will we ever seen granular processing in the h9000?

[taylor12k]

seems like there should be enough DSP for some nice granular stuff in the H9K …. think it will ever happen?

 

[stevepretty]

+1 for this please. Would love to see a granular playground on the H9000. So many possibilities!

[Puppeteer]

Have you tried algorithms with crystals in them, or some of the reverse delay algorithms, or precision delays?

What type of functionality are you chasing?

[taylor12k]

google can explain better than i can… but, granular synthesis isn’t delay…. it chops your sound into tiny fragments and then let’s you re-arrange the playback of those fragments with, generally, control of the shape and pitch of each grain. it’s rooted in sampling more than echo because you can re-arrange fragments in time, not just repeat one particular moment of time.

there have been piles of granular plug ins over the years (google comes in handy here) and with the H9000’s DSP power it seems like real time sampling and granularization may be attainable.

[Puppeteer]

I know what granular synthesis is.  A digital delay is really just a little sampler where the buffer is continuously being written to.  The H9000 has tools that can be repurposed to do granular, and I may have a go at building something, hence why I was asking what features you are chasing.

Do you want a static sample that you mess with in real time (ie load a sample by recording with the H9000, and then mess with that sample), or do you want to process live audio by looping a section of the sample buffer that is constantly changing. Do you want envelopes on each grain, crossfade start and end, ability to reverse direction?

There are quite a few ways I could build this on the H9000, but depending on features, would determine which approach to take.

[taylor12k]

ah, sorry! my apologies.. i read that as if you were unfamiliar with granular synthesis! sorry 🙂 yeah, i mean, you’d think the H9000 would be capable of it.with the horsepower it has, and i would have thought that Eventide would be exploring some new territory, algorithm-wise, at this point…

functionality… i guess i was thinking a real time thing.. live audio input into a buffer with your typical granular controls… grain size, envelope, number of grains, spacing, etc.

you really think it can be done with delay lines? i wouldn’t even know where to begin, so i’m at your mercy and wouldn’t be picky. certainly if you think it would be possible i would be really interested to see how it could be implemented.

[Puppeteer]

I’ll see what I can do.

[Puppeteer]

OK,  here’s my first attempt at granular processing on the H9000. I put together a bit of a demo.

I’m planning on adding MIDI so you can play the pitch in (just need to get back to the studio to plug a keyboard in, development has been done remotely so far)

Here’s a closeup of the controls. It does crunch, nice flanging (that was a bit of a surprise), and cheap reverby effects.

There’s lots of different directions I can take this.  The algorithm should be available in about a week, along with a manual and a bunch of presets. So far it’s a lot of fun. I built it to be quite musical with the envelope and granular window being synced to the pitch.

I’ll post again when it’s available for download.

[taylor12k]

wild stuff! i wish i knew how to do deeper vSig stuff….

 

is it possible to be able to sample into and access a larger sample buffer? the type of granular i’m talking about is the sampling and chopping/rearranging of chunks of audio.. at least a second long.. if not more…

can be seen (and heard) quite simply here:

https://we.tl/t-WvofPOmwQl

 

about 3 minutes in.

i’m sure you’re familiar with these sorts of tools. so it goes beyond just being a pitch shifter but more of a “let’s cut this input into a bunch of small chunks and rearrange them in real time at the output”…

[Puppeteer]

I’m not sure if that was the correct link you sent. The Julie Byrne album is pretty awesome, but I didn’t hear any granular stuff in the first 2 tracks.

As I said, lots of places I can take this.  The first attempt was more a proof of concept, a bit along the lines of the iWish plugin, and it sounded pretty good and useful, so I’ll release it, but I’ll keep tinkering as I get time. Don’t worry, this won’t be the last algorithm of this type that I build.

