Hi,

please remove the sample length limitation of the imported audio samples. It absolutely kills any joy for serious sound design workflow! Especially for cinematic, evolving work.

And I cannot believe that there is no function to record live audio input, eg. from other, parallel sound design processes, singing etc..? Yes, I know there is rolling sampler which makes it a bit quiker but.. common!! Live recording is a must in 2023, especially for this plugin!

Other than that, great product, congratulations!

the sample length is limited for a good reason, or so i would imagine: this is a two oscillator synth with only a handful of modulations internally. long, evolving sounds can only go so far here without external modulation, not to mention the resources and time required to process longer sounds. once it hits a certain point, nothing internally can replicate the sound any further. it has just about all of the limitation of synplant 1 in that regard.

I also think that longer samples might not work with the synths internal feature set or not with the neural network or so. Also regarding your other point in the title: "live input"? You mean like giving it a live input it has to analyse and immediately generate a patch from it? I'd say it's not even possible on a logic basis. I mean: what should the Genopatch magic do with an ongoing chaning sound? It makes no sense. ... but probably I am not quite getting how you meant it, hm? Maybe you could ellaborate a bit how you meant it?

- Manuel Senfft wrote:
I also think that longer samples might not work with the synths internal feature set or not with the neural network or so. Also regarding your other point in the title: "live input"? You mean like giving it a live input it has to analyse and immediately generate a patch from it? I'd say it's not even possible on a logic basis. I mean: what should the Genopatch magic do with an ongoing chaning sound? It makes no sense. ... but probably I am not quite getting how you meant it, hm? Maybe you could ellaborate a bit how you meant it?

He probably wants to record audio from differenct sources to synplant directly to have it in the genopatch as fix sample. From there on you can select an area of the sample and let genopatch do it's usual magic.

In FL I would just record my things into Edison and drag the sample from there to Synplant (when drag&drop hopefully works in the next update).

This makes more sense, I see. Thanks for explaining.

He probably wants to record audio from differenct sources to synplant directly to have it in the genopatch as fix sample. From there on you can select an area of the sample and let genopatch do it's usual magic.

Yes, exactly. Every additional step is a workflow killer if you do real-time sound design, voice-recordings etc..

you're asking for the impossible here. this is very likely not the tool you are looking for. as i said earlier, it is a two-oscillator synth with a handful of modulation sources. it analyzes sound using a trained ML model and creates patches that are as close as possible given the small number of parameters in the synth. it can't simply do more beyond that, and certainly not in real time. it needs the audio in advance to even begin generating the parameter values (this is not even taking into account the multiple iterations it generates, which you then must audit and select) and the parameter values are static.

what you are asking for is a totally different thing on every level and would require infinitely long modulation sources with an infinite number of points due to variation over time. the fact that it can do what it currently does is incredible on its own.

this is, above all, synplant. it is everything it always has been, but with a new tool to generate patch parameters from an audio file. i would imagine that if you were familiar with how the synth works then you would know that this is simply not possible - and if you are interested in sound design, it might be good to master that from the ground up first.

no... this is not what I wrote.
What I mean is, a button in Synplant (next to import wave) that "records" what I am feeding it in eg. my voice or other live input.
The process to first, record my voice, bounce it and then import it or drag/drop in Synplant is just too many steps, thus a workflow killer.
I don´t want Synplant to process anything in real time. Just the addition to offer recording audio instead of just importing it.

then not so much the "real-time" sound design; i'm not sure what that term means here otherwise, aside from real-time, as in, performance, here and now, etc. non real-time is much more feasible if that's your end goal (and i can see it being implemented down the line with not much pain involved). longer patches are definitely out the window though.

- enCiphered wrote:
Hi,
please remove the sample length limitation of the imported audio samples. It absolutely kills any joy for serious sound design workflow! Especially for cinematic, evolving work.
And I cannot believe that there is no function to record live audio input, eg. from other, parallel sound design processes, singing etc..? Yes, I know there is rolling sampler which makes it a bit quiker but.. common!! Live recording is a must in 2023, especially for this plugin!
Other than that, great product, congratulations!

Not sure what plugins you are using, but I have yet to come across one that supports recording samples into the plugin. Now drag and dropping recordings from the DAW or sample library like Serum without using File Explorer sounds reasonable but still your sentiment in my opinion is overstated. Sure maybe one day we will all be able to do everything super easily without any extra steps, just record your voice and output a banger song/synth no work required; but for now try to appreciate the tools we have today. I highly doubt this "kills" your workflow, maybe just get faster? More efficient? Idk. Without this tool it wouldn't even be part of your workflow, so again, try to appreciate the tools you have.

- Manuel Senfft wrote:
I also think that longer samples might not work with the synths internal feature set or not with the neural network or so. Also regarding your other point in the title: "live input"? You mean like giving it a live input it has to analyse and immediately generate a patch from it? I'd say it's not even possible on a logic basis. I mean: what should the Genopatch magic do with an ongoing chaning sound? It makes no sense. ... but probably I am not quite getting how you meant it, hm? Maybe you could ellaborate a bit how you meant it?

You technically can get around the 2 second limit by configuring Genopatch.

I would post the full explanation, but the sentiment of OP has limited my usefulness. I'm sure most will be able to do the extra work to find the parameter and change it or find my other post when/if it is approved.

I would imagine there is a reason Genopatch comes with ≤2 second sample limitation, results may vary

- emma essex wrote:
then not so much the "real-time" sound design; i'm not sure what that term means here otherwise, aside from real-time, as in, performance, here and now, etc. non real-time is much more feasible if that's your end goal (and i can see it being implemented down the line with not much pain involved). longer patches are definitely out the window though.

Exactly, if the team intended to or even could reasonably add the functionality for recording voice into the plugin, they probably would have or maybe it is coming (doubt). You CAN however get away with longer sample length though by configuring Genopatch, but your concerns seems valid. The size of input data that an AI is trained on is important, and they probably had he best results or purposely went with samples under 2 sec, hence the limitation.