Raw hardware oscillator samples
-
OneOfManyPauls OneOfManyPauls https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=382596
- KVRian
- Topic Starter
- 1103 posts since 17 Jul, 2016 from Wales, UK
There are plenty of libraries/packs out there with sampled full patches from hardware synths, but I'm looking for a library of just raw oscillator samples from as large a collection of identified hardware synths as possible - ideally across several octaves (doesn't need to be note sampling), preferably .wav format.
single cycle waveforms would be fine.
This is something I've searched for on and off for a long time now, but never found anything satisfactory.
Does anyone know of anything like this?
single cycle waveforms would be fine.
This is something I've searched for on and off for a long time now, but never found anything satisfactory.
Does anyone know of anything like this?
-
- KVRAF
- 2289 posts since 5 Jan, 2006
Try here, most of their libraries also have samples of the raw oscillators, with loop points and everything, and they are completely free (public domain):OneOfManyPauls wrote: ↑Mon Jan 17, 2022 9:36 pm I'm looking for a library of just raw oscillator samples from as large a collection of identified hardware synths as possible
Does anyone know of anything like this?
https://github.com/publicsamples
-
OneOfManyPauls OneOfManyPauls https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=382596
- KVRian
- Topic Starter
- 1103 posts since 17 Jul, 2016 from Wales, UK
thanks for that - I did take a look at that site before, but I didn't see any raw oscillator samples - only sampled full patches. I didn't dig too deep though, so will have another look.PTV wrote: ↑Wed Jan 19, 2022 8:50 am Try here, most of their libraries also have samples of the raw oscillators, with loop points and everything, and they are completely free (public domain):
https://github.com/publicsamples
waveforms labelled the same by different manufacturers can sound very different.
this isn't some naive attempt at exactly replicating full patches using just single cycle waveforms, but using the right waveforms can get closer to the sound of the original hardware on my sample/wavetable based synths and gives a broader palette in general.
- Banned
- 10732 posts since 17 Nov, 2015
I prob have these on my PC, I'll have a look later if you think they might be of interest
(Links are dead, but you get the idea)
https://forum.vintagesynth.com/viewtopic.php?t=43399
(Links are dead, but you get the idea)
https://forum.vintagesynth.com/viewtopic.php?t=43399
-
OneOfManyPauls OneOfManyPauls https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=382596
- KVRian
- Topic Starter
- 1103 posts since 17 Jul, 2016 from Wales, UK
That'd be great if you can - thanks!
In case this is useful to anyone else, here's a wavetable I made from some of my own hardware synths - 2048 samples per waveform, 40 waveforms in the wavetable:
https://www.dropbox.com/s/r16uswmbiz5e6 ... 8.zip?dl=0
sub 37: tri>saw>sqr>pulse m > pulse n
as1: tri>saw>shark>sqr
perfourmer mk2: sin>tri>saw>sqr>pulse m > pulse n
mother 32: saw>srq>pulse m > pulse n
jdxa: sin>tri>saw>sqr>pulse m
prophet 12: - saw>sqr>sin>tri>others
In case this is useful to anyone else, here's a wavetable I made from some of my own hardware synths - 2048 samples per waveform, 40 waveforms in the wavetable:
https://www.dropbox.com/s/r16uswmbiz5e6 ... 8.zip?dl=0
sub 37: tri>saw>sqr>pulse m > pulse n
as1: tri>saw>shark>sqr
perfourmer mk2: sin>tri>saw>sqr>pulse m > pulse n
mother 32: saw>srq>pulse m > pulse n
jdxa: sin>tri>saw>sqr>pulse m
prophet 12: - saw>sqr>sin>tri>others
- KVRAF
- 2353 posts since 3 Mar, 2010
Yes - I would very much appreciate these, as well!AnX wrote: ↑Wed Jan 19, 2022 5:23 pm I prob have these on my PC, I'll have a look later if you think they might be of interest
(Links are dead, but you get the idea)
https://forum.vintagesynth.com/viewtopic.php?t=43399
- Banned
- 10732 posts since 17 Nov, 2015
-
OneOfManyPauls OneOfManyPauls https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=382596
- KVRian
- Topic Starter
- 1103 posts since 17 Jul, 2016 from Wales, UK
-
- KVRAF
- 2289 posts since 5 Jan, 2006
It's the construction kits you're after:OneOfManyPauls wrote: ↑Wed Jan 19, 2022 11:08 amthanks for that - I did take a look at that site before, but I didn't see any raw oscillator samples - only sampled full patches. I didn't dig too deep though, so will have another look.PTV wrote: ↑Wed Jan 19, 2022 8:50 am Try here, most of their libraries also have samples of the raw oscillators, with loop points and everything, and they are completely free (public domain):
https://github.com/publicsamples
https://github.com/publicsamples/Roland-Alpha-Juno
https://github.com/publicsamples/Modula ... uction-Kit
https://github.com/publicsamples/Moog-Minitaur
https://github.com/publicsamples/Oberhe ... -Session-1
https://github.com/publicsamples/Oberhe ... -Session-2
https://github.com/publicsamples/Roland ... uction-Kit
- KVRAF
- 12555 posts since 7 Dec, 2004
It's very difficult to sample at the required level of detail directly from the oscillator itself to produce anything other than near mathematically perfect waveforms. The later stages are a combination of filters and non-linear buffers... but it's important to remember that the impact these have on the final output may not appear before the filter for instance.
