Signal / Sound processing / Voice recognition -- where to start?

Factual Questions
1

I know a little about image processing/computer vision and some basic pattern classification, but I’ve never done any work with signal or sound processing (don’t ask me how that happened). I’d like to study signal processing, particularly as it pertains to voice recognition. Are there any good book recommendations?

2

Bump once more, then I’ll let it die… sniff

3

NOOOOOOOOOOOOOOO!

[grabs paddles] CLEAR! [zap]

I do a fair amount of digital signal processing, dre2xl, but most of it is real-time image crunching with FPGAs, and the audio DSP experience that I do have is more in the compression area rather than voice recognition (and a few years old as well).

However, every well-intentioned GQ deserves an answer, so… I e-mailed a friend who does work more in the voicerec field. He recommended that you look at two books (the first of which is one that I read myself a few years ago, and was going to recommend to you if I had not heard back from him):

  1. Fundamentals of Speech Recognition, by Rabiner and Juang. It’s old (1993), but is still a classic and an excellent introductory book. It’s expensive ($98 new), and my friend recommended that you borrow it rather than buy it (if possible). The basic ideas are still valid, but some of the implementations (eg Linear Predictive Coding) have been superceded by more modern techniques.

  2. Statistical Methods for Speech Recognition (Language, Speech, and Communication) by Jelinek. Much more mathematical than 1), and incomprehensible if you haven’t already read an introductory text. Still, you’ll need something like this if you want to get serious in the field.

Once you’ve gone through these, you’ll presumably have some idea as to whether you’re going to be interested more in the PC/Mac platforms or dealing with dedicated DSPs. In either case, the really state-of-the-art stuff isn’t going to be found in a book (sound familiar to anyone else in the tech field?).

Hope this helps, and perhaps bumps your question into the sightlines of a Doper with more personal speech-recognition experience than I.

4

SMOOCH Thank you ever sooooo much AB! That’s EXACTLY what I was looking for. I knew the books were going to be in the textbook-cost range, so no worries. I was just looking for something to get me started. :slight_smile:

sulks impatiently for May 7th payday to arrive