dc bias

my problem is that i record, use high pass filter, use low pass filter,compress,normalize, amplify portions :question: of sound that are not loud enough, normalize and end up with an mp3 file that has a dc bias and is often not loud enough.

DC bias should be removed by a high-pass filter (DC is zero Hz).

Are you seeing bias during silent or quiet parts, or at the peaks? True DC bias will show-up during silence. Some sounds/instruments generate asymmetrical sound waves that have higher positive, or higher negative, peaks. Note that some “DC removal” filters will actually add DC bias while “balancing” the waveform… You end-up with bias during silence, which can cause a “tick” at the beginning & end of the file.

…and is often not loud enough.

As you know, when you normalize (or amplify and set the peaks to 0dB), you are targeting the peak levels. But, you may not know that perception of loudness is more related to average] levels (and frequency distribution). One little peak will determine the maximum unclipped (undistorted) loudness of the whole file.

…amplify portions of sound that are not loud enough,

That should help. You can also try some dynamic compression (Effect → Compressor).

Dynamic compression can boost the average level without boosting/clipping the peaks. If you are recording music, be aware that dynamic contrast is an important part of music, and if you over-do it, the whole thing will be constantly-loud and boring. (Most modern music is compressed to death to make it LOUD!.)

Effect > Normalize > Remove DC and de-select everything else.
Install Chris’s Compressor.
Increase the first value, Compression if the show isn’t dense or loud enough. I use 0.77 instead of 0.5.



We had the “DC Removal” discussion a while back. The one in Audacity is remarkably well behaved and doesn’t have beginning and end noise, or any other problems. The Hi Pass Filter technique can generate noise in some shows.

We found it was possible to “fake out” DC Removal only by scientifically generating waveforms that would never appear in nature.


If this doesn’t work for you, then you may have a raw recording that is so bad it needs rescue instead of post processing. Post back. Koz

It may create a new offset and thus a click if the offset is not constant over time or if some sections of audio are offset and some not. Missing features - Audacity Support .


if the offset is not constant over time

In which case the problem is serious low frequency distortion or damage – subaudible audio, not DC offset. Very different problems. In that case you may need both DC removal and the hi pass filter and then Chris. It’s possible with that much damage you may get pops at the beginning and end of the show no matter whose filters you use.

It could be said you need replacing the hardware more than a list of filters. Any hardware that produces damage like that is no longer usable and may create sound files that will damage speaker systems later.


My guess is that the “DC offset” isn’t actually DC offset, but is just natural asymmetry.
Could you post a short sample in WAV format so that we can see for sure? See here for how to post an audio sample: https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1