Am I confused about wavetable

[aciddose]

Sigh. Anything having to do with digitally sampled audio is, by definition, interpolated.

interpolation involves finding a value which approximates a value expected to exist between two points.

"In the mathematical field of numerical analysis, interpolation is a method of constructing new data points within the range of a discrete set of known data points."

when you sample a signal, that signal is filtered before hand and then sampled discretely. no interpolation takes place.

can you clarify what you're talking about?

[whyterabbyt]

Which one of those papers referenced defined it first? The paper itself appears to be well over ten years older than Mr. Palm's research and implementation. The earliest references are contemporaneous with Wolfgang Palm's period of invention but none of the titles seem to reference the subject directly. Although the title, "A Digital Signal Processing Approach to Interpolation", seems like it references a paper that could come close to addressing our topic, at least in theory, when one reads the abstract the paper's relation to the subject at hand seems less so. It seems related more to the Nyquist theorem and early research behind solving the problems related to digital audio sampling as a whole.

There's a fair bit of discussion of that behind the wiki page, it appears. Interesting point raised there that Palm would appear to have originally used the german word "Wellensatz" ie wave set, not anything translatable as 'wave table'. There's also a reference to Waldorf later describing the method as "dynamic spectral wavetable synthesis" which makes it sounds like they presented it as an expansion or special case of wavetable synthesis.
The same discussion makes references to Mathews, 1969, and an article in CMJ Vol 1/1 in 1977. Unfortunately the scanned CMJ archives only go back to Vol 1/3, so I dont have that.

There is however an article from the September 79 issue of CMJ I found called 'The 4C Machine' describing the latest incarnation in a set of digital audio processing systems built for IRCAM. It specifically covers the parts of the hardware architecture designed for handling wavetables (which are described as adder/wavetable units). Other references (Moorer in CMJ 5/1) make references to wavetables in Music-V, indicating that the term was in longstanding use.

But then since you've displayed an intrinsic suspicion of the glory-seeking 'academics' who clearly seek to hog every last bit of the limelight for themselves, that's probably enough. After all, what did Robert Bristow Johnson ever actually do for digital audio himself? Or that Julius O Smith guy who says the same thing.

[whyterabbyt]

Sigh. Anything having to do with digitally sampled audio is, by definition, interpolated.

Its really not, you know. And wavetables dont necessarily have to have anything to do with being 'digitally sampled audio' in the first place.

But feel free to explain to me what's is actually being interpolated when you, say, change the clock speed on a digital circuit which is doing absolutely nothing more than and endless loop of incrementing a pointer into a table of values, and reading out the value from the table.

Because me and my chip can't quite figure it out. Where does the interpolation occur? In the increment, or the read?

[optofonik]

These seem to address the issue. Most are beyond my understanding so you'll probably get more out of it than I could.


https://wiki.brown.edu/confluence/downl ... 7436104000

http://marksmannet.com/RobertMarks/REPR ... Theory.pdf

http://www.aes.org/e-lib/browse.cfm?elib=7806

http://www.aes.org/e-lib/browse.cfm?elib=6334

http://www.ijar.lit.az/pdf/10/2011%2810-111%29.pdf

http://www.mcm.edu/mathdept/lynn_blair/thesis.pdf

http://www.scribd.com/doc/20203269/Inte ... audio-DACs

http://www.earlevel.com/main/1996/10/19/oversampling/

[aciddose]

none of this has anything to do with interpolation though. while you "can" use interpolation, it isn't required.

what i was asking though was why you're bringing this up in the first place? how does it lend anything to the discussion? isn't this completely irrelevant?

this is like me arguing about LED vs. LCD displays being confusing for consumers as LCD with LED backlights are frequently referred to as LED, when they are in fact LCD.

then i start ranting about how you MUST use green phosphors in any display!

(forget about the fact that this refers to a specific technology or method in addition to the fact that there is no justification for saying you "must" use green ones in the first place.)

[whyterabbyt]

But I could also say it does "Transwave" like my old Ensoniq, or ScanWave, or MorphWave, or DWA ( Dynamic Wavetable Alchemy ), or something worse, or better.

Thats actually an interesting point. Different manufacturers had different names for the same thing. It also gets called wavesequencing, and even wavestacking.

If the established term was 'wavetable synthesis' why didn't they use it? Was 'wavetable synthesis' as used by PPG a marketing term rather than a definition, as we still see today with all sorts of different acronyms and terms for different methods of 'virtual analog' ?

[aciddose]

"virtual analog" ?

don't you mean linear arithmetic synthesizers? after all, that name was around first!

[optofonik]

There's a fair bit of discussion of that behind the wiki page, it appears. Interesting point raised there that Palm would appear to have originally used the german word "Wellensatz" ie wave set, not anything translatable as 'wave table'. There's also a reference to Waldorf later describing the method as "dynamic spectral wavetable synthesis" which makes it sounds like they presented it as an expansion or special case of wavetable synthesis.
The same discussion makes references to Mathews, 1969, and an article in CMJ Vol 1/1 in 1977. Unfortunately the scanned CMJ archives only go back to Vol 1/3, so I dont have that.

There is however an article from the September 79 issue of CMJ I found called 'The 4C Machine' describing the latest incarnation in a set of digital audio processing systems built for IRCAM. It specifically covers the parts of the hardware architecture designed for handling wavetables (which are described as adder/wavetable units). Other references (Moorer in CMJ 5/1) make references to wavetables in Music-V, indicating that the term was in longstanding use.

