Syntrillium Backup Forums :: View topic

musicman · Posted - Mon Jun 18, 2001 6:50 pm

Hi, I remembered (vaguely) reading some posts a while back explaining how to extract one particular sound out of a file by pasting the exact same audio (minus that one sound) on top of the "audio with the sound" and in doing so would phase out everything except that one sound. Well, I tried it. I made sure that the copy was the exact same as the file in interest but lacking ONLY in that one particular sound I wanted to extract. However, what ended up happening was that the resultant audio phased/flanged pretty strongly but all the audio was still audible. Did I do something wrong? Please help. Thanks!

SteveG · Location: United Kingdom

If the audio phased/flanged, then there were varying millisecond timing differences between your two files. Even if you aligned the starts accurately, you couldn't fix this. Identical means identical! What was the source of your two files?

Steve

Edited by - SteveG on 06/19/2001 12:57:23 AM

RobertM · Posted - Tue Jun 19, 2001 5:52 am

Expanding on the "identical" thing...

If you are 'recording' into the soundcard, meaning that you are feeding an analogue signal into the card and using the soundcard's A/D converter, then you will NOT get an identical wav file. The file will "sound" the same, but there will be slight differences in each sample point, both in amplitude and in the time-based postion of each point. Even if you were able to get the first sample point of each wav file to line up exactly (as Steve mentioned), there is still no guarantee that all other points will fall in line correspondingly, and this is caused by slight speed varations caused during the analogue playback.

The only way you can hope to make this work perfectly is if the sound that you want to extract was added to the parent wav file in the digital domain, and you have a copy of both the parent wav file, plus the parent-plus-sound wav file, and neither file has been editied (NR, normalize, delete a section at the beginning of the file, etc) after the mix was made. If you have THOSE files, then it shouldn't be hard. If some editing was done, then you may be able to get it to work after a fashion, but you will be left with residual noises caused by the prior editing... it won't be perfect, but it may be good enough for your purposes.

What you want to do is certainly possible. The 'Younglove' script, much discussed in the forum, relies heavily on this invert-mix-paste technique, but that script guarantees that the various copies of the wav file are truly identical except for some very specific, and well controlled, differences.

Hope this info helps.

Syntrillium M.D. · Location: USA

Hi Music. I'm wondering if you're actually referring to "A-minus-B" listening or something along the lines of Cool Edit's Vocal Cut procedure. Not strictly for vocals, it reverses the phase of one channel of audio and sums it with the other channel, causing all center information to 'drop out'. If this is something you desire, this can be found under Transform>Amplitude>Channel Mixer>Vocal Cut.

---Syntrillium Support

musicman · Posted - Wed Jun 20, 2001 7:26 am

Thanks a lot guys. I now understand the issue of "identicalness" that goes into doing this. Actually, my source was from a 12" vinyl, so I guess I can forget about doing this for now. =) Again, thanks!