Easy additive synthesis idea for sample-based, subtractive syntnesizers

[Synthesis]

11 hours ago, Mr -G- said:

Can you be more precise about what is the pre-processing you do to the files before uploading and mixing them? Thank you.

I made an algorithm to prepare the samples. It is applied for all (the same process).

[Synthesis]

7 hours ago, YashN said:

Not sure there is additional processing involved

There is. The key here is sampling  method and sampling algorithm that I call a sample preparation for this easy additive instrument synthesis. 

[Tusker]

19 hours ago, YashN said:

How many bands did you find gives you the more interesting results?

 

Alchemy provides up to 600 partials, and if I need the full spectrum, I just keep that setting.  But you can make great tonal instruments with just the first 20 partials. The vital information is in the first 3-4 partials anyway. Almost all synths allow you to pitch oscillators/operators up and down, so you could extend the number of "partials" by changing the octave setting of an oscillator/waveform if you wanted to.

[Synthesis]

I'm going to make a synthesis journey with this newly built instrument tomorrow:

it's inspired by a Thomas Hitchcock Spinet From 17th Century
https://on.soundcloud.com/8v3qi

 

[Mr -G-]

12 hours ago, Synthesis said:

I made an algorithm to prepare the samples. It is applied for all (the same process).

And what does the algorithm do? Or is it a secret? 🙄

[YashN]

2 hours ago, Synthesis said:

I'm going to make a synthesis journey with this newly built instrument tomorrow:

it's inspired by a Thomas Hitchcock Spinet From 17th Century
https://on.soundcloud.com/8v3qi

 

 

I like the Digital Archaeology approach. It might be the only way barring a full reconstruction or a Physical Model to play these ancient instruments.

[Synthesis]

20 hours ago, Mr -G- said:

And what does the algorithm do? Or is it a secret? 🙄

Not a secret, but it's the know-how part. It's a complex process including Fourier analysis, spectrum analysis, preparation of additive partials and then finally creation of the instrument file itself. Since it's the know-how part, I can't give you more details, but I can tell you, that all input sound sample source go through the same process.

[Mr -G-]

So it is a non-secret that you won't reveal. 😂

 

[YashN]

You can make your own groupings...

[Synthesis]

Here is again a new instrument, a Harpsichord, ready for additive synthesis. I don't know yet what is hidden in it...

https://on.soundcloud.com/KuZrM

 

But this one is sounding...check it yourself, it's a simple portable regal organ from the 18th century:

https://on.soundcloud.com/AGBXg

I know what this instrument hides: e.g.: a real street organ sound.

 

[Synthesis]

 

What about a new synthesis journey with a Jupiter 8 Transisto, it is a Depeche Modish instrument a lot, though my secret garden, is not so secret anymore:

https://on.soundcloud.com/A4AWw

 

[YashN]

It all seems so absurd that this should have occurred..

[Synthesis]

Today I had 30 mins to record, how I synthesized two medieval instruments into one:

The result is many new instruments again. Unfortunately it is impossible to mix and record all sounding variations, if you mix two instruments of mine, the number of possible variations are 9^8= ~ 43 millions and I recorded those only that I found interesting in this 30 mins.

So here is the synthesis "mix" of a harpsichord sounds like this originally

https://on.soundcloud.com/9nZ97

and a spinet sounds like this originally
https://on.soundcloud.com/37MLu

into these instruments:

https://on.soundcloud.com/aujTm

by additive synthesis on my Yamaha Motif XF.

 

[Synthesis]

If we select 8 pieces from a sum of 16 elements, (don't forget, we synthesize 2 instruments = 2x8 elements/layers), there are 12870 pcs of different possible groupings of 8 different elements/layers out of these 16 layers/elements (in this case all layers/elements is on with a sound.) 

[YashN]

Imagine now, that each of these elements can be further processed individually with a real-time controllable DSP block, let's say a Resonant 4-pole Lowpass to simplify, that gives us...

[Synthesis]

On 4/22/2024 at 10:34 PM, YashN said:

Imagine now, that each of these elements can be further processed individually with a real-time controllable DSP block, let's say a Resonant 4-pole Lowpass to simplify, that gives us...

Yes, and yesterday I started to put the cherry onto the top of a synthesized sound this type by adding subtractive dynamic filterings/pitches/lfos on certain layers,  and it's eargasmic. I will record some of these cherried instruments too to show you some results.

