Hey, I'm the guy who invented the algorithm behind The Concatenator. I can affirm that it's not just pure "rng" (random number generation for those uninitiated). If it were, there's no way we'd be getting accurate pitches. Instead, it randomly samples from the corpus and carefully promotes grains that are matching the spectrum well in a "survival of the fittest" type strategy referred to as a "particle filter." It's a classical statistical AI technique from the '90s and 2000s that people forgot about. Happy to chat more and answer more questions if anyone is curious. Lots more improvements coming down the line over the next few months!
@@ctralie did you draw any inspiration from sample brain, because there linguistic similarities… and use targeting in envelopes seems quite similar… I’m kyma and modular guy - it’s just curiosity.. because other than dsp utilization which is comparable I think you crushed it.
@@corticallarvae Oh wow thanks for saying that! I was not personally aware of Sample Brain before doing this; I'm a bit of an outsider to the whole community and mainly took on this on to help Encanti to make the kind of sounds he wanted to make (he found me on github, lol). But the objective is definitely similar. The biggest difference is we target *huge* corpora. The other difference, in my understanding, is that we go for matching the whole spectrogram, rather than trying to match derived features. This means that we get things like pitch and rhythm "for free." In my experience this leads to better sounds over a wider variety of samples, though it's more computationally expensive, so we had to play some tricks (hence the "particle filter"). But shout out still to Sample Brain and to Aphex twin..definitely spiritually similar
Dude, Encanti is a beast and literally pushed the entire field of concatenation synthesis forward in a masters thesis just for this plugin. This is 100% as epic as it sounds. Watch the other videos
Just want to add that, concatenative synthesis is a legit thing in film production and sometimes game audio too. It is used for organic, controllable ambience and texture, like footsteps that require more customization than usual. It is a really niche tool. *I am saying the underlying tech is being used in film for at least a decade, not this plugin. lol
Encanti, the guy in the tutorial video is well established in the IDM world, zebbler encanti experience is a crazy show. He used to do VJ work before producing too I believe. It makes sense they have Mr. Bill and Somatoast on the website they are affiliated artists in that corner of the scene for sure. Probably not a scam, could be next level. these guys are typically making some boundary pushing forward thinking art.
Its not random, it maps the features from the input to the features of a selection of samples from your disk. It doesn’t ‘know’ you want to make sword sounds you have to select the samples it uses. The tech is legit. I just think its very different from what we’ve seen in the past. I really wish you would have researched it more before diving in especially after criticizing creators like white sea studio for doing something similar. While I do think their plugins have a lot of room for improvement its actually very robust considering the technical challenges that go with performing this in real time. The underlying machine learning techniques involved include kmeans clustering which organizes samples according to arbitrary features like spectral similarity
yeah idk I wish he would just admit when he doesn't understand something instead of immediately hating on it. That kind of mindset gets you a fully grown native English speaker who doesn't know the word concatenate. Maybe I'm hating too hard I don't think all his takes are bad
@@zitronekoma30 I hear you, I genuinely appreciate his skepticism and caution myself. But I hope we can all realize that absence of toxic positivity does not necessarily require toxic negativity to go against it. I'm sure its somewhat due to entertainment value, I just feel bad for the developers because people who watch this channel will probably just assume its a scam when they're genuinely putting in an immense amount of effort and cost into developing something cool and unique. He complains that there's no new ideas for plugins but as soon as there is something fresh it gets discarded as a 'scam' because of the pressure to rush out regular content and bypassing research. I do think no trial isn't a great move by DataMinds but they deserve a better rep than this. Developing plugins is harder than making beats and I think we often forget that when we dismiss them as scams.
@@jack_lion I'm the inventor of the algorithm behind The Concatenator, and I appreciate you saying this! By the way, it actually uses something called a "particle filter" to guide choice of the samples. So in a way, it is random number generation, but it then does "survival of the fittest" to determine which activations are working the best and promotes those. This helps it to work at scale on huge corpora. Anyone who sees this should feel free to reach out to me if they want more info. Encanti and I also just presented a paper at ISMIR on the algorithm a few weeks ago
The CPU usage is determined by the user, mainly with the Particle and Polyphony knobs. I like to push the particles up to the CPU threshold for more accuracy, but you can keep it low.
@@encanti okay, because to run through your sample library of sound effects, and to pluck various sounds online, (and that’s not going into if it has it’s own sound effects library (for idm sounds people like Richard Devine, and Josh Eustace (Telephon Tel Aviv) would have a field day with this program) I would imagine this program would be an absolute beast and turn any cpu to mush
@@sawtooth808 Personally, I like it when software has no limit for how hard you can push the CPU. If the performance is bogging down your computer, just lower the particle knob. It's pretty much the same as the Unison knob on Xfer Serum or any given soft synth.