Ultimately you need to look at the complete system and signal path and sample each point with a high impedance (like an oscilloscope probe + buffer circuit) to take accurate measurements. This requires disassembling synthesizers and delving deeply into great levels of detail with great effort.
There definitely are some variations (the 'good' tb-303 'pulse' being a classic one) with interesting properties but generally speaking these are non-static and change at different frequencies or when modulated.
The Alpha Juno for example I could provide samples from the lifted pin from the oscillator chip... but that would be wasted effort. I can tell you right now the only measurable difference from mathematical perfection is feed-through of the digital clock signal.
Ultimately you need to look at the complete system and signal path and sample each point with a high impedance (like an oscilloscope probe + buffer circuit) to take accurate measurements. This requires disassembling synthesizers and delving deeply into great levels of detail with great effort.
There definitely are some variations (the 'good' tb-303 'pulse' being a classic one) with interesting properties but generally speaking these are non-static and change at different frequencies or when modulated.
The Alpha Juno for example I could provide samples from the lifted pin from the oscillator chip... but that would be wasted effort. I can tell you right now the only measurable difference from mathematical perfection is feed-through of the digital clock signal.
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
- KVRAF
- 12555 posts since 7 Dec, 2004
My point being to explain why these sample sets aren't easy to find. They're 99% not useful.
I've looked through these to try to find technical analysis and audible differences in waveforms but I've almost always been very disappointed. Often poor quality samples taken through a long signal path after the output of the synthesizer. Usually including additional stages and effects and into a low quality line input on a sound card. In my experience I've found to get any practical use out of these requires a lot more than just samples themselves - it requires enough technical knowledge to synthesize those waveforms directly myself. In other words to really get any notable results I need to be able to write the math function that creates the waveform!
... and, that is easier said than done!
What you'll generally find is dozens of "saw" waveforms that are exactly alike. You'll find one or two that seem interesting and might look into how these shapes are created in the synthesizer in question. That'll almost inevitably lead to the discovery that the oscillator itself in that synthesizer produces a near perfect waveform up to the output... and that the person who sampled the thing used the headphone amp output which is what really produced the distortion!
I've looked through these to try to find technical analysis and audible differences in waveforms but I've almost always been very disappointed. Often poor quality samples taken through a long signal path after the output of the synthesizer. Usually including additional stages and effects and into a low quality line input on a sound card. In my experience I've found to get any practical use out of these requires a lot more than just samples themselves - it requires enough technical knowledge to synthesize those waveforms directly myself. In other words to really get any notable results I need to be able to write the math function that creates the waveform!
... and, that is easier said than done!
What you'll generally find is dozens of "saw" waveforms that are exactly alike. You'll find one or two that seem interesting and might look into how these shapes are created in the synthesizer in question. That'll almost inevitably lead to the discovery that the oscillator itself in that synthesizer produces a near perfect waveform up to the output... and that the person who sampled the thing used the headphone amp output which is what really produced the distortion!
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
- KVRAF
- 12555 posts since 7 Dec, 2004
Here's an example of what I'm talking about:
Left/white = analog waveform (through the output, not raw from oscillator/mixer!)
Right/pink = digital recreation (from Xhip)
There are two remaining differences. Things you might note are the "flat top" of the waveform and the difference in ramp vs. pulse vs. sub. These are easy to explain.
