Advice sought to replace a voice recognition system running on a Windows Shista PC?

OVERVIEW:
What's entailed to use Linux to access the audio input port and DSP
capabilities on the beagleboard?

How can assembly use an audio input port on the Beagleboard to do data-
acquisition?

How can assembly make DSP calls to the DSP chip on the Beagleboard?

DETAILS:
This is just a simple proof of concept project to recognize a dozen or
so pre-recorded and averaged Power Spectral Densities (PSDs).

I have written some truly rudimentary Digital Signal Processing (DSP)
algorithms using the audio input port to template match a pre-recorded
audio signal with a psuedo-realtime audio signal. I currently have
recorded 3s of mono audio data at 44.1 KHz. I calculate the PSD of 3s
of the incoming signal after level detecting and match it to the pre-
recorded template.

todd.gardner@ecafa.com said the following on 04/26/2009 05:29 AM:

OVERVIEW:
What's entailed to use Linux to access the audio input port and DSP
capabilities on the beagleboard?
  

Check out ALSA[2]. Beagleboard's Audio is supported using ALSA drivers.

How can assembly use an audio input port on the Beagleboard to do data-
acquisition?
  

get a microphone and plug it into the audio microphone jack of beagleboard

How can assembly make DSP calls to the DSP chip on the Beagleboard?
  

you can use dsplink or dspbridge..

DETAILS:
This is just a simple proof of concept project to recognize a dozen or
so pre-recorded and averaged Power Spectral Densities (PSDs).

I have written some truly rudimentary Digital Signal Processing (DSP)
algorithms using the audio input port to template match a pre-recorded
audio signal with a psuedo-realtime audio signal. I currently have
recorded 3s of mono audio data at 44.1 KHz. I calculate the PSD of 3s
of the incoming signal after level detecting and match it to the pre-
recorded template.
  

Does sound interesting.. Though, I wonder if sphinx2[1] could be used on
the dsp side..

Regards,
Nishanth Menon

Ref:
[1] http://cmusphinx.sourceforge.net/html/cmusphinx.php
[2] http://www.alsa-project.org/main/index.php/Main_Page

todd.gard...@ecafa.com said the following on 04/26/2009 05:29 AM:> OVERVIEW:
> What's entailed to use Linux to access the audio input port and DSP
> capabilities on the beagleboard?

Check out ALSA[2]. Beagleboard's Audio is supported using ALSA drivers.

Thank you

How can assembly use an audio input port on the Beagleboard to do data-
> acquisition?

get a microphone and plug it into the audio microphone jack of beagleboard

Where do I research what assembly language calls can I make to read
the audio port?

How can assembly make DSP calls to the DSP chip on the Beagleboard?

you can use dsplink or dspbridge..> DETAILS:
> This is just a simple proof of concept project to recognize a dozen or
> so pre-recorded and averaged Power Spectral Densities (PSDs).

> I have written some truly rudimentary Digital Signal Processing (DSP)
> algorithms using the audio input port to template match a pre-recorded
> audio signal with a psuedo-realtime audio signal. I currently have
> recorded 3s of mono audio data at 44.1 KHz. I calculate the PSD of 3s
> of the incoming signal after level detecting and match it to the pre-
> recorded template.

Does sound interesting.. Though, I wonder if sphinx2[1] could be used on
the dsp side..

I think that Sphinx is way too powerful for my simple DSP but THANKS!

There is no assembly language calls. it is a bit too complex for that.
you need to make alsa calls for it from an application.
Regards,
Nishanth Menon

I tried pocketsphinx on Beagle board (B5) on the top of Angstrom.
pocketsphinx_continuous takes input from audio input(need to connect to audio out of PC not headset mic) of BB and displays spoken text(which is played of PC) on terminal.

i am not exactly clear about you requirement but if you think i could be helpful to you then please let me know.

-Komalsingh

DISCLAIMER:

This email may contain confidential information and is intended only for the use of the specific individual(s) to which it is addressed. If you are not the intended recipient of this email, you are hereby notified that any unauthorized use, dissemination or copying of this email or the information contained in it or attached to it is strictly prohibited. If you received this message in error, please immediately notify the sender at Infotech or Mail.Admin@infotech-enterprises.com and delete the original message.

[snip]

Hello Komalsingh,

I only want to control a dozen (or so) different digital inputs by
voice. I have read that sphinx is for the Linux expert. That I am
NOT but I am motivated. I guess on good solution would be Sphinx
running on Angstrom but I have little idea where tostart being a
moderate newbie with Linux and a complete newbie with the BB.

Your success is encoraging.

My project is currently working on Shista PC in but I want the
portability of the BB. Currently I aquire at 44.1 KHz and then do
some DSP to see if thee is a "match" the incomming voice to one of a
dozen pre-recorded templates. It's rather crude but funtional and
greatly improvable.

I don't really know where to start but I guess the first step is to
buy a BB. Then install Anstrom, then Spinx and voila?

Thanks for the positive note!

Todd

Hello Todd!

Could you help us by making a "Wiki Page" or list of steps for
enabling audio input (using mic); plus a simple example for processing
the input using DSP?

Thanks a lot!

Dio
a complete noob