The aliasing thread

[Majken]

The 8th note in the rising pitch has audible aliasing overtones, so does the 3rd, 13th, 14th and 15th in the descending pitch.

There are likely other overtones as well i think i spotted one at the 11th rising, after that it gets very hard to hear though since the get drenched by the basenotes. Also the 2nd, 11th and 12th descending has some less obvious overtones.

/Majken

[WilliamK]

..

[george]

you will mostly use multi-sampled sounds that won't produce aliasing with any engine

False, unless you are talking about really really big using at least one sample per key. AFAIK AKAI disks used get one or two samples per octave.

However, I agree multi sampled set will make the aliasing not that relevant, but from those graphics you can still see how many digital junk you can get once you play a single semitone up or down.

Still, when using samples with such higher frequencies

Do you realize that a 15khz sine tone is used to see how alias affects fundamental frequency only? What is the point of using other sampled material with harmonics that make the graphics much more confusing instead?

[WilliamK]

..

[autloc]

This thread is entertaining. But seriously, WilliamK, how's about a test with a sample that has harmonics extending right up into the teens, like a bandlimited sawtooth? (or you can use a sine wave at 15kHz, that's fine too)

Personally I don't mind some synths or samplers aliasing (loved my dw-8000 and esq-1), just so long as I have some that don't, but if it came down to a choice between the two I'd obviously take the one without the aliasing.

[arguru]

WK: I guess u guessed is a typo. (edited)

[9b0]

"Sooooooo, listen to those samples, look the picture. And tell me now. Are those other "fancy-tests" acurated? Yes, but are they pratical in musical terms? NO! "

The samples u used are made from 3 (compared to the original 15KHz signals low frequency) sinewaves. Thats not a good musical test. For musicians there is the tambourin test, which explains a lot. The same happans to all sounds, which have fundamentals over the half of the sample rate frequency.

Why i still agree with your viewpoints? Try to use a 256 Mbytes sample package with a bandlimited sampler! Or even better... try to use a 512 Mbytes package.

The same algorythm can't be used for big sample lybraris, like mostly used by VA synths. It makes a big difference to create a 5-6 octave wavetable for a small, 1024 sample long saw wave, than for a huge 512 Mbyte piano library.

(excuse me for my bad english knowledge)

[Majken]

Are you talking about my tests? Did you listened to the SFZ 72 file too? That one is suposed to be alias free. Do you also hear that on those? Just checking.

I didn't see you had posted the SFZ when i was listening to the Wavestation ones. Just compared them though and they are very very easy to tell apart. Especially on the descending notes.

Actually the difference was bigger than I thought it would be. I can't hear any aliasing in the SFZ examples. Very interesting stuff! Thanks alot

/Majken

[george]

This thread is entertaining. But seriously, WilliamK, how's about a test with a sample that has harmonics extending right up into the teens, like a bandlimited sawtooth? (or you can use a sine wave at 15kHz, that's fine too)

Back in the thread:

The only thing that gets in my nerves is all those fancy-mambo-jambo talk about aliasing, that most developers use. To fool people.

And what are you doing then, William?

[Majken]

Hey guys, calm down no flaming n bashing. I really don't want this thread to turn into another locked thread. This is by far the most interesting thread I've read for a long time.

George: He did say something about placing the wavs in a folder a few posts ago. Or maybe that didn't answer the question?

/Majken

[george]

George: He did say something about placing the wavs in a folder a few posts ago. Or maybe that didn't answer the question?

Sorry, I must have overlooked that, but I don't remember seeing it, and have been following this thread almost post by post.

[autloc]

Yeah, I don't want this thread locked, either. I've never seen side-by-side comparisons for samplers/synths like this and it's pretty interesting. Unless the aliasing is really terrible (or really good ), no synth is determined purely by how much it aliases.Still, SFZ is very impressive!

[WilliamK]

..

[autloc]

And I'm still a bit worried, that some so called HQ-NoAliasing could cutt some frequencies you want from the sound. Resulting into dull sound.

It sounds like you grew up using an Amiga I'm pretty sure that the Amiga's "anti-aliasing" filter was analog and came well after the aliasing was already in the signal, though, so that's why it was such a dismal piece of junk -- the aliasing was still there, just attenuated (along with all the musical data, unfortunately).

[Majken]

William count the notes as they go along, on the 8th note you will hear a highpitched noise in the W-station one however not in the SFZ.

I tried queing up both samples in winamp and then activated repeat. Then i played them in the background for a while and turned up the volume to see if i could tell them apart in a blind test. I tried it a number of times and I was correct every time.

The more I listen to them the easier they get to tell apart from eachother. SFZ sounds more dampened overall, as if it had a lowpass filter.

I've been listening in headphones which may make the aliasing easier to spot since i don't get disturbed by room sound. Anyway the setup is what I said yesterday : E-MU 1820m using Sennheiser HD-570 headphones.

William, please make an example based on triangle wavesforms and post the samples without a name. Just number 1 and number 2. Would be interesting to see if I could tell them apart without having heard them before.

/Majken