One of the limitations with real time input is that it’s always going to follow the incoming audio to some extent, because the delay buffer is always going to contain the latest x seconds of audio, and it’s always changing, so windowing doesn’t tend to capture the same snippet of audio. The algorithm I’ve developed to date, will capture the window if you dial up the feedback to 100%, where it ignores incoming audio (you’ll also need it to be 100% wet to stop the dry mix into the outputs). The way I’ve implemented it, this will just loop the delay buffer indefinately, but if you mess with the other controls, it will often die away to nothing, because some of the processing is on the delay loop.  I’ll have a look at implementing some delay offsets into the next algorithm, which will mix things up a bit, and also a different way of freezing an audio slice.  The delay I’ve used allows up to 10 seconds of audio, but there are a bunch of different options there.

I’ll also have a look at puting the envelope inside the loop. Currently the envelope just processes the audio coming into the delay loop.

Some granular tools have a static sample and you modulate the window of sample that is looped.  That’s also possible (99% certain of it anyway), but it will be a different approach. I’ll have a look at that sometime in the future.

[taylor12k]

ha.. oops.. wrong link! (was supposed to be to a youtube video demo’ing some standard granular plugs) but, yeah, your bit about the 100% wet and delay buffer sounds interesting…

and, yeah, the idea of a static sample buffer with scrubbing through with a window… exactly…. but this can also be done in real time these days with a lot of plug ins.. that’s the granular aspect i was talking about…

of course, there’s not necessarily a reason to do it on the h9000 if there are dozens of plug ins that can do it already… but my initial question on this thread was simply whether the h9000 was capable. i haven’t seen the h9000 have more “advanced” effects than the 7600 or 8000 i used to have… just “more of them at once”. i think eventide has dropped the ball a bit in really pushing the platform to more exotic and DSP-hungry types of processes…

don’t get me wrong.. i love the integration of the 9000 and the ability to really stack algorithms unlike we could on the 8000 but i’ve got to think with all that power they can utilize it for even more robust effects, such as granular synthesis..

 

 

[Puppeteer]

That’s the thing.  Somethings are simply easier to do on software and dragging and dropping a sample into a window is definitely not what the H9000 was built for.  You have to record the sample in each time.  I’m interested in what new directions the H9000’s unique tools can take things. Most granular tools don’t process live audio.  Robert Henke’s Granulator allows you to press a button to sample a live input as the new source for the tool, and that’s definitely achievable, along with modulating loop points, playback speed (from -100% to +100%), though unfortunately you can’t directly modulate the playback head, ala Transwave.

One cool thing the H9000 can do is nest processing inside loops of other processing, and send signals from unusual places to new processing architectures.

I’m enjoying the challenge of building completely new effects, especially stuff that pushes the boundaries of stability, like feedback algorithms.  I find VSIG very intuitive and quick to build stuff. Probably something to do with 30 years of programming Kurzweil’s, which have a similar type of modular method of wiring things together.

[vanceg]

After looking into building a Granular algorightm from the existing modules in VSIG for several years, I had personally come to the conclusion that it was going to be prohibitively difficlut to really built an effectvie grainular processor this way.   There is no doubt whatsoever that the H9K (or H90) could handle the processing.   It just strikes me that it would make a LOT more sense and make the process a whole lot more straight ahead if we had a purpose built granular delay module in VSIG that we could work with.

This would essentially be a delay with an adjustable buffer and controls to allow multiple reads from that buffer at arbitrary speeds.   From there we could build a pretty effective set of granulators, with a whole raft of different interfaces, designed for different purposes, using existing VSIG contorl modules,

Alternatively, a full granular algorithm ould be developed aroudn a module like that, with a pretty standard set of granular type controls – Typiclally:  Buffer Duration, Grain Duration,  Number of Grains Playback start point, Playback speed, Envelope shape and then ‘random’ variation on each of those controls.

I really think that having a Grain Delay module in VSIG that we can build around in VSIG would be delightful and would open up a whole new world of processing possibilites which would fit nicely with the Eventide ethos.

I’ve been lobbying for this for QUITE a while as our fine friends at Eventide HQ will attest to.  😉

[parkerfly81]

vanceg wrote:

I really think that having a Grain Delay module in VSIG that we can build around in VSIG would be delightful and would open up a whole new world of processing possibilites which would fit nicely with the Eventide ethos. I’ve been lobbying for this for QUITE a while as our fine friends at Eventide HQ will attest to.

