Calling All Mix Gurus:Extracting or Isolating Vocals In A Stereo Wav File

Guys I have a mix down stereo file I recorded that I wanted to remix. I have both the final mix down stereo file and the mix down without the vocals. I didn’t do the vocals but they have too much Reverb applied. Is there a way to extract or isolate the main vocals in the stereo file in Auria Pro? What do you recommend or is really not possible? Thanks guys!

Comments

  • 25 Comments sorted by Date Votes
  • Definitely not possible on iOS and almost impossible on computer. There is a new plugin for computer which is claimed to do an amazing job but still it will introduce noticeable artifacts when pushed

  • Thanks so much! This is uncharted territory for me. I found this YouTube video:

    https://youtu.be/ NilfCElGJ2c

    Wonder if the Pro-q2 app might do it.

    I wanna isolate the vocals if possible. I know it’s gonna be hard but I wonder if therI s a way to hide for lack of better word, the Bass and hi frequency and leave only the center where the vocals are. I know the snare is centered too but as long as I can get as close as possible. But don’t yet know how.

  • Anytune Pro + works a little bit depending on the material, but you’re not likely to get results you’re happy with. It’s more useful for something like partially removing vocals to be able to sing over them, or maybe isolating a guitar solo or something to learn it.

    You basically only have two variables to work with, Panning, and Frequency. With AnyTune you can set the range of frequencies to remove or isolate within a panning range. So, if you have a nice pominent, centered, piece, you can try carving out the major frequences of that piece within the position in the stereo field it occupies. Carefully tweaked it can sometimes sound kind of OK. But usually you need to take out so much material that it interferes with overlapping material and sounds pretty crappy.

    Another tool that might enable something along the same lines is Directional EQ. ZMors in mid-side eq mode might be useful as well. I really doubt Pro-q2 will be any kind of magic bullet because short of super-sophisticated frequency pattern recognition there just isn’t any way to neatly accomplish that kind of surgery. It’s a bit like trying to separate out the flour and the milk from pancake batter after it’s mixed.

    Probably not destined for success, but a good learning opportunity for sure.

  • I often do remixes where I have to make my own acapellas. Unless there's a lot panning effects on the mix, it should be as easy as matching both versions exactly in a DAW and then reversing the phase of the instrumental track. The tracks should cancel out leaving the vocals naked. You can then use a transient shaper to reduce the amount of reverb. I've done this a billion times.

  • @coolout said:
    I often do remixes where I have to make my own acapellas. Unless there's a lot panning effects on the mix, it should be as easy as matching both versions exactly in a DAW and then reversing the phase of the instrumental track. The tracks should cancel out leaving the vocals naked. You can then use a transient shaper to reduce the amount of reverb. I've done this a billion times.

    Oh damn, that’s brilliant!

  • @coolout said:
    ... it should be as easy as matching both versions exactly in a DAW and then reversing the phase of the instrumental track.

    >

    This is what I love about the forum. Someone declares a thing to be impossible, and someone else reveals a straightforward way to do the ‘impossible.’ :)

  • But it is still impossible to suck a piano up your nose.

  • The phase trick only works if the instrumental is exactly the vocal track minus the vocal. If you did any processing on the vocal track, even levels, it won’t be the same anymore.

    If you have auria you can load both tracks, line them up to the sample, and invert the phase of one chanel.

    You can’t isolate vocals with EQ. Search for DIY acapellas, this technique has years, there’s even a forum acapellas4you you might wanna check.

    Also try M/S processing if your vocal is dead center, but stereo reverb defeats the purpose.

  • Wow thank you all for such great advice! I literally almost fell of my chair laughing at the pull up a Piano up your nose thing! :D :D :D

    Oh my gosh that was soooo funny :D

    Ok now, I like the idea of reversing the phase. How do I do that in Auria Pro?

  • @MusicMan4Christ said:
    Wow thank you all for such great advice! I literally almost fell of my chair laughing at the pull up a Piano up your nose thing! :D :D :D

    Oh my gosh that was soooo funny :D

    Ok now, I like the idea of reversing the phase. How do I do that in Auria Pro?