Seems pretty neat and useful, easy workflow. Not totally necessary and i dont think id shell out for it unless im flush with cash, but if I had it I'd play around with it
i feel like you just spew negative takes for the sake of being a contrarian, been watching your stuff for a while but in recent times you've def lacked any ability to back up your haterisms
he literaly says " I just put in it 3 folders of vocal samples" how is it it dicieving? ofc you gotta put in the kind of sound you wanna mmake, he never said it searches your hardrive automaticaly
it does generate sound based off the input. To massively oversimplify, it matches grains of sounds from the input to the most similar grains of sound from the samples you provide it
I would say the use case wasnt that great. Judging where you are taking the vst from, it makes sense that it looks scammy cause the technology wasnt that "useful" for normal beat making. Or just general production. For experimental music maker, this tool could be a god send. However, I must say there are already similar product using the same synthesis method, Concatenative synthesis, it is essentially a "one grain granulator" that uses only a very short timeframe of individual samples to match it's target. It detects stuffs like frequency and amplitude. The producers that endorses the plugin are mostly experimental music producer, some I would argue are big names in the scene. But nonetheless, not a tool aimed for mainstream.
I would funk with it, admittedly it’s gimmicky, but with the right sample folder it might cook, there’s nothing I’m aware of that can transform a beatbox well rn
you think the way he presented it, that's it not what it is. And this comes from a point that you seem to have a lot of difficulties with reading and listening what people actually are saying, you already make up idea's and depicting idea's without even listening what people actually say about it ... he never said it searches your hard drive man, he says you can use any sample library on your harddrive that you put in the concatenator as an input. This video's and others you do just seem to be attention driven to yourself and your channel, how can you put out video's like this without doing any research before? Consciuous listening and reading is not given to everyone it seems
It uses the same basic concept of concatenative synthesis but, as someone who has both, I’d say they fulfil different functions. Coalescence has features that Concatenator doesn’t and you can perform with it in different ways but Concatenator seems to map the corpus (the chosen folders/samples) a lot more accurately to the input audio. That’s an aspect I’ve really struggled with on Coalescence.
You have no idea what you're talking about. Stop spamming misinformation under every reply. Basing your opinion off a video of some guy poorly reviewing something instead of doing research isn't very smart
nobody who knows their stuff and takes music seriously would ever buy this junk it's so half-baked. if it's anything like synplant, i'd think about it. since it will utilizes actual AI. - recognize what kind of sound it's hearing (ex: a drum loop) - analyze the loop into sections (highs and lows) - collect and load sample libraries according to a theme you choose (ex: wooden perc) - replace the samples being played like slate digital's trigger
Hey, I'm the guy who invented the algorithm behind The Concatenator. I can affirm that it's not just pure "rng" (random number generation for those uninitiated). If it were, there's no way we'd be getting accurate pitches. Instead, it randomly samples from the corpus and carefully promotes grains that are matching the spectrum well in a "survival of the fittest" type strategy referred to as a "particle filter." It's a classical statistical AI technique from the '90s and 2000s that people forgot about. Happy to chat more and answer more questions if anyone is curious. Lots more improvements coming down the line over the next few months!
@@ctralie did you draw any inspiration from sample brain, because there linguistic similarities… and use targeting in envelopes seems quite similar… I’m kyma and modular guy - it’s just curiosity.. because other than dsp utilization which is comparable I think you crushed it.
@@ctralie the targeted samples seem to be longer in capability ❤️
@@corticallarvae Oh wow thanks for saying that! I was not personally aware of Sample Brain before doing this; I'm a bit of an outsider to the whole community and mainly took on this on to help Encanti to make the kind of sounds he wanted to make (he found me on github, lol). But the objective is definitely similar. The biggest difference is we target *huge* corpora. The other difference, in my understanding, is that we go for matching the whole spectrogram, rather than trying to match derived features. This means that we get things like pitch and rhythm "for free." In my experience this leads to better sounds over a wider variety of samples, though it's more computationally expensive, so we had to play some tricks (hence the "particle filter"). But shout out still to Sample Brain and to Aphex twin..definitely spiritually similar
@@corticallarvae Yes, I think that's right
Wow 😲, that's cool. Is it capable of digging your whole hard drive like he said?3:14
Dude, Encanti is a beast and literally pushed the entire field of concatenation synthesis forward in a masters thesis just for this plugin. This is 100% as epic as it sounds. Watch the other videos
Just want to add that, concatenative synthesis is a legit thing in film production and sometimes game audio too. It is used for organic, controllable ambience and texture, like footsteps that require more customization than usual. It is a really niche tool. *I am saying the underlying tech is being used in film for at least a decade, not this plugin. lol
Yep, and the advanced vocalsynths before Ai were typically concatenative too. It's really neat technology
How did they do it without plugins like this tho
Encanti, the guy in the tutorial video is well established in the IDM world, zebbler encanti experience is a crazy show. He used to do VJ work before producing too I believe. It makes sense they have Mr. Bill and Somatoast on the website they are affiliated artists in that corner of the scene for sure. Probably not a scam, could be next level. these guys are typically making some boundary pushing forward thinking art.