 

Btw.: do you have a multilayered subtractive synth? If you had, I might try to convert some easy additive instruments for you of this kind to try it on your own.

[YashN]

8 hours ago, Synthesis said:

Btw.: do you have a multilayered subtractive synth? If you had, I might try to convert some easy additive instruments for you of this kind to try it on your own.

 

Let me show you what I play with:

 

 

 

 

 

Here's how I can do this in real-time on multi-Sampled Keymaps, and eventually further process each band:

 

 

I've included a second ALG (there are 31 different ALGs) where I could potentially FM the original source and then Ring Mod it with the Band if I wanted to.

 

Fun?

[Synthesis]

I made the first instrument with some additional subtractive dynamic settings  (with "cherry on the top")...
https://on.soundcloud.com/vzRt1

 

Kurzweil K2500 is a great synth. Can you read .kr1 files into your machine? How many waveforms (layers) can be played on simultanously (do you have 8 layers)?

[YashN]

4 hours ago, Synthesis said:

I made the first instrument with some additional subtractive dynamic settings  (with "cherry on the top")...
https://on.soundcloud.com/vzRt1

 

Solid cover of A Question of Lust, you've got the Martin vibrato and timbre going too, the synth sound is good too. Impressive. I'm more of a Gahan-style Baritone myself.

 

 

4 hours ago, Synthesis said:

Kurzweil K2500 is a great synth. Can you read .kr1 files into your machine? How many waveforms (layers) can be played on simultanously (do you have 8 layers)?

 

It's more than a synth: it's a Synth-building platform, an Effects-building platform, a Drum Machine-building platform, an Advanced Sampling Engine building platform, etc... In fact, I can use mine as a Filter-building platform too.

 

I've highlighted the max number of Layers in the rightmost column of the first picture above: 32 Layers.

 

Usually, it's when samples overflow a single HD diskette that the file format becomes .KR<n>, but yes I should be able to read that. The usual format would be .KRZ for K2000, and .K25 for K2500, .K26 for K2600, .PC3 for PC3, etc...

 

I believe my K2500 has limited Sample Mem but no worries I also just got another machine with larger Sample Mem so it should load there when I get my External Floppy Drive this week-end.

[YashN]

If you want to have a look at how a V.A.S.T. workstation works, I recommend downloading the PDFs of the K2600 Manuals here (official site):

 

https://kurzweil.com/k2600/#downloads

 

For some reason the DSP ALGorithms Chapter has disappeared from one of the Kurzweil K2500 official Manual download, so for that and a few other things, the K2600 Manual is more complete.

[Synthesis]

Ok, i dig into it, and going to make an instrument for K2500 the weekend.

What is the max filesize you can load into your K2500?

 

[YashN]

Don't worry about Sample Mem size: as mentioned, I have enough on another machine.

 

I was working on something else these past few days but last night I set up a quick one like I described above in 8 octave bands, and manually changed the Keymaps among 2 different ones first - Strings and Voices, then I also did a set of 8 with 3 Keymaps: a mix of two Synth Strings and Sawtooth.

 

Slapped a Chorus, Delay and beautiful Hall Reverb on the last one.

 

Instant Pad Machine and Jarre from the Oxygène - Equinoxe - Les Chants Magnétiques era.

 

Would need some additional work on Layers and Control to make it work completely like what you're doing but doable. The advantage with the implementation on the Kurzweil, as I mentioned, is that I am doing the bands in real-time and in-the-box: no need to preprocess the Samples externally. If I did though, that would liberate some DSP Blocks for building other chains although I can still add some further processing as is or via some clever tricks that I know about on the K2500.

 

I did notice that with acoustic Keymaps, you'd need to tailor amplitude of the higher harmonics somewhat higher than normal for the sustain parts if you want to move the acoustic sources into Synthesis-land and not acoustic-land anymore.

 

BTW, I used to be a huge DM fan. Can't say I like the direction where the last 2 or 3 albums went nor how Gahan sings Live these days. Although Ultra was really good, the band does seem to miss Alan Wilder's seminal contributions in the Studio. I'd rather listen to Speak & Spell, Some Great Reward and Black Celebration than the more recent three albums. I wasn't a big fan of the turn that they took with Violator either although that one brought a lot of new people to the band's work.

[Synthesis]

What about this .krz file?

