Palaver speech recognition app packaged for Arch

Padfoot · 2013-03-23 05:30:48

Palaver (formerly Ubuntu-Speech-Recognition), has been packaged for Arch:
https://aur.archlinux.org/packages/palaver-git/

The git repo is located here:
https://github.com/JamezQ/Palaver

A great video demo of what is possible can be found here:
http://www.techdrivein.com/2013/02/ubun … -demo.html

This is shaping up to be an interesting project, and as long as it keeps on a good development track, could become the Siri of Linux (don't laugh, it could!)

The current code is beta, and will be going through restructuring changes as it moves from git to launchpad, so expect a lot of changes in the near future.

The beta can actually do quite a lot at the moment, especially if you add your own dictionary (which is very easy to do BTW).

Oh, and just a warning regarding privacy, the application uses Googles speech recognition, and requires a network connection to work. Your voice command is recorded locally and deciphered on Google servers.......

Cheers.

Xyne · 2013-03-23 06:11:05

Padfoot wrote:

Oh, and just a warning regarding privacy, the application uses Googles speech recognition, and requires a network connection to work. Your voice command is recorded locally and deciphered on Google servers........

Oh well. Aside from the privacy concerns* I am also disappointed that it is just a wrapper around a web service. A local speech recognition engine would be even more impressive.

Thanks for the privacy warning.

* Seriously, sending speech samples to Google so that they can store and analyse them is crazy to me. Do you really want to live in a future of interactive advertisements that can identify you by voice alone and associate it with everything else that you have ever done online? A conversation with a friend at a bus stop may one day trigger targeted ads that reveal things about you that you consider private. I do not understand how so many people can be completely ok with having their private lives catalogued for companies and governments just to get some non-essential services in return. Beyond that having such tools lying around when your government eventually becomes oppressive will ensure its longevity at everyone's expense. Open, democratic societies have an unpatched memory leak that requires a hard reboot every so often.

Padfoot · 2013-03-23 22:14:57

Xyne wrote:

Padfoot wrote:
Oh, and just a warning regarding privacy, the application uses Googles speech recognition, and requires a network connection to work. Your voice command is recorded locally and deciphered on Google servers........
Oh well. Aside from the privacy concerns* I am also disappointed that it is just a wrapper around a web service. A local speech recognition engine would be even more impressive.
Thanks for the privacy warning.

* Seriously, sending speech samples to Google so that they can store and analyse them is crazy to me. Do you really want to live in a future of interactive advertisements that can identify you by voice alone and associate it with everything else that you have ever done online? A conversation with a friend at a bus stop may one day trigger targeted ads that reveal things about you that you consider private. I do not understand how so many people can be completely ok with having their private lives catalogued for companies and governments just to get some non-essential services in return. Beyond that having such tools lying around when your government eventually becomes oppressive will ensure its longevity at everyone's expense. Open, democratic societies have an unpatched memory leak that requires a hard reboot every so often.

I completely understand, and while a local engine would be teriffic, the only one I can think of on Linux with any potential is sphinx. Unfortunately, (last time I checked) it's not in an easily useable state, yet. I should check out the pace of development on that project. Of course, I would be delighted to be proven wrong on this project.

And while this in no way is intended to dispell any privacy concerns, or justify any possible motives of the companies providing online deciphering, I am guessing this is exactly whar Siri does on Apple products. Also, as Palaver is currently targeting Ubuntu (while still being agnostic enough to easily work on any distro), It needs to have a small footprint considering the push at Ubuntu towards mobile devices. Unfortunately, mobile devices do not lend to local storage of the many samples in multiple languages required to perform the deciphering.

Cheers.

[EDIT] While Palaver is a wrapper around an online service, the wrapper is limited to sending the voice sample and getting a string of text back, the application performs the task of deciphering the meaning of the text and taking the appropriate action based on local dictionaries and plugins.[/EDIT]

Last edited by Padfoot (2013-03-23 22:21:13)

Aegidius · 2013-03-24 17:59:25

I installed it and I associated the F12 key to /opt/Palaver/hotkey. When I try it with Firefox it says:

Error
'Firefox'
is not a recognized command

What is the problem?

Padfoot · 2013-03-24 20:54:27

Have a look at the docs (/opt/Palaver/Docs) for setting up a dictionary of commands. It is very easy to do. Also, have a look at the default dictionary for a list of inbuilt commands.
Bear in mind that some plugins are app specific (have a look at the scripts in /opt/Palaver/Reognition/bin), file browser for example is designed for nautilus.
The docs for creating your own plugins are also very easy to follow.

Cheers.

[EDIT]Try open firefox[EDIT]

Last edited by Padfoot (2013-03-24 20:56:34)

Aegidius · 2013-03-24 21:02:54

I cannot find the ~/.USR.d/personal.dic file.

Aegidius · 2013-03-24 21:09:54

I try with open Firefox but I get the same error message.

Padfoot · 2013-03-24 21:29:35

Aegidius wrote:

I cannot find the ~/.USR.d/personal.dic file.

Create it. Then you can add your own commands as per the docs.

Aegidius · 2013-03-26 11:44:02

I created it and filled up with this example.

how are you computer
say "I am fine thank you"

It recognize the speech but it gives my the same error: is not a recognized command.

Arch Linux

#1 2013-03-23 05:30:48

Palaver speech recognition app packaged for Arch

#2 2013-03-23 06:11:05

Re: Palaver speech recognition app packaged for Arch

#3 2013-03-23 22:14:57

Re: Palaver speech recognition app packaged for Arch

#4 2013-03-24 17:59:25

Re: Palaver speech recognition app packaged for Arch

#5 2013-03-24 20:54:27

Re: Palaver speech recognition app packaged for Arch

#6 2013-03-24 21:02:54

Re: Palaver speech recognition app packaged for Arch

#7 2013-03-24 21:09:54

Re: Palaver speech recognition app packaged for Arch

#8 2013-03-24 21:29:35

Re: Palaver speech recognition app packaged for Arch

#9 2013-03-26 11:44:02

Re: Palaver speech recognition app packaged for Arch

Board footer