Hi!
When listening to some great tracks, I always had the dream to be able to extract drums from a mix…
As a developer and DSP guy, I decided to spend a few years working on this topic, and now we (YellowNoiseAudio) release the VST DrumExtract.
Feedback welcome!
Dynamic EQ is probably your best bet.
or a Dynamic compressor with sidechained from EQ.
You will suffer collateral damage tho, and probably still not get all removed.
luckily [most] traditional kit sounds exist in relatively concise frequency bands, when compared to other sounds. kick/snare are probably some of the easiest to filter.
You could have at least read what he actually said
ya, feel pretty stupid. i just got done dealing with the challenge his tool aims to solve, so it was saturating my brain.
definitely curious to evaluate the results of the product.
hopefully what i typed may serve somebody well in the future in some way, regardless of context. still valid information… just ADD-ridden choice of location to post it…
apologies to OP. at least my post accredits the usefulness of your product by stating that you will likely not achieve success through traditional mastering methodology!
impressive if this can do this easily and consistently
How does it do on tunes where the drums are way less drumkit-sounding?
that’s pretty sick
the venetian snares clip is quite convincing. there’s just always those artifacts which i guess are not possible to get rid of yet. sounds like the “edges” of the sound have been mp3’d at a low bitrate, if you feel me.
Any plans on having a time-based trial version of your product? I’d love to evaluate and review (and purchase if the need ever arose again), being that i was in between a rock and hard place recently and would love to know if this could be my ace-in-the-hole should i face the same challenge in the future.
my recent necessity for drum extraction was probably a bit divergent from your core market, being that i was trying to brighten some drums while mastering a track for somebody, and just couldn’t get at some of it without hyping the wrong elements. my manual efforts did ultimately produce the desired effect, in allowing me to layer extracted drums in additional channels on top of the stereo mix, but it was far from elegant or easy, and still could’ve been better with the right tools.
if nothing else, wanted to share my case study with you, since people mastering tracks frequently need to abstract elements and re-composite them again under varied condition. you touch on this under “mixing: raise or lower volume of drums in a mix”, but somebody that does a lot of mastering of mixdowns may not even see that as relevant if they’re not paying close attention, since you’re really already post-mixing. it could be a good additional way to market to including mastering as an item.
i have no idea, i’m not a marketing guy. but your tool could be huge in mastering, and mastering is not mentioned anywhere in the product description on your page.
hope my feeble input helps in some way!
I don’t exactly support anything created to rip out stuff from other people’s tracks, but I’m kinda interested in what Harkat said. I suppose the algorithm detects peaks and gates out the sound around them. How does it work when the track is mastered to -7dB/RMS (which is nothing uncommon these days) and the peak volume is steady 0dB?
i’m with you on both sides of your statement.
that’s why i have interest in its potential viability as a mastering tool. (not for my own tracks, obviously any need to deal with the drums can be handled easily during the mixing process) but for somebody that masters tracks for a living, if it could solve the challenge you mentioned. today’s tracks saturate the entire frequency range and often hover just under digital clipping through multiple instruments coming and going. and most have been run through an exciter, or a saturator with soft clipping to simulate the additional warmth of analog clipping, which throws out all kinds of harmonics. trying to get the drums away from that to correct problems in the mix is becoming increasingly challenging to people that work in mastering studios. (I’m not one of them, but I’ve taken on the task a few times for some of my m8s, and feel their pain)
I for one would like to have a trial of this before I buy it.
Congrats, previews sound pretty clean, but would also like to try before buying.
looks very interesting and i’d also love to see a demo version of this
this is very cool it sounds pretty good too.
ive messed about with removing the drum tranients from tunes by manual slicing and changing the attack to soften the impact the drums have on the sample. this is long and boring tho and that doesnt give u the drums but the opposite.
U are deffo on to something here well done man
Thank you everybody for your feedback!
Few time to answer in detail now, I’ll do it later, we are busy working on Mac version + update (free for all current customers), but in a few words:
@Omnibros_Productions:
Dynamic EQ is probably your best bet. […] apologies to OP. at least my post accredits […]
No worries, and yes what you typed is also useful! Always good to have two different tools available rather than one!
if nothing else, wanted to share my case study with you, since people
mastering tracks frequently need to abstract elements and re-composite
them again under varied condition. you touch on this under “mixing:
raise or lower volume of drums in a mix” […]
your tool could be huge in mastering, and mastering is not mentioned anywhere in the product description on your page.
