Cycling 74 » Topic: aka.listen assitance http://cycling74.com/forums/topic/aka-listen-assitance/feed Tue, 18 Jun 2013 00:25:14 +0000 http://bbpress.org/?v=2.2.4 http://cycling74.com/forums/topic/aka-listen-assitance/#post-38756 <![CDATA[aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-38756 Sun, 06 Jul 2008 05:34:20 +0000 Mr.Grey I’m interested in doing a project that uses speech recognition. But I can’t seem to get aka.listen to even recognize my audio input on my external interface. It’s configured properly in DSP status as I’m getting a level in Audiotester. Is this object hardwired to a system input or something? Any tips would be appreciated.

]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135406 <![CDATA[Re: aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135406 Sun, 06 Jul 2008 06:15:28 +0000 Rob Ramirez yeah, it takes input from the operating system audio settings.

]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135407 <![CDATA[Re: aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135407 Mon, 07 Jul 2008 01:01:19 +0000 Karl Kliem i needed the same thing for a students project recently. the solution was to use the built in speech recognition of mac os.

1. turn on speech recognition in system preferences

2. calibrate it and change the speech settings for your need

3. put an apple script inside the folder ~/Library/Speech/Speakable Items/Application Speakable Items/MaxMSP/

4. give the script the filename that you want max to react to. for example the following script has the name “yes.scpt”. as soon as you say “yes” it will send the keystroke “y” to max, that you can get with the key object in max.

the applescript “yes.scpt”:
tell application “System Events” to keystroke “y” using {command down}

k

]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135408 <![CDATA[Re: aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135408 Mon, 07 Jul 2008 01:13:58 +0000 Rob Ramirez or just use aka.listen?

]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135409 <![CDATA[Re: aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135409 Mon, 07 Jul 2008 08:23:02 +0000 Karl Kliem when i tried aka.listen only took input from the built-in microphone of a macbook. no way to change the input to an external audio interface.

k

]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135410 <![CDATA[Re: aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135410 Mon, 07 Jul 2008 16:04:19 +0000 Mr.Grey I was able to get it working with the XLR input on my Konnect24D… so it definitely works with external interface.

I was happy that it worked reasonably well, but still really inaccurate. I guess it depends on your expectations, but I need something that is 99.999% accurate without needing to repeat yourself because you said it weirdly or didn’t pause long enough.

Ultimately the issue is that as far as I can determine general voice recognition isn’t very good. Dragon/Dictate have fairly good accuracy if you spend the time to train it to one voice.

Rather than one tool that recognizes anything someone might say as accurately as possible, I need a tool that will recognize a few specific phrases from anyone who might say them… which appear to be discongruous goals.

I’m sure said software exists in proprietary contexts, so it’s unfortunate that there’s nothing openly available for people to integrate into projects.

Please correct me if you know of something. That would be helpful, indeed!

]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135411 <![CDATA[Re: aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135411 Mon, 07 Jul 2008 16:11:41 +0000 Rob Ramirez you have to change it with the system preferences

]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135412 <![CDATA[Re: aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135412 Mon, 07 Jul 2008 16:13:17 +0000 Rob Ramirez > I was happy that it worked reasonably well, but still really inaccurate. I guess it depends on your expectations, but I need something that is 99.999% accurate without needing to repeat yourself because you said it weirdly or didn’t pause long enough.

yeah, that’s free speech recognition software for you. i was actually quite amazed at how well it worked, and how many commands you can store.

]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135413 <![CDATA[Re: aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135413 Mon, 07 Jul 2008 16:17:13 +0000 Mr.Grey Yeah. It’s definitely decent for something that’s free. Even the commercial software hasn’t been that much better until recently. Dictate is very accurate once trained. But again, for multiple users it’s a shame there’s no accessible solution.

]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135414 <![CDATA[Re: aka.listen assitance]]> http://cycling74.com/forums/topic/aka-listen-assitance/#post-135414 Mon, 07 Jul 2008 16:54:21 +0000 Stefan Tiedje Clayton Grey schrieb:
> But again, for multiple users it’s a shame there’s no accessible
> solution.

Maybe its not as easy as you think it should be. Maybe because you can
recognize speech quite well…

To learn how difficult such a thing is to create, read “Goedel Escher
Bach”…
Speech recognition is magnitudes more complex than text recognition…

And even you are not able to understand 99.999 % of a single word. Most
words we understand because they come within a context of other words…

Stefan

–
Stefan Tiedje————x——-
–_____———–|————–
–(_|_ —-|—–|—–()——-
– _|_)—-|—–()————–
———-()——–www.ccmix.com

]]>