DSP for computer engineers

Question

DSP is more kindred with electrical engineering than with computer science, how ever there are topics of DSP like MP3 encoding, JPEG compression and sound analysis can be understood by a computer engineer (with a previous strong math background in Z transform and Fourier series).

For a BS in computer science that only have very basic knowledge in electronic, is there a way to name that kind of specialization in DSP? I.e. someone who is looking to create DSP algorithms that can be run from a computer but not looking to implement them in a custom piece of hardware.

My question is mostly related for a computer science student who wants to apply for a master degree in an international university (no knowledge in analogic circuits, hardware description languages or microprocessors)

Would applied math, be a better approach??

Answer 1

DSP is a domain mixed between EE/CE and Math/CS.
Hardware must be built, but also software written.

Some CS programs do touch DSP at the software level. An applied math program would touch even more the algorithmic side, since pretty much all DSP techniques are based on certain mathematical areas such as Fourier analysis, Complex analysis and Numerical analysis. An EE program might have a specialisation on DSP, covering both the hardware and software sides but at a more "applied", less "proof mathematical" level, than what mathematics programs do.

The field is that wide though that it might also be better to focus on only one thing and do it well, rather than think in a "jack of all trades" way, i.e. too broadly. That is, if your interest is in software and mathematics, then focus on those. If your interest is in hardware, then focus on that. Applied math is sort of a middle ground. It's both theoretical, but it will lack practice such as electronics design, even if it'd give you conceptual tools for understanding electromagnetism and digital logic.

Answer 2

I'm not sure how much math or dsp you already know but I can give my 2 cents. :)

I think there are many of us interested in signal processing but find the initial entry barrier quite difficult. Most books or information on the topic dive straight into complex mathematics and makes it difficult to focus on a starting point. I'm interested in audio dsp and am preparing for a Msc on the subject.

This book is the best I (and many others) have found who try to access dsp with a computer science or software engineering degree.

The Scientist and Engineer's Guide to Digital Signal Processing By Steven W. Smith, Ph.D.

You can get the print book on amazon, but the whole book is online as a PDF, with downloadable chapters. It explains things practically with an applied approach. The topic is still complex, but without the mysterious mathematics.

Along with that you will need to learn Trignometry, Calculus and Complex Analysis. This is a great resource to get a quick overview of the math and core concepts of dsp.

Digital Signal Processing: Instant Access

From what I have experiened from audio coding and audio dsp, is that the dsp is tough, but it's also tricky to get the underlying system working. Deploying C code to custom hardware isn't that difficult, unless you want to add chip specific assembly code.