Totally true, as you mentioned, it can be very useful in mastering. We should add a mention about this.
Currently, you can extract drums, and add this extra layer of drum on top of stereo mix (and balance it with a fader). Even if this “extracted drum” is not perfect, as you blend it with the original mix, very few artifacts will be there in this context.
In the forthcoming update, it will be even easier, because we’ll provide a full range of mix with a fader between 100% percussive and 100% harmonic. Ok you can already do it with the current version (by adding an extra track that you balance with the fader), but with the update, it will become even more handy.[quote=“Ag_U, post:9, topic:8702, full:true”]
I don’t exactly support anything created to rip out stuff from other people’s tracks, but I’m kinda interested in what Harkat said.
[/quote]
Well, it’s a tool that you can use for good things (mastering, drum learning purpose, or even sampling if you ask the original author and “clear” the sample before release of your song, like everyone does now) or bad things, so I wouldn’t say this tool was created to rip out stuff… 
I suppose the algorithm detects peaks and gates out the sound around
them. How does it work when the track is mastered to -7dB/RMS (which is
nothing uncommon these days) and the peak volume is steady 0dB?
As a general rule, don’t expect this tool to work on all songs. It’s like a surgeon tool that you use for specific purpose, e.g. “I want to isolate the drums from this song from 2’10” to 2’15" " etc.
But the song that I used as example on the website (Tame Impala song) is a 2015 song, mastered with a very high RMS if I remember well, and it worked quite well.
@everybody:
trial version
We haven’t considered up to now, because we really don’t like iLok / dongle solutions, this is really annoying for the user. A solution like “2 seconds of white noise” every 10 second wouldn’t be useful as well, because then it’s like making this plugin free and stopping our R&D for future products! Lastly, I have never seen a protection like “30 days trial version” that really works. Change the computer date or uninstall/reinstall it and you’ll have it forever!
DrumExtract could be a solution then (I’m not claiming it would turn your job from hours into a few seconds, but for sure it can help).
could you not cripple a certain function, or make it only work with like 3 seconds worth of audio or something like that for a demo version?
I understand your points but it’s hard to determine what this can do for tracks that are more freq heavy and obviously mastered up to like 5rms or whatever (I believe rms is what it is)
Like, if I were to use it on a VR/Skrill/Getter/etc track etc, how reliable and functional would it be and how it will operate under those conditions.
Again, the tracks you showcased, to me,don’t really show the true power of this plugin, so I’m not really hooked on the idea of buying a 49€ plugin if there’s not enough reliable examples to make more of an idea about how good the plugin is. That’s just my 2 cents but, hopefully you’ll be able to work something out.
This exactly.
Because no matter how you look at it, the successfuless of this plugin would be in people wanting rip off drums from major artists. And seriously, those examples are really dynamic compared to like 90% of the music people want to rip.
In general, don’t expect it to work on any song. I guess it won’t give good results at all on Skrillex songs: the mix are generally very busy, and there are lots of other things than drum itself that are percussive.
Here an example on a XX song
http://gget.it/73ulosqn/xxintro.mp3
http://gget.it/66xlrppp/xxintro_drumextract.mp3
Here an example (removing drums) on Santana
But once again, this tool is not magic, it cannot separate egg and flour from a baked cake all the time
As a rule of thumb, as detailed here, DrumExtract works great on songs where 1) the drums are the only percussive instrument, i.e. no bouncy 1/16 notes bass or 1/16 notes melody etc. 2) no vocal.
Our statement is that it does better than other traditional tools (eq, compression, etc.) about isolating percussive content, and offers state-of-the-art separation technique, so it’s a big step further in the available separation tools (nearly no similar tool existed before), but it’s not a magic software: you will find examples for which it doesn’t work.
Well still, it doesnt really need to be magic.
I imagine if this works like it should you could extract some pretty dope one shots and stuff if you used it creatively.
Some weird percs you hear in jazz or whatever