The Zebbler Encanti experience have legit thrown some of the best live sets i've ever seen and thats one high fucking bar. lol
@@Betbettington54 I’m in total agreement but these guys aren’t listening , the man who wrote the algorithm is in here. I sent him a question or two.
Its not random, it maps the features from the input to the features of a selection of samples from your disk. It doesn’t ‘know’ you want to make sword sounds you have to select the samples it uses. The tech is legit. I just think its very different from what we’ve seen in the past. I really wish you would have researched it more before diving in especially after criticizing creators like white sea studio for doing something similar. While I do think their plugins have a lot of room for improvement its actually very robust considering the technical challenges that go with performing this in real time. The underlying machine learning techniques involved include kmeans clustering which organizes samples according to arbitrary features like spectral similarity
yeah idk I wish he would just admit when he doesn't understand something instead of immediately hating on it. That kind of mindset gets you a fully grown native English speaker who doesn't know the word concatenate.
Maybe I'm hating too hard I don't think all his takes are bad
@@zitronekoma30 I hear you, I genuinely appreciate his skepticism and caution myself. But I hope we can all realize that absence of toxic positivity does not necessarily require toxic negativity to go against it. I'm sure its somewhat due to entertainment value, I just feel bad for the developers because people who watch this channel will probably just assume its a scam when they're genuinely putting in an immense amount of effort and cost into developing something cool and unique. He complains that there's no new ideas for plugins but as soon as there is something fresh it gets discarded as a 'scam' because of the pressure to rush out regular content and bypassing research. I do think no trial isn't a great move by DataMinds but they deserve a better rep than this. Developing plugins is harder than making beats and I think we often forget that when we dismiss them as scams.
@@jack_lion I'm the inventor of the algorithm behind The Concatenator, and I appreciate you saying this!
By the way, it actually uses something called a "particle filter" to guide choice of the samples. So in a way, it is random number generation, but it then does "survival of the fittest" to determine which activations are working the best and promotes those. This helps it to work at scale on huge corpora. Anyone who sees this should feel free to reach out to me if they want more info. Encanti and I also just presented a paper at ISMIR on the algorithm a few weeks ago
He’s using it exactly how he explained its use.
Look at the damn CPU Usage of that thing LOL
Where does it show that? I’m blind
The CPU usage is determined by the user, mainly with the Particle and Polyphony knobs. I like to push the particles up to the CPU threshold for more accuracy, but you can keep it low.
@@encanti okay, because to run through your sample library of sound effects, and to pluck various sounds online, (and that’s not going into if it has it’s own sound effects library (for idm sounds people like Richard Devine, and Josh Eustace (Telephon Tel Aviv) would have a field day with this program) I would imagine this program would be an absolute beast and turn any cpu to mush
@@sawtooth808 Personally, I like it when software has no limit for how hard you can push the CPU. If the performance is bogging down your computer, just lower the particle knob. It's pretty much the same as the Unison knob on Xfer Serum or any given soft synth.
Seems pretty neat and useful, easy workflow. Not totally necessary and i dont think id shell out for it unless im flush with cash, but if I had it I'd play around with it
Do it. Buy the 7 day trial. We all want to see. It will be either epically great or lame, but epic either way.
i feel like you just spew negative takes for the sake of being a contrarian, been watching your stuff for a while but in recent times you've def lacked any ability to back up your haterisms
Punching some stuff in and hoping for the best is not really a good way to work
he literaly says " I just put in it 3 folders of vocal samples" how is it it dicieving? ofc you gotta put in the kind of sound you wanna mmake, he never said it searches your hardrive automaticaly
This plugin would've been so much better if it generated sound based on the input instead of just picking samples.
it does generate sound based off the input. To massively oversimplify, it matches grains of sounds from the input to the most similar grains of sound from the samples you provide it
It’s targeting the samples with an envelope
Aphex Twin's Samplebrain does something pretty similar.
@@blmzndr totally said that as well, I think the samples can much longer here
I would say the use case wasnt that great. Judging where you are taking the vst from, it makes sense that it looks scammy cause the technology wasnt that "useful" for normal beat making. Or just general production.
