Dear Dr. Mitra, I've been hoping to possibly run into you sometime at a professional conference, perhaps the IEEE WASPAA (a.k.a. Mohonk), but I haven't been to that conference since 2001 and I haven't yet seen you at an Audio Engineering Convention (and, admittedly, I'm not in IEEE). I am now working for a company in Calabasas (Music Mastermind) so I hope that someday I can meet you while I am local to USC. I was sorta exploring the dsp.stackexchange.com site and I came upon a short monograph attributed to you called "Applications of Digital Signal Processing". I don't know if this is part of an existing book in print, or is on the way to print. I found the link to at the University of Vermont site, which is quite coincidental since Burlington Vermont is my permanent address and the prof listing it (Gagan Mirchandani) is someone I am acquainted with. The work appears useful and valuable to me, and I immediately noticed Sections 5 and 6, which are about what I do for a living and where I also have published. In fact, one old paper I have written (that later resulted in the Audio EQ Cookbook) called "The Equivalence of Various Methods of Computing Biquad Coefficients for Audio Parametric Equalizers" (long title) came from a paper by Dana Massie that came from your Regalia and Mitra paper about APFs and the bell-shaped parametric EQ. I have also written about wavetable synthesis ("Wavetable synthesis 101, a fundamental perspective") which is what I want to write you about. I don't believe previous authors, nor even the seminal authors themselves, would identify or categorize the Karplus-Strong algorithm for generating harmonically-rich musical tones as "wavetable synthesis". there is a delay-line with feedback, and so also do reverb algorithms have delay lines and feedback, but they are not "wavetables". Wavetable Synthesis is more about what the Palm PPG ( http://wolfgangpalm.com/ ) and Waldorf software synthesizers have been doing. Another, more local example is the Prophet VS synthesizer (Dave Smith Instruments, also the creator of MIDI), There has been, nearly 2 decades ago, a confusion about the meaning of "wavetable synthesis" because of Creative Labs and the Soundblaster sound card that label their sample-playback-and-looping rendering of tones as "wavetable synthesis". It took some years and some editing of certain blogs and sites (including Wikipedia) to set the record straight that simple soundfile playback (even with sample interpolation and playback speed adjustment for changing pitch) with a single loop at the end is not what the original "wavetable synthesis" had been. It was used as a marketing term by Creative Labs. But neither is the Karplus-Strong technique wavetable synthesis. I would judge that Karplus-Strong more closely is categorized as a simple form of Physical Modeling Synthesis, and I think that Julius Smith would agree. Wavetable Synthesis is similar to the pedestrian sample-playback-and-looping technique, except that in wavetable synthesis, it is virtually always in looping mode. And the loops are always exactly one period or cycle of the quasiperiodic tone. Yes, there is at least one paper (by Andrew Horner and one of his students) combining Wavetable and an unrepeated attack waveform of the instrument, but in wavetable synthesis, the tone spectra morphs from one line spectrum to the next simply by cross-fading from one wavetable to the next. If you want, I could help you describe what is fundamentally involved with Wavetable Synthesis, how it relates to sample-playback-and-looping and also to Additive Synthesis (wavetable synthesis can be thought of as a way of performing additive synthesis of quasiperiodic tones where the inverse Fourier transform is done in advance and not at the real-time generation of the tone, the waveforms are pre-computed). Also, if you want, I would be happy to meet with you in person at your convenience (say at your office or even over food somewhere) since I am not local (I am lucky to have found a little place to live in Malibu, by the Backbone trail). Although I haven't written a monograph myself, I have participated in some other efforts including Principles of Digital Audio by Pohlmann and also that large compendium by Mark Khars and Karlheinz Brandenburg (Dana Massie had a chapter on "Wavetable Sampling Synthesis"). I can also be contacted by cell phone: 802/310-4096 if you would want. Best regards, Robert Bristow-Johnson -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
About "wavetable synthesis" as described in "Applications of Digital Signal Processing"
Started by ●July 28, 2013
Reply by ●July 29, 20132013-07-29
i just gotta make sure this doesn't go public... On 7/28/13 7:50 PM, Tim Wescott wrote:>...> > I thought it was very nice and polite. More polite than, for instance, > you ever are to us. >oh c'mon, "ever"? who here do i pick on? (besides the heathen heretical DFT periodicity deniers) oh, i guess that there are the we-must-count-from-one-otherwise-MATLAB-would-break crowd (not too many on comp.dsp but there were a couple, at least) then there was those troll-kinda guys (who were they? i can only remember E.Bob. but there were others). i don't pick on noobs. i don't think i do.> I wasn't sure at first if it was an open letter or a misdirect, but your > above response does at least clarify that point.it coulda been worse. but it's not the first time this happened. some designer or coder should be hurt in retribution for deciding to do it that way. death penalty might be a little harsh.> And in case you're wondering -- this is part of the reason that I prefer > having separate email and news clients.i just want my life to be simple. as simple as possible. i just want something that i can run on this mac and Google Groups became horribly bad and eternal september doesn't do a webpage-based service (but i'm grateful it's free and not icky in some way), it seemed that i had to get a client to do it, and the only one i found was Thunderbird. dunno why Apple's Mail does this dumb RSS that you can't get rid of, but it doesn't do USENET like all of the old email clients used to do. ya know, i used to know how product builds go (although i left it to someone else to be the keeper of the makefile) and how version control works, i used to be able to use the CodeWarrior and SourceSafe to build an app (or a code resource), how to back up, commit, check-in, "push", whatever you wanna call it. now it's just so much more complicated when what i wanna do is just the same as it used to be. that's like negative progress: i have to pay more to do the same, a declining payback. but isn't that opposite of the promise of modern computers (particularly the user-friendly revolution that sorta came with the Mac in 1984)? the computers exist to make my life easier, not the other way around. how come is it so much more complicated to do anything (even a simple graphic or something) now than it used to be? -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."