just bought this. brilliant work. normally I am really skeptical of "AI" enhanced music plugins but this actually implements machine learning in a practical and useful way. if I were Ableton I would be throwing you a nice paycheck and bundle this with Live 12!
I love this. What an incredible and beautiful mind you have. Also got your Inspired by Nature pack with Ableton 12, Tree Tone amazes me! Can't wait to see what you come up with next ;) Thanks for the inspo, Dillon ;)
Hey it should, did you install it in the correct place as per the read me? Make sure it is exactly correct. Let me know if you have further issues, maybe email dillonbastan@gmail.com
@@nrosko Ok we are talking by email already but I want to leave this here so people know, there is a bug at the moment which I think is what happened for you(?): So if you create a new Live set, and make a recording in it, and then drop that recording into Coalescence BEFORE you save the Live set, then it does not recall the sample (this is because the sample at that point is in a Temp location). So I will try and fix this bug but it may or may not be possible because I am unable to use the standard Live path decoding objects for this project for reasons... I will let you know if I fix it though! In the meantime, to work around this issue just make sure to save your Live sets for the first time before dropping recordings recorded in the Live set into Coalescence. If you happen to do that, then does drop and swap out the same sample AFTER you saved the Live set for the first time.
Wellll that depends on what samples are fed into the network. So you have a few options for feature analysis and the closest to what you want is the Chroma mode which for each sample slice gives you the intensities for each tone in a 24 step octave. So in that mode it will group slices based on their intensities in the 24 step octave (for example to help you imagine, lets say instead it was a typical 12 step chromatic octave and you have a sample slice that had a lot of harmonic content in C tone and some in say F#, then its intensity values would look something like (C being the first tone/value): [0.98, 0, 0, 0, 0, 0, 0.5, 0, 0, 0, 0, 0]. Ok so hopefully that gives you a clearer picture of what the chroma values are like. With that understanding then maybe you can see that the ability to group samples based on their key depends a lot on the samples and I don't think it will cleanly do that (and also most harmonic content isn't so clean like the example). Now what it will do is definitely group the tones together, such as sample slices with a lot of harmonic content in C will be group etc. So you will get clusters of sample slices that have the same or similar fundamental, but as far as keys themselves grouped it's unlikely it would cleanly do that. I was considering adding a mode like that and I still might, but not sure if it will work well as is without other changes because notes can belong to different keys but they can only cluster in one space!
You are well on your way to making a Kontakt-like sampler which automatically slices n organizes audio by pitch n velocity, making virtual instrument creation super easy. You could make a whole community around that, like Pianobook has done.
This is definitely on my list of devices I'd like to get. Does it come with documentation too? I'm one of those types that will read docs through till the end to know everything haha
after playing around with this device for several weeks now, i finally realize how genius it is and how helpful it is if one wants to create unheard-before midi-compositions. hey Dillon, please make the pitch of the sample slices mapable to lfo modulation devices. thx
Do you mean be able to modulate individual sample pitches instead of the overall playback pitch (which you already can do)? I'm afraid I can't because then I'd have to have live UI objects for every potential sample of which there are up to 2000 (plus all the other individual sample GUI like reverse, volume, etc)
@@dillonbastan246 thanks for getting back quickly. actually i ment the individual pitch of each sample, but i see the problem with live UI objects with that many/2000 instances. honestly i was maybe to confused by the complexity of this device/coalesce because i overlooked that there's an option to modulate the pitch in the playback-tab, which is great and is good enough for my needs. thanks Dillon to point this out. and i guess it's easy to copy the identical instance of coalesce on different tracks to simulate the ability to modulate the individual pitch of each sample and mute all the non-neede ones. will try to bring ableton to it's limit with 2000 tracks :) all the best and thanks again for your great work, Illya from berlin
@@Illkacirma Cool! FYI unless the samples are short one shots you will fill the network long before you reach 2000 samples because it only has 2500 cells to hold slices!
