A few folk will know that some of us here had taken Chris Liepe’s Discover Your Voice singing course and I’ve kept myself on the mailing list as a few gems drop into my inbox now and then. And this is one.
I’ll let Chris explain how it works and how he uses it in the video below. But basically head over to https://vocalremover.org/ and upload the track you want to strip the vocals from and follow the instructions, simples. In fact despite Chris showing how he gets the edited file on his system, there is actually a SAVE option after conversion and you can just down load the file back to your PC or wherever.
Anyway here is a Greenday track I experimented with and assuming I get the share settings right, the first is the original and the second sans vox. Could make for interesting projects.
That’s awesome! I’m always curious on how this is done. Doesn’t seem too difficult based on his video.
Slightly off topic but how was this course? I’m considering taking a singing course and have 0 experience with singing and of course sing terribly lol. Does the course assume any knowledge and do you feel like it’s helped you develop your singing voice?
Thank you for sharing this video, I was not aware an online version of this thing exists. Cubase has this function built-in since version 11 (the software is called Spectralayers I think and it’s lite version is now built into Cubase), it’s quite a handy tool for transcribing I must say! The first time I saw this functionality was in Adobe Audition though!
How it works is actually very interesting. Instead of just using the waveform, you approach the track from a visual point of view a.k.a. see it as a complete spectrum of frequencies and then you can “photoshop” your track. The machine learning algorithms helps here because they are trained to listen for a specific instrument and isolate it. While we (humans) can do this mentally and focus only one specific part of the song, computer has the capability to isolate this part of the spectrum. I probably butchered the computer science behind it, so if anyone else chips in more accurate information, I’ll be grateful!
@alexisduprey - I’ll come back on the sing question later but yes it is good value for money but I’ll link the 3 taster lessons in a couple of hours when I am free.
Hopefully folks may find this useful. Certainly quicker than creating tracks in Reaper using midi input from GuitarPro but I do prefer to add different instruments and fx when creating BTs and tailor them to my requirements. But definitely a quick and easy way for most peeps.
Thanks for that Toby definitely will be using that. I remember good old days when you had an mp3 track and you were able to load it into your winamp player (anyone remembers it? ) and there was a plug in which was able to strip vocals out. However it wasn’t great
Note that if you just want to find a karaoke track with no vocals, YouTube channel CC Karaoke has hundreds of tracks created using the same techniques.
They mostly sound good. The only problem I saw was that the timing of the vocal text on the screen doesn’t match up very well with the music.
Worth a look if you don’t want to mess with the tech of removing vocals yourself.
(I think Chris Liepe mentioned this channel in his video)
Yes I noted that but wanted to prove the concept. Their tracks and lalal.ai to remove the guitars would work fine but you only get 10 minutes for free and I’ve not dug into the lalal pricing yet. I am looking to see if there is something or someone out there who does vocals and guitar removal for free or less than karaoke.uk where you can buy a custom BT and just use the parts you need.
I haven’t tried it yet but perhaps I will…
The thought is, if you have a VPN connection in theory you can do 10 minutes, then pick another server for the vpn connection, you get another 10 minutes… Rinse, repeat…
That could work? Of course you’d have to already have a vpn otherwise you might as well pay lalal.ai…
Sorry did not take it that way at all but maybe I should have made the reference on the post. I had a quick look and there are loads of tracks on CC K, so for a BT to play with and sing, you would just need to strip the guitar out.
An evening of Open Mic balancing and rehearsing has not allowed me to dig deeper. The lalal.ai audio output is good quality but I’ll have look at their pricing in the morning and search for alternative. There are some vids on YT but not had chance to look at them yet.
I think you need to sign in via email, Google or Facebook, I did using my Google acct so its not IP related and I do already have a VPN as I bounce back between the UK and France for certain “access”. I have 3 email accounts that would give me 30 minutes minus Church On Sunday. Maybe an extra 10 for FB ?
Oh I see! I thought I’d be clever… Little did I know!
This “access” (to free, foreign online tv channels) is particularly good for me to watch F1
But back to the topic, thanks for sharing the knowledge. I might try the websites out ovrr the weekend!
That looks very interesting but not cheap at £79 for the basic version. Vocalremover is free but I’ve not had a chance to dig much further for a similar product or site that remove GTR and gratis. When I get some time I get the shovel out !
Before I go digging, I looked at www.lalal.ai and the fee is ÂŁ30 for 300 minutes. There is also a Lite version for ÂŁ15 for 90 minutes conversion. So strip the vocal out of a track for free using vocalremover.org and then run the voice free track through lalal.ai to remove the GTRs. Sounds like a possible plan if no freebies are out there.
I have a question, perhaps the answer is obvious but…
When they say 300 minutes, I presume they mean song-time and not the time you spend logged in, right?
It’s the limit on how many minutes worth of audio/video are available for splitting. The length of every fully split file is deducted from the package minute limit. You can split as many files as you want as long as their total length doesn’t exceed the minute limit of your package. Minutes are deducted from the account by the following formula: Total file length x stem separation type(s) number.
) and there was a plug in which was able to strip vocals out. However it wasn’t great 