  • Wow that’s so cool! Thanks so much!
    Will try out today!

    Got me looking into the phase thing and found this vid;

    I guess same technique applies with Auria. Gosh so happy and thankful for the suggestions guys I’m learning so much!

  • If you want to use the inverted phase of the instrumental track, it will work well if the tracks are identical, minus vocals on one. Also when you mix the inverted with the regular, not inverted, the tracks need to be perfectly in phase, just opposite phases. If one track is slightly out of phase, you will hear some weird (cool to me :) ) artifacts in the final result.

    If the mixes are different, this won’t work much better than flipping one channel of the stereo file the isolate the mono elements of the mix. You could also try using narrow band EQ to see if it helps. I have tried all 3 ways with decent results, but the method with identical tracks (one w/ vocals, one without) will work perfectly if the tracks are in fact identical, and you line up their phases perfectly when mixing the two together.

  • Thanks! Yes they’re identical. When I recorded this tracks I used an old Alesia Wedge reverb for the vocals. I recorded the whole thing on my old Tascam 688 midistudio and mixed down to an external cd recorder. Wow those days are so gone.

    Well, I should have saved a copy with dry vocals but I guess I didn’t know any better. But I did do a mix down with out main vocals for the singer to use as a backing track. So yes they’re identical.

    Ok so in Auria do I press the phase invert on both tracks?

    @CracklePot said:
    If you want to use the inverted phase of the instrumental track, it will work well if the tracks are identical, minus vocals on one. Also when you mix the inverted with the regular, not inverted, the tracks need to be perfectly in phase, just opposite phases. If one track is slightly out of phase, you will hear some weird (cool to me :) ) artifacts in the final result.

    If the mixes are different, this won’t work much better than flipping one channel of the stereo file the isolate the mono elements of the mix. You could also try using narrow band EQ to see if it helps. I have tried all 3 ways with decent results, but the method with identical tracks (one w/ vocals, one without) will work perfectly if the tracks are in fact identical, and you line up their phases perfectly when mixing the two together.

  • No, just on the instrumental track. Because the vocals have reverb on them, I'm guessing you're not going to be too impressed with the results though. In general the more effects are on the part you're trying to isolate with phase cancellation, the less likely it is to work cleanly. You tend to end up with just the midrange of the vocals, or just the low end. Let us know how it works out though.

  • Are you going to kill the reverb with FAC Transient? I would like to hear how it all comes out. Good luck, and post more questions if you get hung up. :)

  • Thanks guys will do. Daunting task but learning a lot.

  • thanks!

    @Varsham said:

    As with center-channel (vocal) removing, the better the quality of the original tracks, the better the effect works. If you’re using MP3s, do whatever you can to avoid using tracks with a lot of clipping. This will ruin the effect over those parts. You can highlight where clipping occurs on your tracks in Audacity by going to View > Show Clipping. ShowBox Lucky Patcher Kodi

  • edited March 14 Vote Up0

    @MusicMan4Christ said:
    Thanks! Yes they’re identical. When I recorded this tracks I used an old Alesia Wedge reverb for the vocals. I recorded the whole thing on my old Tascam 688 midistudio and mixed down to an external cd recorder. Wow those days are so gone.

    sorry to spoil the party, but 2 playbacks with an analog recorder never (!) end up 'identical'.
    Expect a lot of flanging, as the tape flutters constantly - even if the 688 was in prime condition, it's natural variation is huge for this kind of task.
    A displacement of 4 samples is equivalent of 1/10,000 of a second and would be clearly noticable.
    On the other hand it would be interesting how it comes out as tape is very 'random'.

    Transient shaping (shorter decays) can in fact reduce perceived ambience significantly.
    (as already mentioned it also was the main task of the original Transient Designer in my case)

    But meanwhile true reverb removal algorithms exist, which do a better job.
    I use Zynaptiq UnVeil/Unfilter, there's also Izotope's De-Reverb, Acon Deverberate and probably some more.
    According to my experience it's not some click-and-forget process, but needs proper tweaking for convincing results.

