Hi All,
I was wondering how to successfully remove beat and instrument sounds from a sample, leaving as clear as possible just the vocals.
I read a couple of other forums, it mentioned one needs an EQ with lots of bands.
I experimented with the sample used in Prodigy's Break and enter, (Baby D Casanova.) using various Caustics EQ / filtering, but the end result was either unwanted sounds in the sample, or an excessivly tinny vocal. (Admittedly I didn't spend much time on it).
So... Is caustic a suitable tool for this kind of sound manipulation, or should I look at a seperate app.
Any suggestions for such an app under android/windows?
Cheers
i dont think caustic has the tools to extract vocals from a sample, in this currant incarnation...you can try using a couple parametric eqs to remove some frequencies...then resample by exporting the processed sound to wave...then process it again through more parametric eqs...but i doubt that will be adequate...youll probably need a standalone aplication that cleans vocal tracks...its my understanding its much harder to remove music from a vocal than it is to remove vocals from music...i doubt youll ever be able to get a clean vocal this way...no matter how good the program is.
You are right about that dan. Vocals are often the only sound panned to dead center, so programs like Audacity (which I highly recommend) can simply remove all frequencies that are equally in both channel. This sometimes leaves the remaining track sounding a little thin, but audacity can take care of that too (the remainder of the waveform is still as perfectly staged and mastered a track, so boosting and amplifying it yields surprisingly clear results.
Now, in audacity you can do a lot more than that if you're a coding prodigy fluent in Nyquist, but I couldn't begin to tell you how it's done.
EQ-ing out instruments to leave only vocals will only frustrate you, unless the clip is extremely sparse and well defined. I used to attempt such things, only to waste 20 hours of my life with nothing to show for it -- not like now! (: -- but I'd love to hear about your progress.
All that being said, their may be some innovation I've never heard of, as I am getting old and out of the cool loop...
¿que onda, pachuco?
Mmmmm what you are dreaming about is generally not possible, even if you invert phase on one channel and sum mono..... Some eq tricks can help very little, but extracting one bit of audio from a mix..... You cant unmix whats already been done, period.
I know a Nyquist programmer who claims he can take the drums out of most 80s songs with audacity. I've heard the end result - it's pretty seamless. Something about the gated reverb of the post collins/gabriel drums makes it easier to isolate the specific bits that are drums (and drum reverb) and remove them from the waveform. Now they don't sound "good" at all, I merely said seamless, lol... Once you start erasing information in the digital waveform what do you replace it with? Empty space?
My unverified and admittedly amateurish hypothesis is that within two decades the programs will exist for anyone to "unmix" the average digitally recorded pop tune.
¿que onda, pachuco?
Hmm, interesting.
Begs the question. How do bands like the prodigy do it?
And as for the trick of removing the stereo field's center, the only time I used it was to make an old happy mondays tune into a karaoke track for a friend, and we were all surprised how well it came out! Did I get lucky? Yes. If the vocals were panned (even a little bit) it wouldn't have worked. Likewise, if anything else was panned to center it would have been removed. If I needed anything besides lead vocals removed I could not have done it. Also, I understand that it leaves you with swiss cheese for a waveform, but it didn't hurt that particular song in any noticeable way...I'll e-mail you, or post a clip, if I can dig up that hard drive.
¿que onda, pachuco?
How do they do what, specifically?
¿que onda, pachuco?
Use clean vocals for samples.
If you take the example I mentioned , Break and enter, the sample sounds clean as a whistle in the track.
The original sample by Baby D, there is a beat, and lots of backround instruments playing during the vocal section.
I have audacity, , and will have a tinker with that, see what results I get.
Certain parts of the prodigy track has very little beat, during the sample, so there has been a very good job of cleaning it up.
Depends what mix of Casanova they used. I've sampled from 12" inch mixes and other remixes found on singles as they often have extra bits like an acapella vocal section to fill it out.
A quick google search suggests the "Raising Hell Mix"
A quick phone call from the record company of one well known act to another usually secures access to multi track material.
http://m.soundcloud.com/metatronic555
Ha, I'm sure they are not going to arbitrarily release multi-track material to just anyone.... for a variety of reasons.... Heck I wouldn't if I had material like that, that I worked hard to put together..... been there done that....
Inverting phase on one channel to "remove" dead center content will yield a bunch of noise, and not much useable material. But isolating just mono-centered material is near impossible in alot of situations, unless you get lucky to get solo hits of instruments or drums.... which rose in popularity in the 80's for old-school hip hop, and in the 90's with jungle and drumnbass etc... etc....
In the end, when trying to extract content from rendered mixes, sometimes you get lucky sometimes you don't.
But if the Prodigy asked if they could sample you, would you say no?
http://m.soundcloud.com/metatronic555
Great point about other remixes, that's something I had not considered.
. Using keyword acapella brings up alot of new material.
Speaking of phase inversion, about a year ago I figured I'd mess around with phase inversion to see what I could come up with. I still had a copy of the 5.1 split audio tracks from NIN - The Downward Spiral. I wanted to see if I could create a solid instrumental version of "Ruiner", so I loaded all 6 of the audio tracks in FL10 and spent a lot of time in "Trial and Error Land". I didn't succeed. I got a lot closer than I expected... however, THAT little bit of success was cancelled out by the renmants of the vocals leftover... The first chorus smacked me in the face with harsh, robotic-sounding vocals......