+1 for this!

seems that with all the power the H9000 has it must be possible…!

[afewster]

+1

[stevepretty]

+1

[vanceg]

One might think that since granulation is really just a matter of playing back multiple short segments of audio from a buffer, each with its own pitch, duration and amplitude envelope, thst a polyphonic Sampler module could be repurposed to become a granulator.

I would think that we’d be less interested in dropping existing soundfiles as source material, and more interested in using a real-time buffer to record into and play out of.  Why would we concentrate on creating a file based granulator when software does that so well – I’m of the opinion that we’d really want this to be based on a real-time buffer where we Write into and read from the buffer “simultaneously”

I have done some experimentation with using the Sampler modules as granulators whereby I would read into a buffer, stop the recording, and then play back short samples of audio from that buffer.
This approach holds some promise since the Sampler modules will easily play back multiple copies of audio from arbitrary points in a buffer, at arbitrary pitches and can apply an ADSR envelope to each sample (grain).
That said – creating a control mechanism in VSIG capable of triggering  and controlling the playback of many many samples at the same time,  each with its own start and end point, speed/pitch, envelope, is a daunting task at best. My attempts using the Sampler modules were challenging to say the very least – what became clear to me was that I was Simply trying to shoehorn the Sampler module into a role it just wasn’t designed for.

It became clear to me thst a dedicated GrainBuffer/GrainPlayer module was really the way to go.  Writing a module that can be run in VSIG is beyond my programming capabilities.  I have considered doing this in max/msp where such modules DO exist, and then porting this I’ve too run on the h9000 but that has some drawbacks as well, the largest of which is that it is the. Essentially a stand alone algorithm and can’t be reused by others In VSIG.

It strikes me thst Granulation is an effect time that Eventide  would excel at. It also seems to fit squarely in the creative/technological wheelhouse that Eventide has defined and been the market leader in through the years.  It’s the most “Eventide” type of effect I can think of: Time/buffer based processing, deals with pitch, near endless creative potential, capable of creating an enormous range of effects from one core concept/algorithm (everything from pad sounds to glitchy  delays to reverb like processing or even sample chopping effects) and (almost) nobody else is doing it in a real time processor format.

I can’t stress more how much I think this would be a great move.  Especially if we’re able to get this to run on the H90 as well (which I assume would. E entirely possible).  I can image a whole family of algorithms thst could come out of this, or one algorithm with such a wide range of capabilities that it could be used for many purposes.

clearly a strong advocate of this idea!

[parkerfly81]

vanceg wrote:

I would think that we’d be less interested in dropping existing soundfiles as source material, and more interested in using a real-time buffer to record into and play out of. Why would we concentrate on creating a file based granulator when software does that so well – I’m of the opinion that we’d really want this to be based on a real-time buffer where we Write into and read from the buffer “simultaneously”

 

totally agree with this.. and think an important part of granular is the ability to be able to monitor the live input and not have to record something and then play it back – although, it is also great to have the ability to also play back from a saved sample (or ideally multiple samples..) as well.

[Puppeteer]

I’ve been doing some more work on the Pitchy Granulator that I demonstrated above.  I’ve implemented MIDI and also a freeze function that will grab the current window on a MIDI note on and hold it while a note is held (and it’s monophonic, so you should be able to repitch it on the fly).  The freeze can be triggered with MIDI note, without using the note number for pitch determination.

This is the current flow sheet of it (Orange arrow is MIDI, dotted is control). It’s not possible to load or save samples except by baking them into the vsig code, but you can sample a live input by freezing it. I’m out of the studio at the moment, so won’t be able to export the algorithm and do final bug testing for a couple of weeks, but this hopefully meets a bunch of the wishes in the thread above.

If there are any cool features or changes that you can imagine over the next 2 weeks, let me know and I’ll see what I can do.

I’m also going to implement a different approach to granular with the audio sampler which will allow for  real time sample reversing and pitching, but that will be for another day.

Here’s the current Emote View

Attachments:

You must be logged in to view attached files.

[Author]

Posts