Freesound Forums

Hi.

I would like to see on Freesound a pack of basic waves (and others) to feed our oscilators, like:

-sine,square,sawtooth, triangle.......

And other ones, less typical like noises and... well... any kind of 512 samples(i think) long wave for feeding our OSC.

i suggest everyone that haves one, to upload it and write here, so Bram can go adding them to a pack as they come.

What do u think ?

I like!

I could record the wave forms of some classic analogue synths like the ARP 2600 if you;d like. SO the wave form needs to be 512 samples hhhm, thats gonna take some tuning i gues, but im willing to do the effort if you think it will be nice.

Cheers
Anton

I am not really sure about the 512 samples, it-s what i use on maxmsp on the oscillators. What about other hard or softsynths ? or ... Reaktor, what does it uses?

And, Anton, it would be really cool to have those waves, yeah. I was wondering, tho, if one can record things like that from a commercial synth, and still being able to use a CC license on it, what do we know about it?

If it is legal i would love to see those waves, Anton!!!

plagasul
And, Anton, it would be really cool to have those waves, yeah. I was wondering, tho, if one can record things like that from a commercial synth, and still being able to use a CC license on it, what do we know about it?

As far as I have been told it's illegal to use the preset sounds from PCM synthesisers (i.e. 'rip' the ROM-waveforms from a PCM synth), however it's not illegal to use analog waveforms as in essence you tweaked the synthesiser yourself to create that waveform.

But - please - be careful what you sample.

And - multisamples are a must for small waveforms.

Bram

And - multisamples are a must for small waveforms.

you mean at different frequencies?
Plagasul: could you give me an example of a waveform sample so that i have something to aim for?

Anton
you mean at different frequencies?

yes.

single (or multiple) waveform cycles at different frequencies!

What kind of range should i do, one per octave? or more?

[edit]: fixed

I will make some, not sure about the range, i will do it arbitrarily, if someone can, enlight us.

Hi there.

I seek some answers, let me see if i can ask the right questions:

-Is there any reccomended lenght per cycle of a wave, for people to use it on their gadgets?

-Correct me if i am wrong, but, when increasing the frequency of a wave aren't we just increasing the cycles per second? If i am right, then what is the point of providing the wave at various frequencies?

Thanks.

plagasul
Hi there.

I seek some answers, let me see if i can ask the right questions:

-Is there any reccomended lenght per cycle of a wave, for people to use it on their gadgets?

-Correct me if i am wrong, but, when increasing the frequency of a wave aren't we just increasing the cycles per second? If i am right, then what is the point of providing the wave at various frequencies?

Thanks.

1. length per cycle dtermens pitch, it doesnt really matter i think as long as you map it correctly
2. providing several frequencies might be interesting because the characterstic of an analog oscillator might not be linear across the spectrum.

Just uploaded the Arp waveforms, bit long recordings. I recorded them @ 96khz 24 bit, but Adobe Audition made them 32 bit floats for reasons i dont care to investigate at the moment. I guess this isnt the format most people prefer their waveforms in (esp since its about 5s samples of each waveform) so I suggest somebody who knows what to do with them to remix them

Enjoy

First of all, hello to you all at Freesound Project.

plagasul

-Correct me if i am wrong, but, when increasing the frequency of a wave aren't we just increasing the cycles per second? If i am right, then what is the point of providing the wave at various frequencies?

It all depends on what you mean as there is a difference between physical acoustic tuning and digital tuning.

Digital tuning doesn't just alter the hertz or frequency, it resamples or reassigns a load of zero's and one's in order to emulate the effect of altering the hertz or frequency. The more extreme the alterations to a digital sample, the more the binary data gets squashed or stretched, depending on the direction of pitch-shifting. This can result in the loss of some of the original binary data and you end up with quite a different sound from what you had originally.

I have an old Emulator 3 hardware sampler from the late 80's-early 90's. When I am assigning a sample to a zone, if I use just one sample and zone it across the whole octave range, the sample assigned to the higher notes is either just plain awful or corrupted. This anomaly is known as the Chipmunk Effect.

Therefore I have to make various samples at different pitches in order to have a usable full keyboard range of the source sound.

I usually use various source samples of the same sound at one octave intervals, i.e. C1 C2 C3, etc. and zone them accordingly. I have even used samples pitched by 5th's/7 whole-tones when sampling my guitars to get a more natural sounding instrument.

Something like that anyway.......... :roll:

nice explanation of the issue. And i guess perhaps this even applies to sampling just a few key wave forms on something like an arp.

Cheers
Anton.

Thx...

hello.

analog oscillators vary in character (waveform) over their range.
partly this is from the freq response / bandwidth of the synths circuitry.

conventionally, i think one sample every 3 semitones is about optimal.
this i have picked up from various sampler manuals and tutorials etc.

once you transpose a sample more than a couple of semitones, you can easily hear the formant shift.

formants are the resonances that are not dependant on note-pitch.

i would guess that synth oscs don't vary as much (over their range) as acoustic voices, one (crossfaded) sample per octave may suffice, but up the top things could get critical, like there's probably a fixed cutoff at about 16khz-ish.

btw: 44100/512=86.1328125 which is just flat of F1.
don't know how relevant that is.

maybe i should clarify something:
when i say "crossfaded" waveforms, i mean that the person assigning the samples to keygroups on their sampler could x-fade the keygroups (zones) to even out the transitions.
nothing to do with the recording side of things.

if you're sampling raw waveforms, I suggest you loop them. Either using a crossfade of some kind, or preferably by just truncating the file to fit.

I do multisamples in minor thirds, but not always. If you for some reason can't do multisamples (eg if it's an industrial machine going an "musical" frequency) then make the recording as long as you can. This will make it sound a lot better looped and pitched at high frequencies. 5-10 seconds at least, imo.

When I can find the time, I'll sample my analogues and my beat-up accordion for yall.

BTW; if any of you are contemplating doing some serious sampling of midi gear, then check out this product:
http://www.sonicstate.com/news/shownews.cfm?newsid=2068
-it's a program that automatically does multi-samples of midi instruments.

Andreas

I saw it before, although it looks nice it shouldnt be that much work to create something like it with pure data or max/msp. Besides the creation of Halion/sf2 etc. files.