DSP theory

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
OK, for the benefit of anyone interested in sinking their teeth into the meat of digital audio signal processing, I thought I'd put together this list of references. It's certainly not a comprehensive list, and one can find many highly reviewed titles which I do not list here. I've seen and read many of these books, but I've chosen to list only those which I felt met some basic criteria:

- easy to read and understand
- accurate formulae and no conceptual errors (surprisingly, I've come across a number of books in which the author has made major conceptual errors, or obvious mathematical mistakes!) This can be very counterproductive for someone attempting to learn the material for the first time.
- concise, useful information without all the pointless 'exercises' and irrelevant fluff which can balloon a small text to preposterous length.
- contains hard-to-find or specialized information
- are very cheap/free and readily available

A cautionary note for those who havn't touched DSP theory before: I don't want to scare anyone off, but some of this material gets highly theoretical, and a university level mathematics education will come in very handy, and in some cases is essential. Many of these books contain good primers on the necessary math, but may still rely on a solid undergraduate level foundation of calculus, linear algebra and differential equations. Much of DSP theory stands on complex mathematics, and a conceptual understanding of the equations goes a long way. Also very useful is some basic signal theory. You should know what an impulse is (conceptually and mathematically), and what the impulse response is. Convolution is a good concept to know, and some knowledge of transfer functions doesn't hurt either.

That said, I encourage everyone to have a peek anyway because while the equations and some of the concepts may seem completely abstract, one can still gain a great deal from the explainations and diagrams etc... so, here we go:

1. "Understanding Digital Signal Processing" by Richard G Lyons. ISBN 0201634678
This is definitely my favorite book on DSP. No other DSP author I've read has the ability to explain difficult abstract topics as clearly, concisely, and correctly as Mr. Lyons. The book doesn't deal specifically with audio too much, but does include a very enlightening discourse on complex math as it relates to DSP. And to put the cherry on the cake, it's reasonably priced by comparison to other DSP texts. My #1 recommended read!

2. "Digital Signal Processing Using MATLAB" by Ingle & Proakis. ISBN 0534371744
This one's really helpful for testing filters in MATLAB. One thing I really like about digital audio is that most of the time, a simulation is exactly the same as the real thing, unlike a SPICE simulation of an analog circuit where the real-world implementation may contain some hidden surprises.

3. "Digital Audio Signal Processing" by Udo Zolzer. ISBN 0471972266
Ok, this one contains some highly theoretical stuff, but for those who already have a solid understanding of DSP, it's an invaluable resource of info you won't find anywhere else.

4. http://www.dspguide.com/ - there's a free book here (yes free, all 640-odd pages!). I havn't read it all, but from a brief browse, it looks ok. You can't go wrong with free...

OK, happy reading people!
oh yes, a word or two on reference #4:

If you havn't looked at DSP before, I suspect you'll start with this before dumping any money into books. Chapters 30, 32 and 33 are a good start on the math, and should complement chapters 5-7 and 13 well. You may want to skip the Fourier transform stuff initially, as it's not particularly applicable to audiophile applications (useful if you want to understand MP3 compression though).

The author of this book does not deal explicitly with FIR and IIR filters, though the filters he describes all belong to one or the other category (with the exception of the FFT techniques). I'm also a little disappointed to see some generalizations stated as absolutes. For instance, some of you will recognize "windowed sinc filters" as lowpass FIR filters. The author states that this class of filters has "poor performance in the time domain, including excessive ripple and overshoot in the step response", when in fact these filters usually (if properly designed) exhibit by far the best time, frequency and phase behaviour of any digital filter type. Anyway, the general concepts should all be there.

The Custom Filters chapter shows some of the interesting things that can be done with FIR techniques...
Re-use something else?


What if you took a unit which is relatively inexpensive these days -- such as the Sony TAE9000ES which not only has an in-situ programmable Sharc (ADSP-21065L, very likely most of the http://www.analog.com/techsupt/prod_briefs/dolby_dig.html board as well but for Analog Devices DAC's of course), but various DAC's, DTS, AC3 etc. decoding. It also has a built in equalizer which is set up so that it starts at about 100Hz which is kind of sad. Anyway, this would be a great way to start off modding if you could get into an inexpensive secondhand unit -- looks good, has good PSU, DAC's etc.

Sony link: http://www.sel.sony.com/SEL/consumer/ss5/generic/homeaudioes/espre-amplifiers/index.shtml

Check out this link as well http://home.online.no/~espen-b/ta-e9000es/index.html

And the info forum http://pub7.ezboard.com/faussiedvdandhtforumsonytae9000esinformationforum?page=1

You should be able to get one cheap (my guess $700 or so), especially if you get it without the remote (which eats batteries anyway).

You get 30 days evaluation software (for the SHARC alone) for free from Analog Devices ....

Is this a cool evalution board or what?

Yeah, this looks really promising! One thing I like about this unit is the possiblity of snagging the proprietary code (DTS, AC3, etc.) that you're not supposed to get... ;) Of course, it'd all be in assembler code without any comments, but that's still better than nothing. Their digital volume control scheme is rather nifty too...

Obviously, you'd need schematics or maybe a service manual before tackling any hardware mods, but from the looks of it, maybe just some software hacks are all that's necessary? This could be a really interesting way to go... the Stereophile stamp of approval (not that Stereophile is by any means the last word) is nice to see too.

The tricky bits will likely be figuring out the software interfaces Sony used with it's host controllers and display / user interface, and whatever other mysterious preipheral ICs are also included. This will likely have to be done by painstaking assembler code reverse engineering. Now if you can simply avoid tinkering with this part of the code, then so much the better. I'd be interested to see what size flash the SHARC chip is booting from, and also how their RS-232 interface is hooked up. With sufficient space in the flash, you could simply insert your own code in there, and if they piped the RS-232 straight into the JTAG port, that could be really handy for programming and debug too.

Thanks Petter, I'll keep my eyes peeled for more info!
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.