Amaranth audio Cycle, your impressions ?

VST, AU, AAX, CLAP, etc. Plugin Virtual Instruments Discussion
Post Reply New Topic
RELATED
PRODUCTS
Cycle

Post

No a lot of people, including myself at first, were confused by this. The imported sample is just meant to be used as a reference from which you draw your own waveform (bit like tracing paper) - the sample does not make a sound itself intentionally, and there is no resynthesis going on, it's entirely synthetic.

Post

aMUSEd wrote:No a lot of people, including myself at first, were confused by this. The imported sample is just meant to be used as a reference from which you draw your own waveform (bit like tracing paper) - the sample does not make a sound itself intentionally, and there is no resynthesis going on, it's entirely synthetic.
That's the part which disappointed me the most actually.

Post

aMUSEd wrote:No a lot of people, including myself at first, were confused by this. The imported sample is just meant to be used as a reference from which you draw your own waveform (bit like tracing paper) - the sample does not make a sound itself intentionally, and there is no resynthesis going on, it's entirely synthetic.
OK. Now i see why i don't get a sound after importing a sample and playing notes but only when i use the play button.

I get the impression that while spending more time with Cycle it gets more confusing than more clear. I also get the impression that spending more time with it is not really worth the effort.
While there are some nice presets in terms of doing your own sounds this currently seems to be a kind of nightmare.

It looks like i have some other synths where i get comparable and/or better results easier and faster even if there are maybe a few sounds that are not so simple to be replicated.


Ingo
Last edited by Ingonator on Tue Mar 25, 2014 7:47 pm, edited 2 times in total.
Ingo Weidner
Win 10 Home 64-bit / mobile i7-7700HQ 2.8 GHz / 16GB RAM //
Live 10 Suite / Cubase Pro 9.5 / Pro Tools Ultimate 2021 // NI Komplete Kontrol S61 Mk1

Post

Yeah, but I'm pretty confident automated resynthesis will be coming down the road... It's like all the parts are already in place, it just needs the code to link it up. For a first release from a new developer it's incredibly polished, especially considering how deep it is.

I for one like a lot of the presets, even if I do feel that Cycle's true power has yet to be fully unleashed. A lof of the presets so far seem to follow very 'traditional' tonal guidelines. Considering what you have access to with vertext modulation some pretty amazing things should be possible. The 'Cheat Sheet' has made it much more accessible IMO, I can't wait to see what more talented people than me can do once they get head around it.

I also can't wait for the distortion/IR modelling effect (Firestorm). If it has the vertext modulation on an IR from the Synth it should be amazing. 8)

Post

Interesting thoughts. I'm in the process of bringing up to level the help/tutorials side of things.
It takes a lot of time though, especially when its not really my area of expertise and I don't have a great workflow for creating that kind of content.

Cycle's workflow is meant to be a more scientific, predictable method of sound design. I've done some deep synthesis stuff with a few other general synths and I've struggled with how hard it is to create particular sounds. How can you get a reed sound with a 6x6 FM matrix? It's probably there, but it's lost in a myriad configurations.

The sample analysis function is to give you a scientific, measurable point of reference. If you're after a sound you hear on youtube, you can record it, analyse it, and have it in a matter of minutes. That's the goal of it at least, and with familiarity with Cycle this process does perform well.


I finished a few videos that should help along the interested:
https://www.youtube.com/user/AmaranthAudio
Image

Post

On the subject of casual sound design, due to Cycle's open-ended nature you do need purpose instead of impulsiveness. You can liken it to Photoshop... there's a million ways to make a nasty mess quickly because it doesn't limit you to rigid aesthetic boundaries.
Image

Post

DavenH wrote: The sample analysis function is to give you a scientific, measurable point of reference. If you're after a sound you hear on youtube, you can record it, analyse it, and have it in a matter of minutes. That's the goal of it at least, and with familiarity with Cycle this process does perform well.
I think what people are wanting though is for at least some of that process to be automated, after all while it is possible to import a wave and then draw an envelope over it to capture something of what it sounds like, surely that is also something the computer could do, and possibly more precisely? Then you could edit it further, it would be closer to resynthesis I suppose (or a sort of midway point between pure resynthesis and synthesis) but could take some of the initial clunkiness and imprecision out of the process.

Post

I can take a picture of a building with my camera or I can sit hours in front of it with pencil and paper to capture the geometrical structure of it.