What you created looks amazing! but im wondering your specs , how heavy does this run on your system? my system is kinda dated unfortunately. Ryzen 1700x w/16gb of RAM :(
Hey just review the manual or sections of the video about how the network works and that should answer your question. Here is a deep dive to really cement how its working: czcams.com/video/RpEj7_RKQDo/video.html&ab_channel=DillonBastan
@@dillonbastan246 Sure! In Skatart, when you play your generated corpus of sounds, you encounter different modes. Instead of waiting for a point to conclude, the sound already plays when it's triggered on the XY pad, without any "voice stealing," allowing the voices to overlap. This overlap contributes to a consistent and sustained texture rather than a choppy one. czcams.com/video/cWXdTlu_n44/video.html
@@ok.frasco its a polyphonic instrument first of all so you can overlap the voices in a number of ways depending on which playback mode you are in. There are quite a number of ways to playback the "points". But dealing specifically with a single playback voice, if you were to loop playback with one voice and move to another point the voice's set loop length would finish before moving to the beginning of the next point. If you are in the XY playback mode specifically, you also can have the option to latch to a single point if desired. I'm not a user of skatart so i cant really give you a good comparison of the workflows. I will make a deep dive on playback modes of Coalescence eventually... I also have a deep dive on how the spectral analysis and neural network works here if that's helpful: czcams.com/video/RpEj7_RKQDo/video.html&ab_channel=DillonBastan
hey you just choose the chroma mode which is on by default, check out this video for clarity on the network and slicing czcams.com/video/RpEj7_RKQDo/video.html&ab_channel=DillonBastan
??? I think you are in the wrong place, this is the preview: czcams.com/video/MvL1mXXch1Y/video.html&lc=UgyCTd14xgbdvY3PlDp4AaABAg&ab_channel=DillonBastan
I really hope Dillan will never accept Ableton live contract. For me it would mean that such a device will never exist . Or in a economical "acceptable" plan. Like one new feature of this device would exist in each new version of live . Then the whole device would be "complet" in many years ... And if Dillan leave because of his creativity bloked by stupid polities , they will just change the browsing system to make it new... no thanks .
omg just show me what sounds it can make I’ve listened to you talk for like 5 minutes without hearing a thing lol. For future reference if you wanna advertise your M4L device better just get to the point with the sounds that it makes (that’s what we are all here for) and then provide the explanation after for those who want to listen to it
just bought this. brilliant work. normally I am really skeptical of "AI" enhanced music plugins but this actually implements machine learning in a practical and useful way. if I were Ableton I would be throwing you a nice paycheck and bundle this with Live 12!
for real, this would be a killer device to include in 12 suite
as long as Dillon makes m4ls I will never leave Live 😍
insane work, dillan! congratulations and thanks for the release!
so many uses. awesome work as usual. instant buy.
This is going to be crazy helpful for me, thank you!
wow! incredible device! great job Dillon.
Instant buy your product . SOOOO incredible . Such a genius !
00:00 Intro
02:13 Point Mode
05:12 Rings Mode
09:23 Paths Mode
12:00 Sample Tab
15:48 Network Tab
24:56 Playback Tab
31:23 Modulation Tab
34:54 Input Tab
Thank you FemiFleming
Thanks
Such a clever and inspiring device, dragged me out of a deep creative rut in minutes.
I needed a muse and i got one for Christmas!
Phenomenal work!
I love this. What an incredible and beautiful mind you have. Also got your Inspired by Nature pack with Ableton 12, Tree Tone amazes me! Can't wait to see what you come up with next ;) Thanks for the inspo, Dillon ;)
Thank you for this video ..I was a little lost but this was perfect
What an amazing plugin!
Legendary instrument Dillon!
holy shit dillon... this is crazy
Stunning!!!
I think this is brilliant :)
Thank you I can't wait to try it
Impressive work
Having some crazy fun with this, thank you.
Dose not recall the sample from a project. shame i lost a good track.
Hey it should, did you install it in the correct place as per the read me? Make sure it is exactly correct. Let me know if you have further issues, maybe email dillonbastan@gmail.com
If it is not installed correctly then it may have issues recalling samples, also protected samples don’t work in it atm
@@dillonbastan246 Thanks i did not do that, fallen fowl to my own impatience.
