Discrete-time Signal Processing (Prentice Hall International Editions) [Englisch] [Taschenbuch]

I have both the DSP version and the 1st edition of DTSP, and used them for coursework. From my personal recollection: I tried to take an introductory graduate-level DSP course twice, once with the old DSP text (but did not get to finish), and the second time completing the course from another college with the DTSP text. DSP, in spite of its introductory and pioneering nature, turned out to be the more readable and better organized textbook among the two. It is shorter by many pages, less verbose, less "heavy and circular in arguments", and quicker to come to the point, not to mention being clearer, more lucid, and well-illustrated with good examples and diagrams without overwhelming the reader. The beginning chapters are well written with introduction to the applications of Complex Variables in the context of DSP as an EE subject. In short, DSP by Oppenheim and Schafer is the less ambitious book of the two, but really accomplishes much more by doing less and focusing on the essential concepts underlying DSP.

I can see why and where many beginning EE Signal Processing students tend to get frustrated. DTSP, 1st edition, is actually a rewritten version of DSP, albeit meant to be a major extension (or expansion, if you wish) with more applications appended and "heavier expositions". (This reminds me of many other books by MIT EE faculty on the market.) I find the most frustrating part of DTSP is the chapters on transcending from the Discrete Fourier Transform (DFT) to Fast Fourier Transform (FFT). Why? All you have to do is look at the way the authors present the DFT, then get to understand the truncated DFT, and try to follow the steps that supposedly get one toward FFT's - not at all clear or well-connected conceptually. It is even more obvious when one tries to do the moderately difficult and somewhat tedious exercises at the back of each chapter. While it is true that DTSP appears superficially "precise and rigorous", I find it a bit convoluted and too ambitious in its treatment of digital signal processing topics, indicative of a desire to reflect trappings of the latest applications by demonstrating the power of the theory presented. Why not just return to the fundamentals of DSP and limit the topics to a manageable level, thereby retaining a high degree of visibility for the more average EE student. (Maybe DTSP was meant to be taught by the authors themselves, or by their Ph.D. students. That way, the obscure points, discussions and topics in the textbook could be explained away in live classroom lectures via student interactions. Right?) In spite of all this, I think DTSP has intrinsic value in EE Signal Processing, being that it retained much of the key concepts in the older DSP text with some original flavor.

Suggestion to authors: Write another book on DSP, but limit the book to concepts and some elementary applications, just like the 1970's original, or better still edit and update the original to be published as an inexpensive classic by, say, Dover. Rewrite DTSP as the follow-on textbook. How about it? I really miss the lucid style of the more original DSP textbook by the authors. And I mean it, seriously!

Specific topics covered so far in the course, which are mentioned, but NOT covered in the text: Closed form solution of Difference Equations, complex convolution theorem, unilateral Z-transform and their use in the solution of Difference Equations with non-zero initial conditions, and Contour Integration.

I can see where some people use this as a reference, because there are many brief, accurate statements. But if you're trying to learn this material for the first time, look elsewhere.