Anomaly in W7 recording baseline?

I would have searched the Web/forum for this but didn’t know the jargon to describe what I see. I recorded a Web conference using Audacity (PC is running 64-bit Windows 7). The project file looks normal except for one oddity: the centerline for the blue waveform is slightly below the black horizontal line that starts at 0.0 and extends to the right. Specifically, it’s anchored at -0.1… Is there a way to fix this? Prevent it from happening again?

This is called “DC-offset” und is usually caused by using a minor-quality soundcard for recording.

Select the entire audio track and from the Audacity “Effect” menu, apply the “Normalize” Effect with the “Remove DC offset” checked.

Use a better-quality soundcard.

  • edgar

Thanks for the diagnosis AND the way to fix the project file. Don’t suppose you have any ideas about talking my employers into upgrading my sound card? Kidding.

Blackmail? (kidding :wink: )

And right now don’t be tempted to do the DC removal and the normalization amplification in the same pass. From 1.3.14 and in 2.0 there was a bug introduced whereby if you selct both options in Normalize the amplification is done first followed by the DC removal. It should be dne the other way around and 1.3.13 and previoud did it that way. I have reported this bug earlier today.

The workaround is to make two passes the first for DC removal and t’other for normalization amplification. In fact if you have a soundcard that produces DC on the signal (as you and I both do) then your basic workflow should be:

  1. capture
  2. DC offset removal
  3. all other editing …
  4. normalization/amplification
  5. Export

I personaly would like to see the two parts of the current normalize split apart - and I have written a proposal in the Wiki for that - see:

My real solution was to buy an external USB soundcard for my vinyl/tape transfers and FM captures - I only use the onboard card for occasional steaming audio capture.


To be fair, the bug does not seem to apply in “normal audio” cases, though I see that you have produced a test case that indicates there is an issue so it certainly should be investigated.

Yes but it’s still good practice to do the DC removal early and the amplification late on in the editing process, so you should be using Normalize twice anyway.

It was only testing that dug that bug out for me (I sometimes test the noisefloor by recording a signal and the Normalizing/Amplifying) - it just wouldn’t happen in my normal audio production workflow.