Here is that very result... https://www.youtube.com/watch?v=h0J-Vy__sok
Also, in previous sessions of manipulating these same 6 audio files, I came up with a couple other decent results, but no phase inversion was used here.
https://www.youtube.com/watch?v=9JcPITP3FXo
...and...
https://www.youtube.com/watch?v=q9-I9bEtbUM
VIOLENT NEUROSIS (YouTube) ...https://www.youtube.com/user/ViolentNeurosis
better technology and separate tracks.
think about it. dozens of equalizer frequencies. the issue is removing the ones that are close enough that you cant remove them without removing the voice. if the voice sample ( at the smallest scales) equals the frquency of say a snare then you dont need to remove the snare because it wouldnt make a difference. 10000 hz is 10000 hz.
but i really have no knowledge base on this. so. im really just going on logic. which doesnt always apply.
but. really. it makes sense. i think.
Taking material from 5.1 or 7.1 makes it a bit easier, but I think the OP was about extracting content from a simple stereo mix.
Yeah, my comment was slightly unrelated, just you mentioning phase inversion got me started on a tangent. Sorry, I'm bad about that.
VIOLENT NEUROSIS (YouTube) ...https://www.youtube.com/user/ViolentNeurosis
Izotope iris.
Www.soundcloud.com/keeperz
What is the used for? How effective is it? Better than Elevayta plug-ins like ExtraBoy?
VIOLENT NEUROSIS (YouTube) ...https://www.youtube.com/user/ViolentNeurosis
Hey, no worries..... that's the nice little thing about music and video dvds encoded in 5.1 and 7.1, if you can isolate those channels, you can get some bits that can be useful for sampling purposes...... depends on how crafty the engineers are in their mixing abilities.... (I briefly dabbled in 5.1 with Nuendo some years back -- it is a challenge to work with no doubt)
Yeah, unfortunately the optical drive on my old laptop decided to croak and my new laptop doesn't have one, so I am limited to split audio tracks, or *true* 5.1 remasters... Found a couple DVD Audio remasters several months back, one for NIN and one for Korn... Loaded them into DVD Audio Extractor and found that all 6 channels pretty much contained the same audio data, just at different levels. In other words, basically just a much wider sound field. For my purposes, they were useless. They were custom remasters, but still, perhaps I was going about the process incorrectly. Hopes dashed. Dreams crushed.
VIOLENT NEUROSIS (YouTube) ...https://www.youtube.com/user/ViolentNeurosis
Spam alert
Not to do with vocals but with noise based things on tracks, risers etc.
Where the track is maybe just a beat and a riser, like the beginning, you can apply well controlled compressor distortion to just the areas where the beat is playing, this effectively changes it to noise to blend with the rest of the riser sample, and then you can just highpass it... So get creative, I don't think vocals are possible in caustic right now, but I often distort the hell out of things to make them sound cleaner :D
I make creative EDM utilising Caustic and many other Daws. See for yourself on my SoundCloud, you may like it :) https://soundcloud.com/xandyaudio thanks!
lol and now explain to me why a spambot chooses a dead thread from January 2014 to post its advert xD
Feeling collaborative or just need help? You can always message me: einsyui@gmail.com
I'm also always happy about messages on soundcloud!" --αlpha Yui
Crap, here we go again:
smithanderson78
I have audacity and don't have a fukking clue how to use it!!! (Apart from record n export to mp3) wud love to be able to dive into it. Recorded a vinyl with audacity n it sounded shite. Vocals screechy, bad sound quality. Tried playing with a bit of eq but cudnt fathom it out. Shame
AngryBulldog
I'm still fairly new to caustic but when it comes to extracting vocal samples and unwanted frequencies I found that the best way to do it is to try to cut out all of the frequencies that you want save that file and then load the file again and invert the WAV file so that way you cancel out all of those good frequencies in your left with only the bad that also works with pulling vocals out which is what I'm tinkering with right now myself since I don't have a good vocal editing program or wave editing program on an Android device what I've been doing is going through one of the karaoke making apps pulling out the song I want through that since hit extract vocals and then taking that inverting the wave and feeding it i and then taking that inverting the wave and feeding it into a way that it in progra and then taking that inverting the wave and feeding it into a way that hitting program that then cuts out all of the background instrumentals and you're left with just a few random wild frequencies that you have to cut out yourself but that takes a lot of the guesswork out of it and makes it a less lengthy process. That's not necessarily to say it's the best way to do it, melodyne is really good for since that's one of the things that that program the specificall since that's one of the things that that program was specifically designed for but when it comes right down to it really there's no good way to extract samples from a song unless you happen to have the root stems. Acapellas4u is one place I've gone to get a lot of really good stems for remixes also you've got to remember that if you can't get the vocal stem from a track but you can get all of the other stems from it you'll still be able to invert the wave and remove all of those tracks giving you the leftover bits like the vocals ECT. Like I said I'm still new to caustic and getting a feel for it so I'm not completely sure what all the program itself can do but as I go along I'm sure I'll figure it out and if I do then maybe I can post something else up later but if anybody else figures a good way out to rip vocals out of a track by all means hit me up LOL
Jon Holistik