i know how to end aliasing

[Architeuthis]

solving aliasing is the start of audio synthesis. whatever we have now was the beta test. it's an extremely important endeavor that will underpin the future of technology

the future: products will be based on intelligent chaotic algorithms that create sound and vibration in 2d 3d (4d?) space, whoever has the mind to come up with novel algorithms wins. eventually ai will discover novel chaotic algorithms that will solve human problems, trained on the algorithms we as humans come up with now. it is the last human problem to solve.

this will affect every technology we have: medicine, ai, entertainment, science, physics, biology.

[9b0]

bandlimiting is impossible to calculate [perfectly] because it requires a chaotic equation that you can't know the answer to until it occurs.

OMG. based on this sentence, i know it does not make sense to reply to you ever, so this is my last post in this topic.

as i wrote before, BLIT is an ancient technique, but it's not an approximation, like polynomial based methods.

here it is in action. this is not additive synthesis, this is a single oscillator (and is pretty fast too, takes fewer resources than a polyBlep). the output is a filtered version of a bandlimited impulse train. this is perfectly bandlimited, since the impulse trains are the result of driving a sinc function (sin(x)/x) with a ramp.


and you might also want to watch this:

[Architeuthis]

ill include that in the repository for yet another band limiting technique. and again, the goal is not perfect waveforms. it is to discover chaotic equations that approximate whatever sound i want and tuned to be alias-free.

the band limiting techniques included in my engines like BLEP, DSF, etc. are just for "i want all algorithms available" because all of them are useful for experimentation and asking questions, not to mention just fun to play with.

• you said: since the impulse trains are the result of driving a sinc function (sin(x)/x) with a ramp.
• proposal: calculate sin(x)/x as a chaotic equation (that means calculate it using a feedback based equation). I'll figure out how to turn sin(x)/x into an over-time calculation, again I don't know if it's already doing that. Maybe ill look at the code...

[mystran]

My two cents: generating band-limited waveforms is the easy part. Keeping them band-limited as you do interesting processing with them is the difficult part.

[antto]

ill include that in the repository for yet another band limiting technique. and again, the goal is not perfect waveforms. it is to discover chaotic equations that approximate whatever sound i want and tuned to be alias-free.

i thought the goal is to "end aliasing", as the thread title says

"i know how to end aliasing"

well, i know how to build a time machine. i just need time and money

[antto]

this reminds me of a (voxel?) renderer technical demo video on youtube (years ago) where it was claimed "infinite detail" is possible
it also reminds me of that dude here who claimed he knows how to compress data infinitely

EDIT: oh, i have an idea
with that same idea "blackhole for aliasing" - why don't you make a "blackhole for bugs", bug-free code!

EDIT2:

my reaction from yesterday:

[Architeuthis]

when i say infinite detail i mean infinite like pi, like a circle. it's infinite detail because of self-similarity, like vector. an analog sawtooth or BLIT perhaps (if it is perfectly aliased) has infinite detail. It's still just a half a triangle, very simple shape, but it still has infinite detail. You will never zoom in enough to find the exact edge point in spacetime of that sawtooth (in a a mathematical unbounded simulation) just like you will never zoom in on a circle enough to find an edge. it's infinite detail based on the timestep (samplerate), the tradeoff is that it's infinite within the bounds of the simulation. it's exactly as if you limited your frequencies to multiples of the samplerate, you will never alias. the problem is you also can't play any music. you can imagine it as: im unlocking the frequency spectrum by using infinitely smooth waveforms (again, infinite based on the bounds of the simulation, we still have to convert A/D <-> D/A at some point).

[mystran]

this reminds me of a (voxel?) renderer technical demo video on youtube (years ago) where it was claimed "infinite detail" is possible
it also reminds me of that dude here who claimed he knows how to compress data infinitely

Perhaps you mean "unlimited detail"?



If I'm not mistaken, that tech actually found a market for itself in point cloud visualization for engineering purposes, so it's not really completely ridiculous, though it obviously never made a huge impact in the gaming market (and these days Nanite can push so many polygons you basically have unlimited detail that way as well).

