Topic: Audacity "amplify" does not produce equal amplitude

[thunderbolt]

Problem:  (Audacity screenshot enclosed)

Unequal emplitude between tracks 1 and 2 despite amplifying both to -4dB, which I do for individual tracks prior to mixing and rendering a matrix.  Visually and audibly, they are not "matched."  If I listen on headphones, I need to bring up channel 2 by some 3dB in order to balance it.  Recorder is a 744T.  This is happening with different mics, and only on channel 2.  I generally remove spikes and things before amplification or normalizing.

It would be really nice to be able to match levels quickly in preparation for listening to determine the audience-board ratio.  I have generally used Amplify to bring each of the four tracks to -4dB.

Any ideas about what is happening here?

[runonce]

on the trackset drop down > Split Stereo Track

Then Amplify each channel independently.

Otherwise it picks the highest peak of the L/R pair and uses that for reference.

[thunderbolt]

Thanks for the reply.  I did split the tracks and amplified each one separately.

[page]

are you selecting each track and running amplify individually or are you selecting everything and running amplify?

got a 1min sample we could tinker with in google drive?

[runonce]

make sure you are selecting just one track...

Split > Click into the track > hit the back arrow - Edit > Select > Cursor to End

[thunderbolt]

I'm 99.9% sure  I'm selecting just the one track.  I click in the space to the right of "32 bit float."   (See JPEG) the box where you can split tracks, etc.  I start with the raw 4-track and don't mix to stereo until I have brought up equal amplitude in all four channels.  Then I create stereo tracks of the AUD and SBD, then time align.  Then I pull out the mixer board and adjust SBD gain to preference..  I then Mix/Render.  Then normalize to -.4 dB.

The thing is, I've been using Audacity (2.0) for 2 years, and while I'm not as good as a lot of people here, I have learned to time align, mix/render, EQ, stretch, etc., and haven't encountered this kind of obvious difference.  If it wasn't so visually and audibly obvious, I might not have noticed it so quickly.

What's interesting, too, is that a couple of weeks ago, when I first noticed it, I brought up the (right track) level using the slider on the mixer board, and it was quite a jump, perhaps 3-4 dB.  When I then went under "Amplify" to check how much it affected the gain, it had not changed: it still read -4 dB, even though I increased the gain about 3-4dB using the mixer board function (but this was prior to mix/render).

I'll post a clip later (cooling off from cutting the farkin' grass).   Thanks for looking at this guys, really appreciate it.

[runonce]

Im not sure just matching two channels to a certain peak is going to guarantee balance?

[page]

Im not sure just matching two channels to a certain peak is going to guarantee balance?

I thought about that, and thats why I wanted the sample. I was going to check RMS levels and peak levels and compare results to what thunderbolt posted. Could be any number of things from how it's getting selected or run, to a crossing of peak and RMS.

[runonce]

Im not sure just matching two channels to a certain peak is going to guarantee balance?

I thought about that, and thats why I wanted the sample. I was going to check RMS levels and peak levels and compare results to what thunderbolt posted. Could be any number of things from how it's getting selected or run, to a crossing of peak and RMS.

Right - a split second event on one channel - might create an imbalance when using peak amplify.

Not trying to thread jack - but I've always thought the lighter blue in the middle of the wave form is the RMS? And the dark the peaks...

[page]

Not trying to thread jack - but I've always thought the lighter blue in the middle of the wave form is the RMS? And the dark the peaks...

yeah, it's akin to that at least. I'm not sure it's calculated the same way or if it's just an analogous rating more than RMS specifically.

[runonce]

I suppose the possibility of something being out of phase is in play also - that can make a mix sound a bit lopsided...despite being visually balanced.

[page]

I suppose the possibility of something being out of phase is in play also - that can make a mix sound a bit lopsided...despite being visually balanced.

yep, that's another possibility.

Need a sample.

[sacchini]

My workflow:
- split stereo in two separate tracks
- calculate RMS for each track using "Contrast" in "Analyze" menu (I tried to guess English term since my Audacity is in italian)
- "Amplify" from "Effects" menu to balance L and R tracks
- create a stereo tracks again

[thunderbolt]

Link to my dropbox.  It's 16-bit WAV, 2 min of the mic source, 23 MB. 


https://www.dropbox.com/s/v4nuac2w9pp9llh/2-trk%20snip.wav

[thunderbolt]

I understand exactly what you're saying, Jon, but in this case, it's amplified music recorded from the PA, mixed with a SBD, so I feel like the soundstage (how sounds are "recreated" in the sound field) is pretty much toast relative to any accuracy.

I guess what has surprised me is that until very recently, applying (presumably) uniform gain has not resulted in what I'm calling an "imbalance."  I'm able to match levels at the time of recording, and that's reflected in the initial waveform I see when I load it up in the DAW.  It's only when I've amplified to -4dB that I see the imbalance.   The challenge, too, is that when adding a SBD to the mix, it can change things around, and to my mind, it destroys soundstaging (in the audiophile sense) even though I mix it in pretty low (-3 to 4dB on average, relative to the AUD signal).

This recording was somewhat unusual in that the performers marched around the [outside] venue, which is cool from an audio standpoint, but it made me decide not to use the SBD feed until they all arrived back at their stage mics.  The latency destroys the magic of hearing them marching around in a 360 degree loop.  Anyway, that fact may have contributed to the amplification anomalies.