WAV export amplified and distorted - sometimes

[kozikowski]

Let me rewrite that a bit.

<<<@Koz I think what Tony has done is to record one Stereo Performance with the "Right Channel control sweep from his audio system" + "microphone (no signal)>>>

Followed by...

<<<recorded a second Stereo Performance with "Left Channel control sweep from his audio system" + "microphone (no signal) which is positioned directly underneath the first performance>>>

And in that case, yes, any difference between the top and bottom captures would only appear in the output summation.

Since I really like simplicity, I would have exported the first performance, cleared the decks and then captured and exported the second one. You'll be doing something like that anyway because it's crystal clear the top and bottom tests are not coincident, hence the comb effect.

I'd still give much chocolate to know what the requirements of the room analyzer equipment are. I can see the storm clouds of the poster trying to export four channels in one sound file.

Koz

[tony22]

@Koz I think what Tony has done is to record one track with the "Right Channel control sweep from his audio system" + "microphone (no signal), and in the same Audacity project recorded a second track with "Left Channel control sweep from his audio system" + "microphone (no signal).

@Tony - is that correct?

You are correct. The test sweeps were already in Left Channel Only / Right Channel Only form. In point of fact I have a "real" set of sweeps in which the Right Channel in Audacity does have the mic waveform included. I just ran this variant to see if eliminating the mic feed would have any effect on things. Since it did not and they showed essentially the same effect, I used these traces as the example.

@Koz When he then Exported from Audacity, the two tracks were mixed together (this is what is supposed to happen when you Export from Audacity) and created the comb effect seen in the third screen shot.

@Tony, what you should have done is to select one of the tracks and then use "Export Selected" rather than "Export".

edit: If this is the case, is there a way for me to take the clean right channel WAV (which I exported originally before capturing the left channel), the combed left channel WAV, and "de-combify" that waveform using the right channel waveform somehow?

Yes you could - but if you use "Export Selected" to export one track at a time, then you will not need to.

I'm sure you've hit it, Steve. But for the sake of argument, what would I do in Audacity to revert the signal back into itself? At this point I'd like to have that little tool handy.

@Koz, I did it this way (again keep in mind I have real Audacity captures with the mic pickup in the Right Channel) because DRC needs separate Right and Left captures - each one made up of mic feed + control signal - in order to do its thing.

[tony22]

Since I really like simplicity, I would have exported the first performance, cleared the decks and then captured and exported the second one. You'll be doing something like that anyway because it's crystal clear the top and bottom tests are not coincident, hence the comb effect.

That's actually what I would normally do, Koz, so we're not so far off. In my haste to re-do sweeps I ran the second capture inside the same Project as the first, and clearly did not understand enough to do an Export Selection. Good lesson to keep things simple, as you say. So the trick is to unmix the first stereo track from messed up second track. Steve suggested this is possible. I'd like to see if I can do that, even though I will probably recapture the tracks again for real at some point.

[steve]

for the sake of argument, what would I do in Audacity to revert the signal back into itself? At this point I'd like to have that little tool handy.

This requires that you still have the original Audacity project, or that you have Exported at least one of the individual tracks (and that is why in practice it is pointless, because if you have the original project, then you can just use "Export Selection", but for the sake of education )

Let's call the original 2 recorded tracks "A" and "B".
The track that you Exported was a mix of "A" and "B", so in effect it was "A+B". We will call this Exported track "C".
If you use the "Invert" effect on one of the tracks, it turns the waveform upside down (positive "peaks" become negative "troughs", and negative "troughs" become positive "peaks". The value of each sample becomes the numerical inverse of the original value).

Thus, if we invert "A" and mix it with "C" we will get "C + (-A)", which is "(A+B) + (-A)". Thus the "A" component of "C" gets "cancelled out" to leave "B".

Similarly, inverting track "B" and mixing it with track "C" will result in audio that is identical to the original track "A".

This does of course rely on the positions of the tracks along the time line to be in exactly the same position as when you made the export, so that corresponding samples still correspond.

[tony22]

A-ha! I think this will work. I have A and C (which is A+B), all in their original timelines. So if I open a new Project, import A and C into it, invert A and then Export the whole thing, I should get B back. Sound right?

[steve]

As long as they are in their original time positions, then yes it should work.

[kozikowski]

You can't just run the sweep again? We've been how many days fixing this? Each patch step invites mistakes. Of course, I never make mistakes like this, but not everybody is so lucky.

You understand we're going to want to see at least one live total performance, right? With microphone track visible.

If you have A and C, then at one time you must have had A and B. Will Project UNDO help any here? I'm perfectly clear that Project tools fail at the worst possible time, which, I'm guessin' would be now.

Koz

[tony22]

Hmm, didn't seem to work. I am attaching two images. The first contains (for real) the two stereo tracks that started this whole thing. The upper track is the "clean" capture (A), and the lower track is what happened when I did the original Export while A nd B (now lost for good) were up in the Project space. Let's call the lower track in this image (C). The lower image is what I got after I inverted the first track in the upper image and then did an Export for those two. If it were to have worked the lower image should look very close to the upper track in the first image. The timelines of A and C have not been touched since they were originally captured / created.

Tracks A and C

combined.jpg (228.67 KiB) Viewed 495 times

Export after inverting A

combined1.jpg (155.12 KiB) Viewed 495 times

[tony22]

You can't just run the sweep again? We've been how many days fixing this? Each patch step invites mistakes. Of course, I never make mistakes like this, but not everybody is so lucky.

I can do it again, Koz. But at this point it's a matter of principle!

If you have A and C, then at one time you must have had A and B. Will Project UNDO help any here? I'm perfectly clear that Project tools fail at the worst possible time, which, I'm guessin' would be now.

Yeah, that'd be about now.

[steve]

Hmm, didn't seem to work

That doesn't really surprise me. There are lots of reasons that it can go wrong....and it has.

For some reason (and there are too many intricate possibilities to go into) the "A" and "C" tracks are slightly out of synch, leading to the mess in the final picture.

If you are interested in proving the "C - A" theory:
Record yourself talking (track A)
Record a second track of anything (track B)
Use Ctrl+Shift+M to create a mix of the two tracks (track C)
No do either.....
Invert track A
Select track A and track C and use Ctrl+Shift+M to reveal track B
or......
Invert track B
Select track B and track C and use Ctrl+Shift+M to reveal track A
or......
Invert track C
Select track A and track C and use Ctrl+Shift+M to reveal track B
or......
Invert track C
Select track B and track C and use Ctrl+Shift+M to reveal track A

If you are interested in getting back to your original task:
then I think it's time to abandon the first run through and make fresh recordings.

By the way, I'm using Audacity 1.3 not Audacity 1.2. That makes no difference to the theory, but I'm not certain if "Ctl+Shift+M" works on Audacity 1.2 or not.