Post

Sampleconstruct wrote:I can take a picture of a building with my camera or I can sit hours in front of it with pencil and paper to capture the geometrical structure of it.
That's what recordings and samplers are for. If you want flexibility, you need to abstract the data into a simpler form. It's a 10 megapixel photo of a building vs it's architecture CAD file.

aMUSEd wrote: I think what people are wanting though is for at least some of that process to be automated, after all while it is possible to import a wave and then draw an envelope over it to capture something of what it sounds like, surely that is also something the computer could do, and possibly more precisely? Then you could edit it further, it would be closer to resynthesis I suppose (or a sort of midway point between pure resynthesis and synthesis) but could take some of the initial clunkiness and imprecision out of the process.
Something I intend to do. :)
Image

Post

DavenH wrote:
Sampleconstruct wrote:I can take a picture of a building with my camera or I can sit hours in front of it with pencil and paper to capture the geometrical structure of it.
That's what recordings and samplers are for. If you want flexibility, you need to abstract the data into a simpler form. It's a 10 megapixel photo of a building vs it's architecture CAD file.

aMUSEd wrote: I think what people are wanting though is for at least some of that process to be automated, after all while it is possible to import a wave and then draw an envelope over it to capture something of what it sounds like, surely that is also something the computer could do, and possibly more precisely? Then you could edit it further, it would be closer to resynthesis I suppose (or a sort of midway point between pure resynthesis and synthesis) but could take some of the initial clunkiness and imprecision out of the process.
Something I intend to do. :)
That would be interesting. I guess the CAD analogy is a good one, pure resynthesis tends to generate large data messy files, this would be more like vectorising the data to create a wireframe model that can then be subjected to various mathematical transforms.

Post

DavenH wrote:That's what recordings and samplers are for. If you want flexibility, you need to abstract the data into a simpler form. It's a 10 megapixel photo of a building vs it's architecture CAD file.
Except that with sound, given the right methods of analysis, you pretty much have access to all the information you need. It's not really a good analogy: there's nothing hidden behind a facade. ;)
DavenH wrote:Something I intend to do. :)
:tu:

Post

Breeze wrote:
DavenH wrote:That's what recordings and samplers are for. If you want flexibility, you need to abstract the data into a simpler form. It's a 10 megapixel photo of a building vs it's architecture CAD file.
Except that with sound, given the right methods of analysis, you pretty much have access to all the information you need. It's not really a good analogy: there's nothing hidden behind a facade. ;)
That not quite the thrust of the analogy, which is that flexibility is inversely proportional to the size of data.

Think of 3D animation. You can edit images in photoshop with filters, add noise, smudge it, fade in other images, and so on. But would you animate a scene in photoshop?
No one does that, because it's too much data to orchestrate. When you have a 3D model made of a small number of points and lines so it's many degrees more malleable. You animate the flexible, defining data for the scene, then render it to a series of images.
Image

Post

DavenH wrote:That not quite the thrust of the analogy, which is that flexibility is inversely proportional to the size of data.

Think of 3D animation. You can edit images in photoshop with filters, add noise, smudge it, fade in other images, and so on. But would you animate a scene in photoshop?
I know what you mean, but then you wouldn't take a skeletal Max animation, import it into Photoshop and work with it there either.

Analogies aside, it should be possible to get the essence of the waveform captured even without extreme detail and get practically the same result, if not better, as tracing it. That would be a question of resolution: deciding how much detail you want to capture. Think of it as "waveform mocap".

To me the tracing part of capturing waveforms in Cycle is a rather laborious "boring bit". Even if rounded-off for convenience and manageability, the interesting part is what can be done with the waveform afterwards.

Post

Yeah I get it. I'm getting some good ideas about how to do the auto trace. :)

What I want is if you have sample sets like these:

Sax soft [c1 ... c6]
Sax med [c1 ... c6]
Sax hard [c1 ... c6]

... that with some dialog Cycle would figure out not only the time-wise tracing but also the key scale and modulation topology. A challenge is that for the topology to connect fluidly each trace needs the same resolution, yet more complicated shapes need more points than the simple ones, so the latter would be "overdefined" to some extent if resolution was constant.

It's a failing of the curve design; if the vertices behaved like gravity bodies on a rubber sheet, the waveform surface would still be continuous without a fully connected topology and you could add points only to the regions that needed more detail. The way it is, any region that requires more points sets the lower limit on the resolution everywhere. Hope I can figure out the math to do that while keeping the nice properties of the current curve design.



btw, no need for anyone to hold their criticism because I entered the thread, 'sall good. :wink:
Image

Post

Ingonator wrote:
UPDATE:
Just noticed that there is also a tutorial about how to create a certain sound from scratch:
http://www.amaranthaudio.com/tutorials/modelWave.php

Ingo
I´m having a headache just looking at these pictures. I wanna create sounds and make music, not having a quantum mechanics class.

Post Reply

Return to “Instruments”