Wavetable Synthesis Question

[audiojunkie]

I world like to try to simulate, to a rough extent, the ideas behind the Roland D50, except with wavetable synthesis instead of LA synthesis. Using two oscillators, I hope to use one Wavetable oscillator for the attack portion sound, and one oscillator for the sustain sound.

At 2048 sequential frames, one gets somewhere around 40 milliseconds of audio in single cycles. Generally, an attack of a sample can be recognized at between 100-300 milliseconds.

My question is:

Is it possible to sweep slowly enough through the wavetable cycles for an attack sound of something, like a piano hit or a bell or something similar to be audibly recognizable? In other words, can there be enough audio material in 2048 frames to create a recognizable attack phase?

Put simply, has anyone ever successfully created a wavetable of the attack portion of an acoustic piano key being pressed? Can it be done successfully?

Thanks in advance!

[zerocrossing]

I think you’re barking up the wrong tree. Wavetable synths are so good at this point that you could just sample a piano note and turn it into a wavetable using Icarus or any number of synths with wavetable editors. No real need to do what you’re taking about.

Also, attacks are generally a lot of noise and inharmonic sound. Some wavetable synths actually let you bring little samples in to use for such things. I often will use FM wtih a very short decay to get something like that.

[audiojunkie]

I have yet to find a wavetable that has an attack from an acoustic instrument....such as a string plug, an initial blow from a wood wind, a brass attack, etc. That's why I had the question. All of the wavetables that I've found are just modulated sounds, which is great for the sustain phase. I don't doubt you, I am sincerely interested...do you happen to know of any URLs to wavetables like this, or to youtube (or other media) that I can hear these wavetables? I use Hive 2, and it doesn't import samples, or else this would be a non-issue. As I understand it, the standard wavetable nowadays contains about 2048 frames. Can enough single cycle waveforms be sequenced to achieve the whole initial sound of an acoustic instrument? I'm wanting a true acoustic instrument sound, not an FM noise attack.

Imagine that I want to have the sound of a blown digeradoo that morphs into an evolving sustain. Can a wavetable contain enough single cycle waveforms for this sound to be recognized if I sweep through the wavetable? This isn't sampling, remember. I've got a sampler that can do this, and I can easily layer my sampler and my Hive 2 with two tracks in my DAW and morph them from there, but I am trying to achieve this sound within a single patch.

[bluesawsq]

I have yet to find a wavetable that has an attack from an acoustic instrument....such as a string plug, an initial blow from a wood wind, a brass attack, etc. That's why I had the question.

Check 'Instruments' category in the free Echo Sound Works Core Wavetables collection.

[elassi]

... if I sweep through the wavetable?

... and especially imaging this didgeridoo played backward after a modulated period of time.

Therefore +1 for your interesting question. Nice impulse.

[zerocrossing]

I have yet to find a wavetable that has an attack from an acoustic instrument....such as a string plug, an initial blow from a wood wind, a brass attack, etc. That's why I had the question. All of the wavetables that I've found are just modulated sounds, which is great for the sustain phase. I don't doubt you, I am sincerely interested...do you happen to know of any URLs to wavetables like this, or to youtube (or other media) that I can hear these wavetables? I use Hive 2, and it doesn't import samples, or else this would be a non-issue. As I understand it, the standard wavetable nowadays contains about 2048 frames. Can enough single cycle waveforms be sequenced to achieve the whole initial sound of an acoustic instrument? I'm wanting a true acoustic instrument sound, not an FM noise attack.

Imagine that I want to have the sound of a blown digeradoo that morphs into an evolving sustain. Can a wavetable contain enough single cycle waveforms for this sound to be recognized if I sweep through the wavetable? This isn't sampling, remember. I've got a sampler that can do this, and I can easily layer my sampler and my Hive 2 with two tracks in my DAW and morph them from there, but I am trying to achieve this sound within a single patch.

I don’t intimately know Hive’s wavetable selection, but I bet there’s at least a few that have sections that should work well for you. If not, you should try something like Icarus and let it make a wavetable from any attack phase of any sound you want. In all honesty, though,

It’s ultimately futile, though, as in practice, a decent multi-sample will give you better results.

[seafire]

You would be better off (imo) forgeting WTs and find a synth that has a wav osc and use that ( in one shot mode, no loop obviously) for the initial/attack stage, and the synth engine for the rest of the sound.

Isn't that how the D50 does it anyway?

[audiojunkie]

You would be better off (imo) forgeting WTs and find a synth that has a wav osc and use that ( in one shot mode, no loop obviously) for the initial/attack stage, and the synth engine for the rest of the sound.

Isn't that how the D50 does it anyway?

Yes, that's how the D50 does it as far as I understand. I have TAL-Sampler, which would work if I absolutely have to, but I was hoping to succeed with this idea. TAL-Sampler has three envelopes, but they are global envelopes, and not available on a per layer basis. I guess it is possible to use envelope 3 (the MOD envelope) to modulate the layer 2 volume. That would give me two layers with separate envelope controls.

[pocvecem]

Maybe you should take a closer look at granular synthesis. To me this always has been another "taste" of wavetables.
Experimenting with grain width and depth you can make the sample as smooth a you like to. Then move through the sample by controller and/or modulators.

To me that's the only reason why I keep Pigments because I can also add fm and ringmod to granular though TAL Sampler sounds so much better.

[bermudagold]

I have yet to find a wavetable that has an attack from an acoustic instrument....such as a string plug, an initial blow from a wood wind, a brass attack, etc.
Imagine that I want to have the sound of a blown digeradoo that morphs into an evolving sustain. I am trying to achieve this sound within a single patch.

cant karplus-strong with wavetable get you close?
check out yuli yolo presets for vital

[BertKoor]

I world like to try to simulate, to a rough extent, the ideas behind the Roland D50, except with wavetable synthesis instead of LA synthesis. [...]

At 2048 sequential frames, one gets somewhere around 40 milliseconds of audio in single cycles. Generally, an attack of a sample can be recognized at between 100-300 milliseconds.

regarding the D50, I found this piece of information:

33k sample rate
[...]
Most samples are either 2048 (~60ms), 4096 (~120ms) or 8192 (~240ms) sampling points long. The longest sample is the noise sample, 16384 sampling points long. All sample lengths are multiples of 2048, which makes the addressing hardware easier to implement.