2005, 09:50 PM

Quote: Originally Posted by mattdisaster

Oh my lord that has just added gallons of fuel to the fire of me getting a minimac. This will be the best project ever!

What's so special about speech recognition? Windows PCs have been doing it for a long time, and we have quite a few frontends that do speech. Some do speech from the get go, like centrafuse, and others like Frodo do it with the help of support programs, like NaviVoice

mattdisaster · 01-22-2005, 10:17 PM

No way? I honestly had no clue it could be done on a pc, I'll have to check into it then, I cant think of anything cooler then "play The Strokes, Is This is" "Mute" "volume Up" oh man, I'de be the coolest person on the block! hah.

-matt

Lake · 01-22-2005, 11:01 PM

It's not that Macs are unique in their ability to do speech, but it is incredibly well integrated into the operating system and available in the unix shell environment, applescript, and cocoa, as well as ready-made for the user.

Turning on a few options in the system preferences lets me do some rather impressive stuff right off. I can activate buttons on the frontmost window by saying their label, or navigate items in the menu bar (for example in iTunes: "Controls menu; Next Song" will use that menu item to skip the current song).

So you don't even need to specially support Speech in your own application, just take care to 'play nice' and it's automatically supported.
--platform rant--
I gave up on preaching about Macs to 'non-believers' a long time ago. But since you brought it up: It's just one more thing that works the right way on a Mac. It's not impossible in Windows, but it's generally a real pain in the *** and doesn't live up to its potential.
--/platform rant--

~Lake

P.S. I've actually developed for Windows and run a WinXP box for (certain) games, and happen to be typing this on a WinME machine. So it's not as though I'm ignorant of what Windows really is :P

0l33l · 01-23-2005, 12:03 AM

Quote: Originally Posted by mattdisaster

No way? I honestly had no clue it could be done on a pc, I'll have to check into it then, I cant think of anything cooler then "play The Strokes, Is This is" "Mute" "volume Up" oh man, I'de be the coolest person on the block! hah.

-matt

You can't say "play song", but you can say play, pause, stop, and you can also control navigation

Budman · 01-23-2005, 08:04 AM

Quote: Originally Posted by 0l33l

You can't say "play song", but you can say play, pause, stop, and you can also control navigation

I've been using voice in the car for about a year with a pc. In WMP I can control: Play, Pause, play artist/genre/playlist, view/hide media info, view/hide playlist, skin/full mode, among others. With GPS the main commands I use are show next turn (visual), next turn (verbal), Where Am I, among a few others.

They can be used simultaneously by setting different call signs for each app.

Once the voice training is done it's fairly accurate.

kandyman676 · 01-23-2005, 11:42 AM

Actually, the speech parser has been built since Mac OS (since 1997 i believe...I remember playing with it back in high school) way before it was even considered a stanard package in the default Windows XP as an accessibility tool. speech recogniation for Windows has always been on and off for years. I remember the first version of Dragon Natually Speaking. :-) when I played with the apple back in high school, it was several jumps ahead of the windows software.

remember, we are suppose to be discussing APPLE/MAC speech, not PC. its great to know the PC options, but please keep them to the PC threads:-)

0l33l · 01-23-2005, 12:04 PM

Quote: Originally Posted by kandyman676

remember, we are suppose to be discussing APPLE/MAC speech, not PC. its great to know the PC options, but please keep them to the PC threads:-)

Yeh, yeh, yeh

thecapitalizt · 01-23-2005, 12:26 PM

Theoretically, you don't have to even prorgram anything in to make speech recognition work. Turn on the 'press buttons' option in the accessibility pane, and it will allow you to press any button in a User interface, or a menu option.