@@nrosko Ok we are talking by email already but I want to leave this here so people know, there is a bug at the moment which I think is what happened for you(?): So if you create a new Live set, and make a recording in it, and then drop that recording into Coalescence BEFORE you save the Live set, then it does not recall the sample (this is because the sample at that point is in a Temp location). So I will try and fix this bug but it may or may not be possible because I am unable to use the standard Live path decoding objects for this project for reasons... I will let you know if I fix it though!
In the meantime, to work around this issue just make sure to save your Live sets for the first time before dropping recordings recorded in the Live set into Coalescence. If you happen to do that, then does drop and swap out the same sample AFTER you saved the Live set for the first time.
lets go DILLON !!
this could be used to analyze samples and find the key they are in with the network! thanks dude. wild stuff i’ll be buying.
Wellll that depends on what samples are fed into the network. So you have a few options for feature analysis and the closest to what you want is the Chroma mode which for each sample slice gives you the intensities for each tone in a 24 step octave. So in that mode it will group slices based on their intensities in the 24 step octave (for example to help you imagine, lets say instead it was a typical 12 step chromatic octave and you have a sample slice that had a lot of harmonic content in C tone and some in say F#, then its intensity values would look something like (C being the first tone/value): [0.98, 0, 0, 0, 0, 0, 0.5, 0, 0, 0, 0, 0]. Ok so hopefully that gives you a clearer picture of what the chroma values are like. With that understanding then maybe you can see that the ability to group samples based on their key depends a lot on the samples and I don't think it will cleanly do that (and also most harmonic content isn't so clean like the example). Now what it will do is definitely group the tones together, such as sample slices with a lot of harmonic content in C will be group etc. So you will get clusters of sample slices that have the same or similar fundamental, but as far as keys themselves grouped it's unlikely it would cleanly do that. I was considering adding a mode like that and I still might, but not sure if it will work well as is without other changes because notes can belong to different keys but they can only cluster in one space!
@@dillonbastan246 thanks for the explanation, makes perfect sense to me! cheers for the detailed reply!
Wonderful Dillon, again. ❤❤❤
Super cool
Congrats! I hope you are a millonaire already!
Haha not even close but thanks!
You are well on your way to making a Kontakt-like sampler which automatically slices n organizes audio by pitch n velocity, making virtual instrument creation super easy. You could make a whole community around that, like Pianobook has done.
M4L 🐐 also, when Dillon says “obvious” this means you need to get out your physics 2 course ware…thanks friend 🫶🖤😎🐓 bought 💰
nice
💪💥🗯🔥💭❤💨💪
.....stable....nailed.....
omfg transient sensitivity! i just realised a live feature i never knew existed! whattaday
It's just for my device. I believe Simpler has a similar parameter for their slice mode
I’m imagining this in a Daisy Patch Eurorack module 🤤 Do the learning on computer and export it. Definitely gonna ask the Daisy community about this
SOM network is pretty lightweight so chances are you can run it all on there especially if you just reduced its size
@@dillonbastan246 The Daisy can run gen~ so I have honestly no idea how close it would be possible to make it even if it wouldn't be 1 to 1.
@@jimspectre I coded the SOM in gen
This is definitely on my list of devices I'd like to get. Does it come with documentation too? I'm one of those types that will read docs through till the end to know everything haha
There is a user manual and several walkthroughs on YT
woahhhh
after playing around with this device for several weeks now, i finally realize how genius it is and how helpful it is if one wants to create unheard-before midi-compositions. hey Dillon, please make the pitch of the sample slices mapable to lfo modulation devices. thx
Do you mean be able to modulate individual sample pitches instead of the overall playback pitch (which you already can do)? I'm afraid I can't because then I'd have to have live UI objects for every potential sample of which there are up to 2000 (plus all the other individual sample GUI like reverse, volume, etc)
@@dillonbastan246 thanks for getting back quickly. actually i ment the individual pitch of each sample, but i see the problem with live UI objects with that many/2000 instances. honestly i was maybe to confused by the complexity of this device/coalesce because i overlooked that there's an option to modulate the pitch in the playback-tab, which is great and is good enough for my needs. thanks Dillon to point this out. and i guess it's easy to copy the identical instance of coalesce on different tracks to simulate the ability to modulate the individual pitch of each sample and mute all the non-neede ones. will try to bring ableton to it's limit with 2000 tracks :) all the best and thanks again for your great work, Illya from berlin
@@Illkacirma Cool! FYI unless the samples are short one shots you will fill the network long before you reach 2000 samples because it only has 2500 cells to hold slices!
