Any Ai or Software that can take in many samples/drum loops etc, sift thru them and accurately output them as one-shots?

[Brisbot]

UPDATE: I have found some promising software called 'AudioStellar'. It not only does a great job of accurately chopping samples, but it will also map them by similarity, similar to Atlas, however unlike Atlas it has some very unique and fun ways of triggering/manipulating samples. BTW i am not sure if someone mentioned it to me on here or elsewhere.

https://audiostellar.xyz/lang/en/index.html

I would love software that could accurately:

1. Take in many samples

2. Cut them up accurately

3. Then output them in a new folder as one-shots.

The reason I want this is due to Ai plugins like Atlas that will map all your one-shots accurately by how similar they are. So if I took like EVERYTHING, all my samples, turned them into one shots, I will now be able to find samples that sound most alike. That would be a game changer as finding similar bits of audio would be a breeze. Whether it be similar drum samples, or 1 or 2 second glitch bits. Would be AMAZING for layering.

On a side note I've already used Atlas to make some awesome monster kicks by layering 16 kicks that are very similar and creating these beast kicks.

edit: I feel like I may have asked this already, but I can't find a thread. Not sure if I asked it here or elsewhere.

Edited January 9, 2023 by Brisbot

[Brisbot]

On 12/22/2022 at 10:56 AM, Summon Dot E X E said:

I asked something similar here: 

 

Did you get any good results?

[zlemflolia]

use c++

https://docs.fileformat.com/audio/wav/

[zlemflolia]

if u have an iterator that calculates in realtime or via sliding metrics various statistics about the audio stream then you can note segment time ranges and extract them.  if you can calculate filtering criteria you can filter samples down later, or perform sample search using metrics to compare timbre

[zlemflolia]

i had this idea once but i never followed through.  once you do that and can define a similarity metric, then you just need to do it in reverse and sample match against an existing song to "remix" it and you can randomly iterate through samples until you find a setting you like, to reorient the track you're remixing sonically without changing it fundamentally.  a type of glitch sound effect but it can also be used for various purposes, like humming any song and having it remixed into synths afterwords or beatboxing and having it turned into real drums. 

[zlemflolia]

with an audio to midi converter this is made easier, by filtering for sequences of certain types of midi events and sampling for the durations of the selected sounds. 

[exitonly]

Myriad is a batch sample processor that, among other things, has an "Extract at thresholds" action. With Myriad you can set up workflows of processing so if you wanted to  extract all of these files and then normalize them, conver sample rate, and then rename them based on the detect base pitch of the sound then you can.

https://zynaptiq.com/myriad/

AI isnt going to be needed for simple transient detection. I could see someday hopefully being able to pull out drum or other instruments (i.e. have AI generate a simulated extraction) from a complex sound would be amazing. For day to day drum slicing, transient detection is often fine.

 

[ignatius]

when i made sample libraries of one shots i had to use more than one program since there isn't one single bit of software that does all the things. i think i used /strip silence/ -> make new audio files out of separate regions in DP or something. but also did some stuff in live because when ya edit a bit of audio you can make it into its own sample and it goes into the subfolder in teh sessions called "processed audio" or whatever it is so then it's easy to dig out the samples. 

but.. if any DAW or editor would just do 'strip silence-> make new audio files' and save to a folder it'd be way easier. i think there's a script or something but i forget. been a minute since i spit out a bunch of one shots. 'renamer' was great for renaming things though. 

[Brisbot]

On 1/7/2023 at 2:04 AM, ilqx hermolia xpli said:

use c++

https://docs.fileformat.com/audio/wav/

Thanks for the recommendation. I've always wanted to get into programming if only because I have all these workflow ideas that I can never find a plugin or software for. Though that would be a big time commitment, lol.

I do also have Melodyne which does a decent job at what you are talking about. I wish though this kind of software was more accurate.

I have found some promising software worth checking out to anyone reading this. It's called 'AudioStellar'. https://audiostellar.xyz/lang/en/index.html

On 1/7/2023 at 10:37 AM, ignatius said:

when i made sample libraries of one shots i had to use more than one program since there isn't one single bit of software that does all the things. i think i used /strip silence/ -> make new audio files out of separate regions in DP or something. but also did some stuff in live because when ya edit a bit of audio you can make it into its own sample and it goes into the subfolder in teh sessions called "processed audio" or whatever it is so then it's easy to dig out the samples. 

but.. if any DAW or editor would just do 'strip silence-> make new audio files' and save to a folder it'd be way easier. i think there's a script or something but i forget. been a minute since i spit out a bunch of one shots. 'renamer' was great for renaming things though. 

So if you are interested I've found software that kinda gets at what I am wanting to do. And goes a step further by doing what Atlas does and mapping one shots by similarity. It's open source. https://audiostellar.xyz/lang/en/index.html

[Brisbot]

On 1/7/2023 at 9:36 AM, exitonly said:

Myriad is a batch sample processor that, among other things, has an "Extract at thresholds" action. With Myriad you can set up workflows of processing so if you wanted to  extract all of these files and then normalize them, conver sample rate, and then rename them based on the detect base pitch of the sound then you can.

https://zynaptiq.com/myriad/

AI isnt going to be needed for simple transient detection. I could see someday hopefully being able to pull out drum or other instruments (i.e. have AI generate a simulated extraction) from a complex sound would be amazing. For day to day drum slicing, transient detection is often fine.

 

Myriad DOES seem like it would work as well. Do you know any other good Batch Processors? The list of things it can do is very impressive. And needed

Edit: Ah it's just for Mac it seems. But this kind of plugin is definitely in the right direction.

Edited January 9, 2023 by Brisbot

[ignatius]

3 hours ago, Brisbot said:

Thanks for the recommendation. I've always wanted to get into programming if only because I have all these workflow ideas that I can never find a plugin or software for. Though that would be a big time commitment, lol.

I do also have Melodyne which does a decent job at what you are talking about. I wish though this kind of software was more accurate.

I have found some promising software worth checking out to anyone reading this. It's called 'AudioStellar'. https://audiostellar.xyz/lang/en/index.html

So if you are interested I've found software that kinda gets at what I am wanting to do. And goes a step further by doing what Atlas does and mapping one shots by similarity. It's open source. https://audiostellar.xyz/lang/en/index.html

thanks for the link. that app looks interesting. will check it out asap. 

[Brisbot]

you 

On 1/9/2023 at 8:35 AM, Summon Dot E X E said:

I'm trying out AudioStellar on Windows now. It looks promising. It can split up a WAV into slices prior to its processing/grouping, so that's exciting.

I'm still trying to figure out how to play it like in the demo videos, but it automatically spit out all the slices in a new folder, so that does exactly what I wanted!

You ever incorporate this into your music or anything? I have gotten some great results running audiostellar thru FX and granularizers as it picks similar sounds close to each other so the granularity tends to fit texturewise.

On 1/9/2023 at 7:16 AM, ignatius said:

thanks for the link. that app looks interesting. will check it out asap. 

You ever try Audiostellar? I can't recommend it enough.