  • @Telefunky I just looked up the Tascam 688. 8-track cassette! I think there are going to be problems as well.

  • Yeah just realized that during my lunch time :(

    They are almost I mean once zoomed in almost identical but as the song progresses, I hear double effect.

    I’m almost thinking it’s gonna be best to re record vocals which is what I wish I didn’t have to ask the singer to do plus time is tight.

    @Telefunky said:

    @MusicMan4Christ said:
    Thanks! Yes they’re identical. When I recorded this tracks I used an old Alesia Wedge reverb for the vocals. I recorded the whole thing on my old Tascam 688 midistudio and mixed down to an external cd recorder. Wow those days are so gone.

    sorry to spoil the party, but 2 playbacks with an analog recorder never (!) end up 'identical'.
    Expect a lot of flanging, as the tape flutters constantly - even if the 688 was in prime condition, it's natural variation is huge for this kind of task.
    A displacement of 4 samples is equivalent of 1/10,000 of a second and would be clearly noticable.
    On the other hand it would be interesting how it comes out as tape is very 'random'.

    Transient shaping (shorter decays) can in fact reduce perceived ambience significantly.
    (as already mentioned it also was the main task of the original Transient Designer in my case)

    But meanwhile true reverb removal algorithms exist, which do a better job.
    I use Zynaptiq UnVeil/Unfilter, there's also Izotope's De-Reverb, Acon Deverberate and probably some more.
    According to my experience it's not some click-and-forget process, but needs proper tweaking for convincing results.

  • @coolout said:
    I often do remixes where I have to make my own acapellas. Unless there's a lot panning effects on the mix, it should be as easy as matching both versions exactly in a DAW and then reversing the phase of the instrumental track. The tracks should cancel out leaving the vocals naked. You can then use a transient shaper to reduce the amount of reverb. I've done this a billion times.

    So, Auria can do this.

    Any other iOS DAWs too?

  • @tja said:

    @coolout said:
    I often do remixes where I have to make my own acapellas. Unless there's a lot panning effects on the mix, it should be as easy as matching both versions exactly in a DAW and then reversing the phase of the instrumental track. The tracks should cancel out leaving the vocals naked. You can then use a transient shaper to reduce the amount of reverb. I've done this a billion times.

    So, Auria can do this.

    Any other iOS DAWs too?

    I am pretty sure you could do this in AUM with the built in Stereo Processing effects and File Players. Have not tried it, but it has all the necessary tools, and since it is AUM, I am confident it will work.

  • @CracklePot said:

    @tja said:

    @coolout said:
    I often do remixes where I have to make my own acapellas. Unless there's a lot panning effects on the mix, it should be as easy as matching both versions exactly in a DAW and then reversing the phase of the instrumental track. The tracks should cancel out leaving the vocals naked. You can then use a transient shaper to reduce the amount of reverb. I've done this a billion times.

    So, Auria can do this.

    Any other iOS DAWs too?

    I am pretty sure you could do this in AUM with the built in Stereo Processing effects and File Players. Have not tried it, but it has all the necessary tools, and since it is AUM, I am confident it will work.

    But with AUM you’ll have pretty hard times lining up the tracks

  • @mschenkel.it said:

    @CracklePot said:

    @tja said:

    @coolout said:
    I often do remixes where I have to make my own acapellas. Unless there's a lot panning effects on the mix, it should be as easy as matching both versions exactly in a DAW and then reversing the phase of the instrumental track. The tracks should cancel out leaving the vocals naked. You can then use a transient shaper to reduce the amount of reverb. I've done this a billion times.

    So, Auria can do this.

    Any other iOS DAWs too?

    I am pretty sure you could do this in AUM with the built in Stereo Processing effects and File Players. Have not tried it, but it has all the necessary tools, and since it is AUM, I am confident it will work.

    But with AUM you’ll have pretty hard times lining up the tracks

    Yes it would be practically impossible. I was thinking of the case where there is just one digital file used, do the process on one channel, and mix together. The usual mono extraction method. The more daunting task involving the analog sources would be a headache.

Sign In or Register to comment.