Looks sooo great, any plans on vst support ?
if so, not for quite some time...
I'm waiting with bated breath!
the last part : instant buy
What you created looks amazing! but im wondering your specs , how heavy does this run on your system? my system is kinda dated unfortunately. Ryzen 1700x w/16gb of RAM :(
awesome plugin!
Is there any way i can map the xy coordinates when using the ring mode? It works fine in point mode
There is no XY coordinater in ring mode, in Ring mode the ring regions are assigned to different midi notes
Downloaded. It's super cool. You might want to fix the spelling of "Repetitions" under the Network tab ;)
Wow I've read it so many times while making the device and didnt even catch it hiding in plain sight, thanks!
ML4M4L❤
It would be easier doing patches for the nasa lol, and thanks now i know what i want for christmas
Curious - when it finds snippets to associate with keys (ie C or D) - is it finding samples that are in that frequency / note?
Hey just review the manual or sections of the video about how the network works and that should answer your question. Here is a deep dive to really cement how its working: czcams.com/video/RpEj7_RKQDo/video.html&ab_channel=DillonBastan
Hold uuuupp. Just saw the short promo and this... this can get beautifully weird. You should get 50% of what I'm
paying to upgrade to Suite... haha.
Is there a way to make sounds in the corpus overlap instead of waiting for every point to end, like fence mode in skatart?
Sorry I'm not really familiar with skatart, can you explain what you want a bit more detailed?
@@dillonbastan246 Sure! In Skatart, when you play your generated corpus of sounds, you encounter different modes.
Instead of waiting for a point to conclude, the sound already plays when it's triggered on the XY pad, without any "voice stealing," allowing the voices to overlap. This overlap contributes to a consistent and sustained texture rather than a choppy one.
czcams.com/video/cWXdTlu_n44/video.html
@@ok.frasco its a polyphonic instrument first of all so you can overlap the voices in a number of ways depending on which playback mode you are in. There are quite a number of ways to playback the "points". But dealing specifically with a single playback voice, if you were to loop playback with one voice and move to another point the voice's set loop length would finish before moving to the beginning of the next point. If you are in the XY playback mode specifically, you also can have the option to latch to a single point if desired. I'm not a user of skatart so i cant really give you a good comparison of the workflows. I will make a deep dive on playback modes of Coalescence eventually... I also have a deep dive on how the spectral analysis and neural network works here if that's helpful: czcams.com/video/RpEj7_RKQDo/video.html&ab_channel=DillonBastan
Can you make it so it slices according to pitch, so it could slice a piano loop?
hey you just choose the chroma mode which is on by default, check out this video for clarity on the network and slicing czcams.com/video/RpEj7_RKQDo/video.html&ab_channel=DillonBastan
please vst/au version :(
please vst/au version…
bastan back at it again
Neural network? What is that?
www.google.com/search?q=Neural+network
system seller device
audioz please
??? I think you are in the wrong place, this is the preview: czcams.com/video/MvL1mXXch1Y/video.html&lc=UgyCTd14xgbdvY3PlDp4AaABAg&ab_channel=DillonBastan
I really hope Dillan will never accept Ableton live contract. For me it would mean that such a device will never exist . Or in a economical "acceptable" plan. Like one new feature of this device would exist in each new version of live . Then the whole device would be "complet" in many years ... And if Dillan leave because of his creativity bloked by stupid polities , they will just change the browsing system to make it new... no thanks .
dont worry they arent offering me a job haha
omg just show me what sounds it can make I’ve listened to you talk for like 5 minutes without hearing a thing lol. For future reference if you wanna advertise your M4L device better just get to the point with the sounds that it makes (that’s what we are all here for) and then provide the explanation after for those who want to listen to it
Hi i think you are confused, this is the walkthrough, the promo is here... czcams.com/video/MvL1mXXch1Y/video.html&lc=Ugx2QnWkIlHzDq4zUTZ4AaABAg
@@dillonbastan246 gotcha thanks