I'm curious if this works for you on your Kurzweil K2500. It's 22.9 MBytes. This instrument is capable for this easy additive instrument synthesis. If you change the volume of the layers. I think there will be one problem for sure: this seems to be 16 mono layers and not 8 stereo layers. Does this mean, that a layer can only have a mono waveform on Kurzweil K2500?

Please let me know your results.

Download it here:

https://drive.google.com/file/d/1gITm7BenLiRI0ZFB5mRaOKEMhaVuFj-7/view?usp=drive_link

 

[YashN]

I get an 'access denied' on this link.

 

Mono works too - a LAYER's Keymap in Kurzweil land works like this:

- Default = Mono

- Setting Stereo = On on the Keymap Page makes it a Stereo Keymap, so that you can choose two Multi-Sampled Keymaps and further Pan them differently. Both Keymaps here are routed into a single ALG in the LAYER

 

Another way to do a Stereo Program on a Kurzweil is to use two different LAYERS, each one with a Mono Keymap, and then Panning each separately differently.

[Synthesis]

Download it here:

https://drive.google.com/file/d/1gITm7BenLiRI0ZFB5mRaOKEMhaVuFj-7/view?usp=drive_link

 

Sorry, now it is shared and available.

 

[YashN]

Got it. Thanks

[Synthesis]

Have you had any success in loading this .krz file and had any synthesis result? I only know about Yamaha Motif Xs/Xf, these instruments sound there interesting, but I don't know about how a .krz file can be heard in a K2500?

[YashN]

6 hours ago, Synthesis said:

Have you had any success in loading this .krz file and had any synthesis result? I only know about Yamaha Motif Xs/Xf, these instruments sound there interesting, but I don't know about how a .krz file can be heard in a K2500?

How did you write that .KRZ file?

 

Success, me? Never!

 

Here how it should work:

- I should have kept my old SCSI CDROM drive, then I'd probably be able to write that file to a CD from another drive, then transfer it over via the SCSI Drive. However, I left behind the SCSI CDROM drive when I moved.
 

- An alternative would be to have several empty HDD floppies at hand, then find the utility to split that single file over to several floppies and write the floppies from my external USB-to-Floppy drive that I got second hand from a nearby store. However, I don't have enough empty floppies, nor can I find the external floppy drive, so...

- ...since I will need that external floppy drive ability for a few projects, I ordered one from Amazon a week ago with two more small things. Amazon somehow automatically tracked the order, only to tell me a few days later that they found a 'technical error' had been made in that order and they auto-refunded the purchase. They did tell me to keep it all if it were delivered. Nothing was delivered though.

- I should have one of those modern SDCard to SCSI emulators, and done the transfer from that, assuming my K2500 has enough SRAM (I think it does not but that isn't a big issue, I have another machine to test it on). I still don't. I am planning to though.

That last one is my best bet, so I should be able to test it when I get one.

I did manage to have a look at the structure through Awave that supports .KRZ and its Sample structure.

Let me know If I got this right:

 

• Multi-Sampled on 6 Keys
• Pre-processed with Bandpass for most if not all of them, Bands do not overlap
• Mapped so that you can easily Mix between two multi-sampled Keymaps per Band

 

My test the other day on the K2500 required no external samples, I used the internal multi-sampled Keymaps, and required no external pre-processing: it's all done in real-time.

 

I am thinking that the results I got which seem quite good for String machines is not only from the effect section but also the Phase changes thanks to the BPFs.

 

Of course, it's not restricted to that in the Kurzweil as you can do all sorts of combinations of any 2 Keymaps, and in fact, it's really easy to extend that to 8 different Keymaps, and if I want to get crazy with it: up to 32 Keymaps and many more bands, all in real-time, and add even further processing if I want  to.

 

The little test I did, although it's not with your samples nor the full implementation of your approach, does already give an idea of the method though.

[Synthesis]

I tried to make that .krz file in awave, but I don't know what result it would give loaded into a K2500... It's great you have BPFs on K2500, that don't remove too much of an original sounding of a factory patch. I need to prepare the samples so as not to loose frequencies/amplitudes from the original sample, and moreover, I don't like hearing the step in sounding  between two different sampled sound of an instrument, because it is there also, if "partials" are reused. Preparing the samples you can avoid as much artifacts as possible .

[Synthesis]

And I shall stay at low size files to load them into mem of synthesizers. 

 

Ok, but now you guess, what this is like.

 

If you load prepared instrument files, you avoid any time-consuming settings, just have fun, that's important too, I think.