For experimental music maker, this tool could be a god send. However, I must say there are already similar product using the same synthesis method, Concatenative synthesis, it is essentially a "one grain granulator" that uses only a very short timeframe of individual samples to match it's target. It detects stuffs like frequency and amplitude.
The producers that endorses the plugin are mostly experimental music producer, some I would argue are big names in the scene. But nonetheless, not a tool aimed for mainstream.
I wasn’t sure about this plug-in when I saw it before and I think I just ignored it after I saw the CPU usage part. And I kept moving on moving on.
What a podcast, this episode was insane
Looks like XO for Drum sounds
Wait, I tought XO IS for drum sounds?
@@AdamElteto It is. I never did try to point to say, a folder of Vocals or Horns etc... But I'm fairly sure that wouldn't work.
So it just plays a sample chromatically?
I would funk with it, admittedly it’s gimmicky, but with the right sample folder it might cook, there’s nothing I’m aware of that can transform a beatbox well rn
beardyman using objekt is the best beatbox transform I have seen
@ yeah I saw that, but isn’t that only for reason? Plus it’s beardyman he’s got all kina sauce on his beatbox
It is a rip-off of a max4Life device called coalescene from Dillon Bastan
@ interesting..I use live, so thx
@@kittycloudz813 Dillon is more or less known for his "Inspired by Nature"-Pack with Vector/Emit/Tree-Tone, which is free in Live
It’s an update to ATwins plug in same concept
It’s samplebrain yo
Encanti is a legit artist and a good guy in general. Just saying 🙂
This seems like a more convoluted version of Krotos Audio’s Reformer Pro, which is actually a cool and useful sound design tool
It's a clone of Coalescene from Dillon Bastan
I thought it was a Linux command line utility when I heard the name!
can do the same thing by feeding too many commands into my sp404.
Zing!
Encanti is a genius.
doing the same thing as th xox plugin with sorting ur sounds
you think the way he presented it, that's it not what it is. And this comes from a point that you seem to have a lot of difficulties with reading and listening what people actually are saying, you already make up idea's and depicting idea's without even listening what people actually say about it ... he never said it searches your hard drive man, he says you can use any sample library on your harddrive that you put in the concatenator as an input.
This video's and others you do just seem to be attention driven to yourself and your channel, how can you put out video's like this without doing any research before?
Consciuous listening and reading is not given to everyone it seems
Ok I’ll be the one, R and G?
This seems similar to morph by zynaptiq but with automatic sample selection.
What’s rng
Looks like The Melda MMorph to me
P
O
R
T
A
L
L
reminds me of XO by XLN Audio
Big Krotos vibes
Ai linus tech tip 😂 as the hero image
just synplant 2 on steroids, might be great for video productions
THIS IS COOL.FOR NATURAL SOUNDING EFX THIS IS NICE.
This is a shit version if thr max device Coalescence by Dilllon Bastan.
It uses the same basic concept of concatenative synthesis but, as someone who has both, I’d say they fulfil different functions. Coalescence has features that Concatenator doesn’t and you can perform with it in different ways but Concatenator seems to map the corpus (the chosen folders/samples) a lot more accurately to the input audio. That’s an aspect I’ve really struggled with on Coalescence.
whattttt
nvm i'm disappointed lmfao
Udio your own shit. You guys love it 😂😂
It's a clone of the Max4Life Instrument called Coalescene from Dillon Bastan
You have no idea what you're talking about. Stop spamming misinformation under every reply. Basing your opinion off a video of some guy poorly reviewing something instead of doing research isn't very smart
stupid and cringe plugin, no one's using this for film sound fx omg lolol
Hmm a bitchrushing / pixlating glitch filter... ok ... eh maybe. Should not be more than about $49
What pixelating and bitcrushing? This is something completely different.
nobody who knows their stuff and takes music seriously would ever buy this junk
it's so half-baked. if it's anything like synplant, i'd think about it. since it will utilizes actual AI.
- recognize what kind of sound it's hearing (ex: a drum loop)
- analyze the loop into sections (highs and lows)
- collect and load sample libraries according to a theme you choose (ex: wooden perc)
- replace the samples being played like slate digital's trigger
you don't know what you're talking about this is a completely different type of tool from synplant they have like basically nothing in common
This plugin and synplant are completely different and do different things
@@zitronekoma30 i think you need to learn how to read. i didn't say they had things in common.
i said if it's anything like synplant that utilizes AI.
@@zoid3515 are you guys collectively tarded? in what world did i say they were in any way similar? LEARN TO READ.
@@Liza.Wharton that fact is that plenty of professional artists and sound designers use these tools already so learn your craft before talking big
you can "BUY" if it's available at certain sources 😉😉
this is bad