[Architeuthis]

unlimited detail, infinite detail. i like the word infinite because unlimited implies there was a limit in the first place. infinite is more exact, it describes a number: ∞ or we can drop the whole infinite/unlimited thing and just say:

"it's a chaotic equation that i tune not to alias"

ill say that moving forward.

also, this requires you to abandon classical audio synthesis, so in other words, I'm only ending aliasing in my simulated universe rather than for arbitrary audio. you have to actually jump in to get the benefit, tradeoff is you are not allowed to calculate sin() in this universe, you have to tap into the "free energy" source of "spacetime fluctuations" (i.e. the sinewave oscillating over time and the noise floor)

the reason i say "ending aliasing for all" is because i am positioned to grab a lot of the audio market, they will use my engine for their audio, be it a game, an audio plugin, etc... hmm, ok that's too optimistic. my engine will be used alongside classical synthesis im sure.

classical synthesis: good for getting something exact
chaotic synthesis: good if you don't need exact (especially organic / foley sounds).

[Architeuthis]

im working on fast sinewaves and gpu rendered audio for an additive synth that will run in browser and be part of my clap export feature (the base engine will be free, my fine tunes won't be, but anyone can do fine tunes, that's the point of this thing, anyone can make something and sell it in this alias-free universe). the alias-free universe is the whole point of this, jump in, make something, jump out with your new audio plugin or whatever.

[Benedict]

…this is the last human problem. ai will solve everything else after this.

There is no known solution to human arrogance that I am aware of.

You can try culling them as they appear

The problem being that it simply causes more to appear, and rather than the intention of removing the problem, you simply breed a whole new rage of them.

So yup, you can't solve it other than to point out their incontinence and then ignore their stupidity so they either go away or find a pond that suits their format.



ye cannae change the laws of physics, captain...

[Architeuthis]

let me know, benedict, if you have a question about what im doing.

the additive synth will be:
-phase
-frequency
-amplitude

control all arbitrarily with formulas and per sine modulation. this is extremely powerful, very musical, very useful. ive used it to model cymbals and hi hats, as well as do extreme sound design. it could probably be repurposed as a resonator and eventually ill do a vocoder when i learn how. imagine an in-browser vocoder that competes with vocodex.

[cpr]

I don't understand your time and money issue. unless you are dying soon, you have time. and writing code doesn't cost money, so why is that an issue? If you have a great idea, stop arguing with people here and just get it done. and let people know when you have some pudding where we can find the proof.

As well, there is no slack for people touting ideas. The natural response to just telling people you have an idea is for them to consider what you are saying and challenge you on things they either don't find clear, or just don't make sense based on their experience. If you want a true dialog to prove your theories, then you need to engage that with data, not acting all pissy that people don't believe what you are saying.

[Architeuthis]

cpr: "I don't understand your time and money issue."

response: i don't have a job, i lost my job and decided to use my savings to keep me going: spent the last month and a half working on what i have now, i am working on it as we speak, it's a lot of work, i don't want to demonstrate something half-baked. i want the next demonstration to be interesting and compelling. because i don't have money it means im spending time on survival: applying for jobs, finding scraps of cash, relaxing after a month of and a half of constant coding as long as i can keep my eyes open, etc.

follow up: i just nailed an interview, i might have a job soon.

[Benedict]

Congrats on the possible job. I hope it works out. As I hope your synth will work out too.

I am interested in any new synth, esp one that brings something new, seeing nothing new has appeared for decades now. I think the worry about the synth is that your expectations seem waaaaaay beyond the usual. Either you are the new Nikola Tesla or just another gunner... Time will tell.

Everyone else is struggling with Nyquestianism as a reality of physics, like gravity. Making grand claims with nothing to back them leaves you exposed, looking like a wacker. Getting bent at people who question you makes it worse.

If you have indeed come up with a method:
a) show it working,
b) keep quiet til you have the next (real) game changer, or
c) speak with some humility that your next doohickie will be showing a different method that may well be a successful alternative to the usual methods, e.g., Band Limiting and Oversampling (which do work - if imperfectly).