projects in DSP

"Rohit Pydimukkala" <rohit@ee.iitm.ac.in> wrote in message
news:1bf94b60.0401281943.9f658d3@posting.google.com...
> I just started out with DSP and I am quite excited with my learning.
> Now that I am interested in doing some small projects in DSP. Can
> anyone let me know where I can find DSP projects on internet or
> anywhere?

Rohit,

You didn't say if you want to do things with a DSP chip or just doing things
using DSP.  So, this is not about using a DSP chip:
Assuming that you have a PC and a sound card, and if you're interested, you
might develop a program that implements a sonogram.  The idea would be that
you could play real time sounds into it from a microphone, other analog
source or from a file.

It would involve spectral analysis, selection of time frames, overlap
processing, possible speed settings, generating a display, etc.

Then you could experiment with a variety of filters between the input and
the spectral analysis.

see:  http://www.dfki.de/~clauer/programming.html and other Google hits for
ideas.

At any rate, the project would be nicely divided into pieces and there would
be some real-world aspects in implementation.    For example, you could
implement a spectrum analyzer a bit easier than the sonogram.  If there's
too much real-world implementation work (such as lining up the data for the
display and setting the display parameters) then you may learn something
about where your interests really are and *are not*.

For the spectrum analyzer you may want to low-pass filter each output
position / "bin" so the display changes gracefully.  Or, you may find that
the analysis bandwidth does that anyway.  Overlapping input segments before
analysis processing might have a similar effect.  So there are lots of
things that can be experimented with without breaking the fundamental
operation.

For the sonogram, you will need to integrate the analysis window length with
the scan rate and will need to create a scan type of display - which might
be a horizontal waterfall with frequency on the vertical axis and time on
the horizontal axis and the most recent time on the left edge.

Fred

>>>>> "Rohit" == Rohit Pydimukkala <rohit@ee.iitm.ac.in> writes:

    Rohit> I just started out with DSP and I am quite excited with my learning.
    Rohit> Now that I am interested in doing some small projects in DSP. Can
    Rohit> anyone let me know where I can find DSP projects on internet or
    Rohit> anywhere?

I think a really good project is a DPCM vocoder.  The basic idea is
fairly simple, and with an audio card, you can record whatever you
like, run it through your DPCM vocoder, and listen to the result.

And if plain DPCM is too simple, you can add a fixed LPC filter to
it.  Or make the quantizer adaptive.  Still too simple?  Make the LPC
filter adaptive and the quantizer adaptive.   Still too simple?  Start
quantizing the filter coefficients and transmitting them to the
receiver/decoder.

And you get to listen to see how these things affect the quality of
the result.  That's way better than looking at some SNR value or a
plot of the audio signal or difference signal on some graph. :-)

And you learn a lot and have fun at the same time.  What could be
better? 

Ray

Yes, perhaps it does. Try the fish idea?

"Rohit Pydimukkala" <rohit@ee.iitm.ac.in> wrote in message
news:1bf94b60.0401300953.6a3556f2@posting.google.com...
> "Airy R. Bean" <airy.r.bean@lycos.co.uk> wrote in message
news:<bvanj9$pjkgd$1@ID-217727.news.uni-berlin.de>...
> > Now that computing power is so cheap, why not make your
> > own NMR/CAT scanner. Then you can look at the internals
> > of your own body, and save a fortune at the doctor's.
> >  > I just started out with DSP and I am quite excited with my learning.
> > > Now that I am interested in doing some small projects in DSP. Can
> > > anyone let me know where I can find DSP projects on internet or
> > > anywhere?
> Hey, I am actually looking to do something more of comp based and with
> few sample DSP chips I can manage. I cant afford any expensive
> hardware. NMR/CAT scanner does make a very good project but, I guess
> it needs some expensive hardware.

Jerry Avins wrote:

(snip regarding NMR and signal processing)

> That's truly inspired. I tell you what: you provide the scanning
> hardware and I'll do the software. Then we can both use it.

> I'm really curious. When a someone asks for serious help, even in a way
> that may seem a little naive, why are some people moved to tease? Some
> people offer a kid candy, then snatch it away when it's reached for.
> You might be able to explain the pleasure they get in that.

I believe it is most often done when people ask what seems to be
a homework question, though not stating that fact.

If people state that it is homework, and show that they have actually
done some work on the problem, they usually get good answers.

It is a little less obvious in this case, but it does seem that
the person hadn't really thought about the problem much.

-- glen

Rohit Pydimukkala wrote:

> "Airy R. Bean" <airy.r.bean@lycos.co.uk> wrote in message news:<bvanj9$pjkgd$1@ID-217727.news.uni-berlin.de>...
> 
>>Now that computing power is so cheap, why not make your
>>own NMR/CAT scanner. Then you can look at the internals
>>of your own body, and save a fortune at the doctor's.

(snip)

> Hey, I am actually looking to do something more of comp based and with
> few sample DSP chips I can manage. I cant afford any expensive
> hardware. NMR/CAT scanner does make a very good project but, I guess
> it needs some expensive hardware.

Well, just to develop the software you can generate some simulated
data, add a little noise for good measure, and then transform that.

You should be able to do it on ordinary PCs, unless you actually
want DSP (the P is processor not processing) experience.

-- glen

"Airy R. Bean" <airy.r.bean@lycos.co.uk> wrote in message
news:bvanj9$pjkgd$1@ID-217727.news.uni-berlin.de...
> Now that computing power is so cheap, why not make your
> own NMR/CAT scanner. Then you can look at the internals
> of your own body, and save a fortune at the doctor's.
>

Would that be NMR based on YOUR famous Nuclear Emissions?

GW

"Airy R. Bean" <airy.r.bean@lycos.co.uk> wrote in message
news:bvdcks$rfvo7$1@ID-217727.news.uni-berlin.de...
> If not NMR then try sonic scans.
>
> Here's another related idea....
>
> Use sonic scans with an aperture just
> large enough to put a fish through.
>
> Detect those remaining bones in a supposedly
> filleted fish! That'd be useful! A lot cheaper
> than suing the food supplier.
>
> What is it about modern students that they
> cannot think for themselves? However, having said
> that ISTR that all the 3rd year projects had to
> be selected from a list, so perhaps we didn't think
> either.
>

But not everyone actually finished their third year, did they "Airy"?

GW