The difference in amplitudes simply comes down to minor differences in mixer levels. I didn't bother to try to tweak it to perfection because I already know it's possible and I'd be wasting my time on something pointless. Obviously while you might set what you think is exactly "full" gain in a synthesizer, that doesn't mean it's "1.0" or 0 dB or anything remotely like that. It's just an arbitrary level where the knob/fader maxes out! In software like Xhip, obviously (I hope!) if you set exactly 0.0 dB that's exactly what you get!
The "flat top" is due to the DC blocking high-pass filter capacitor in the mixer/amplifier/output and later stages outside the synthesizer. If you sample a different frequency you'd find this varies. It turns out at this particular frequency and with the particular capacitors and other circuit elements in the analog synthesizer, this sample of this note at this moment in time just so happened to produce a very nearly "flat" top on the waveform. It's nothing but a 6 dB high-pass filter! In Xhip that was recreated by applying a 6 dB high-pass filter on the output.
Additionally the anti-aliasing filter in Xhip differs from the sound card filter. The sound card used a zero-phase filter while Xhip uses minimum phase. These differences aren't audible and amount to rounding error in most cases or ultra-sonic harmonic differences in others. Either way you won't be hearing the difference.
There is also an invisible phase difference between the sub-oscillator, ramp and pulse waveform in the analog synthesizer output vs. the perfect zero-phase alignment in Xhip. This could be modeled by very precise fine-tuning of tiny 1/100ths or less phase offsets for each to recreate a perfect identical spectrum... but again that would only model that exact particular instant in time in the sample and while notable has little to do with what makes the analog waveform "analog".
When you get into the nitty gritty details of these things they become more like a child's fascination with clouds in the sky, ripples in a pond, bark on a tree, moss on a stone or stars in the night sky. For someone who understands these things they no longer carry the same majesty they once did, much like staring at the paint on a wall!
Now let's all take the time to appreciate the sublime beauty of the textures left in the paint on the wall and become child-like and mystified in awe as we once may have been and forgot all those years ago.
I'm a spoil-sport, I know.
Left/white = analog waveform (through the output, not raw from oscillator/mixer!)
Right/pink = digital recreation (from Xhip)
There are two remaining differences. Things you might note are the "flat top" of the waveform and the difference in ramp vs. pulse vs. sub. These are easy to explain.
The difference in amplitudes simply comes down to minor differences in mixer levels. I didn't bother to try to tweak it to perfection because I already know it's possible and I'd be wasting my time on something pointless. Obviously while you might set what you think is exactly "full" gain in a synthesizer, that doesn't mean it's "1.0" or 0 dB or anything remotely like that. It's just an arbitrary level where the knob/fader maxes out! In software like Xhip, obviously (I hope!) if you set exactly 0.0 dB that's exactly what you get!
The "flat top" is due to the DC blocking high-pass filter capacitor in the mixer/amplifier/output and later stages outside the synthesizer. If you sample a different frequency you'd find this varies. It turns out at this particular frequency and with the particular capacitors and other circuit elements in the analog synthesizer, this sample of this note at this moment in time just so happened to produce a very nearly "flat" top on the waveform. It's nothing but a 6 dB high-pass filter! In Xhip that was recreated by applying a 6 dB high-pass filter on the output.
Additionally the anti-aliasing filter in Xhip differs from the sound card filter. The sound card used a zero-phase filter while Xhip uses minimum phase. These differences aren't audible and amount to rounding error in most cases or ultra-sonic harmonic differences in others. Either way you won't be hearing the difference.
There is also an invisible phase difference between the sub-oscillator, ramp and pulse waveform in the analog synthesizer output vs. the perfect zero-phase alignment in Xhip. This could be modeled by very precise fine-tuning of tiny 1/100ths or less phase offsets for each to recreate a perfect identical spectrum... but again that would only model that exact particular instant in time in the sample and while notable has little to do with what makes the analog waveform "analog".
When you get into the nitty gritty details of these things they become more like a child's fascination with clouds in the sky, ripples in a pond, bark on a tree, moss on a stone or stars in the night sky. For someone who understands these things they no longer carry the same majesty they once did, much like staring at the paint on a wall!
Now let's all take the time to appreciate the sublime beauty of the textures left in the paint on the wall and become child-like and mystified in awe as we once may have been and forgot all those years ago.
I'm a spoil-sport, I know.
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.