-
Separating the vocals from an audio track
Hi guys,
I am using this one song for my video and there are parts in the song for which the vocals don’t match the theme of the video. I wanted to eliminate the vocals from these segments in the video where they don’t match while keeping the background music at the same time. I used this software called Audacity to separate out the vocals from the song and then on FCP, I set up the original audio file with the vocals on track 5 and the edited one without the vocals on track 6. For the parts where I would not want the vocals in the video, I would cut the original audio files using the blade tool while keeping the edited file without vocals unchanged. This method did the trick for me, however the music volume drastically lowers down at transitions from the original audio file to the edited ones at the instances where i made the cut and I have a feeling that some details are also lost in the process. I thought this might be because the edited audio file is mono in nature and hence I placed a exact copy of the same edited audio file on track 7 as well, just so there might be a better match in db as compared to the original audio file with the vocals. This didn’t seem to make any difference either.
Would anyone happen to know the correct way of doing this?
Any help would be really appreciated.
Thanks 🙂
-
5 Replies
-
This approach is never going to be perfect because the process of stripping out the human voice from your master audio track unavoidably also steals away some frequencies shared by the music. Adjusting volume alone isn’t going to be enough to truly make it “match”; to the human ear – you also have to somehow restore some of the missing frequencies in the non-vocal track, thru tools like expansion or compression or multi-band EQ, etc.
This is the kind of work a dedicated sound designer does, and they earn their money.
You might get better results by using more powerful tools than Audacity, going to ProTools or Apple Soundtrack pro, or Izotope, might help some. I don’t suppose you can go to the source of the original track, and ask them for an instrumental version of it?
Doubling a mono track the wrong way can make things worse by creating phase-cancellation errors, so be careful there.
-
Hi Mark,
Thanks for the information.
I doubt I can approach the source for the instrumental track because they are mainstream and since I don’t intend to use the video for any commercial purposes, I decided to use their track.
I understand that a perfect match is something that i probably cannot achieve, however, do you know of any way within FCP that I could make the frequencies/db/volume of the edited audio file and the original audio file match?
Regards,
Tanmay Chowdhary -
Try the things I listed in my previous post: multiband Equalizer filter, audio compression, etc. They are in the audio filters section of FCP7.
-
Vocal stripping in Audacity etc involves flipping phase in one channel to cancel the centre which is usually the vocal. However it can also be a lot of other instruments. When you edit between you are creating phase differences on the cuts which will mean you cannot do a match edit.
This is becasue there will be a difference between stero and mono versions of the audio. So unless you make it mono only you don’t stand a chance. Even then a match edit will sound weird as more than the vocal will be dropping out.
-
Hi Michael,
Thanks for the input. I ended up simply eliminating the vocals part and it worked for the better.
Regards,
Tanmay
Reply to this Discussion! Login or Sign Up