Troubleshooting audio peaking problems

Here’s a question I received about troubleshooting audio peaking problems in Audacity and Soundforge…

Hi Kenny, 
Here’s one for you! I use Soundforge 8 for audio editing. I downloaded Audacity Beta, imported a track, normalized to 0.20db….Perfect. I exported in Wav to file, opened the Wav file in Soundforge and pressed  play and made the sound maximum 0.20Db Max. Perfect. I then “saved as”  MP3 at 128KB, opened the mp3 in Soundforge…and guess what? The track meters were back in the “red” again! Your thoughts please?! 

Hi, thanks for your question.

If your wav file, when normalized, is showing you a good healthy signal and isn’t peaking (i.e. your meters are not flashing in to the red) then you know that your wav file is good.

If you then encode that wav file to mp3 and find that the mp3 file is peaking, then there are two possibilities:

1) The mp3 encoder engine in your audio processing software is introducing an increase in the files output volume when it encodes from wav to mp3.

2) Your mp3 player or audio processing software is reading the mp3 file differently to the wav file and so is “seeing” peaks that aren’t actually there.

Different mp3 players and mp3 audio processing software will “read” mp3 files in different ways, much in the same way as a DVD movie will play perfect in one DVD player but jump, skip or freeze in another (yip, we’ve all been there)! 

If you listen to your mp3 closely and can’t hear any clipping (distortion) then you know that the peaking isn’t really there and it’s just your software that’s not displaying the true output of the mp3 correctly. 

Fortunately digital distortion is easy to detect and hear as it is a very harsh, uncomfortable distortion compared to the more subtle distortion you get from older analogue sound. 

If it is peaking, your ears will tell you, so get a pair of good quality headphones and listen carefully.

Generally, you can expect a good mp3 encoding engine to preserve the sound of the original file pretty well, although some mp3 encoders give better results than others.

Arguably the two most common mp3 encoders which software manufacturers build in to their software are the lame or the fraunhoffer – most good audio software use one or the other. I hear many people speak highly of the lame mp3 encoding engine – I’m not one of them. I much prefer the fraunhoffer encoder. 

To find out which encoder your audio processing software uses, look through the techinical spec for your software and it’ll most probably tell you which type of mp3 encoder they use in it.

Also need to bear in mind that mp3 is a compression process, so to make a file 1/10th of its original size, the mp3 encoder needs to make some complex changes to that file. While the changes it makes may not be too noticable in the sound of the new compressed mp3 file it creates (which is the whole point of mp3 of course!), it is still making some pretty dramatic changes nevertheless. You may find it is simply struggling a bit to keep the normalization of the original wav file intact when it encodes…