So what we're left with is a type of synthesis that was being researched and developed on several different fronts around the same time. Really, it's all academic at this point.

Excellent debate, though.

BTW, I am not at all suspicious of the venerable, highly respected, Computer Music Journal. I think I read a couple of issues when I was still a student in the early 80's and, after reading a couple of essays, subscribed to Mix magazine.

[Shabdahbriah]

But I could also say it does "Transwave" like my old Ensoniq, or ScanWave, or MorphWave, or DWA ( Dynamic Wavetable Alchemy ), or something worse, or better.

Thats actually an interesting point. Different manufacturers had different names for the same thing. It also gets called wavesequencing, and even wavestacking.

If the established term was 'wavetable synthesis' why didn't they use it? Was 'wavetable synthesis' as used by PPG a marketing term rather than a definition, as we still see today with all sorts of different acronyms and terms for different methods of 'virtual analog' ?

Indeed.

... Having a plethora of options, with many being JUST as plausibly 'accurate' a definition as any other, 'they' may have written down a bunch of possible names (for it) on little pieces of paper, and put them all in a hat, and pulled one out as the "winner".

OR - one of their 'reps' at NAMM, may have blurted-out something he thought was 'clever", and the maddening throngs ran with it.

Kidding aside:

Somewhere, I have an interview with 'Palm' where he gives numerous insights about it, and uses several different terms to explain the process, which gave the impression that "wavetable synthesis" was not wholely accurate nor his prefered term, but became the common term through "use". That would be "in house", use.

naturally, I can't find that reference...

[himalaya]

But I could also say it does "Transwave" like my old Ensoniq, or ScanWave, or MorphWave, or DWA ( Dynamic Wavetable Alchemy ), or something worse, or better.

Thats actually an interesting point. Different manufacturers had different names for the same thing. It also gets called wavesequencing, and even wavestacking.

If the established term was 'wavetable synthesis' why didn't they use it? Was 'wavetable synthesis' as used by PPG a marketing term rather than a definition, as we still see today with all sorts of different acronyms and terms for different methods of 'virtual analog' ?

and so there are different names to describe that process of modulating the wavetable and sweeping through all of its waves. That is granted and not the issue.

The term Wavetable Synthesis points to those methods, to the fact that there exists a mechanism to scan/sweep/move/etc through a wavetable. All of it, part of it.

[himalaya]

Somewhere, I have an interview with 'Palm' where he gives numerous insights about it, and uses several different terms to explain the process, which gave the impression that "wavetable synthesis" was not wholely accurate nor his prefered term, but became the common term through "use". That would be "in house", use.

That term seems very descriptive, I think. It tells us that using the modulation process we can synthesize new sounds by sweeping through all (or selection) of the waves contained in a wavetable. Without the 'scanning' bit what is left? Simple playback of a single cycle wave? Then we are back in the 'subtractive world' for shaping the tone.

[Shabdahbriah]

Somewhere, I have an interview with 'Palm' where he gives numerous insights about it, and uses several different terms to explain the process, which gave the impression that "wavetable synthesis" was not wholely accurate nor his prefered term, but became the common term through "use". That would be "in house", use.

That term seems very descriptive, I think. It tells us that using the modulation process we can synthesize new sounds by sweeping through all (or selection) of the waves contained in a wavetable. Without the 'scanning' bit what is left? Simple playback of a single cycle wave? Then we are back in the 'subtractive world' for shaping the tone.

I do as well. Assuming we know there is such a thing as a "modulation process", and are willing to explore it. The association(s) with "additive" synthesis is rather telling, me thinks.

[ENV1]

a more accurate interpretation of the term is that the edges of the wave are at 90 degree angles, making the angles "square".

Well 'square' certainly isnt entirely wrong. In a way it certainly makes sense, its just that since the form in question is neither really a square, let alone a wave, that the term Symmetric Linear Pulse would have been a better choice since it, unlike the term 'square wave', does not contain any nonsensical elements.



But that train has departed a long time ago.

The terms are what they are now, so what can you do...

[aciddose]

no, it is exactly square in terms of the angles involved. the only issue is that it leaves a lot open to question regarding more complex waveforms. as i said, "semi-square" waveforms.

it hasn't departed for me. in fact you don't need to use the term "linear" at all. "pulse" is the widely agreed upon generic term and when you're in the context of subtractive synthesis we generally assume a very narrow definition which works out just fine.

if someone were to call a sinc im"pulse" a "pulse" wave, that would also be correct as it's just a variation of a pulse with specific parameters.

what people are complaining about here is that they've gone and used a generic term and now they're upset that something other than the very specific sub-set they've used the term to describe can actually be just as well described by exactly the same generic term.

that's just silly.

when we use the term "fruit" in a supermarket we assume bananas, apples, oranges, pears and all these typical northern hemisphere fruits.

that doesn't make it wrong to sell less common wild fruits in a supermarket and put them in the fruit section.

then we can take a trip to the southern hemisphere, whether africa or south-america or persia, greece and india, east-asia or australia and boom, we're hit with a whole new slew of "fruits" without including a single one of the ones we're used to.

the common term "fruit" for people from different areas means a specific set of fruit. the technical term "fruit" includes tomatos.

[optofonik]

that's just silly.

&


For your continued enjoyment, the full, "cough cough", Monty...


&