I've created this page as a rough guide for myself, and possibly others?, as a guide to determine whether a wave file has been encoded to mP3 at some stage. Some people go to extreme lengths to disguise mp3 sourced shows. Intricate editing, hiss reduction, level boosting...Any number of methods to deceive people. The people who do this and try to pass it off as non-mp3 are cunts.

I don't have a problem with mP3 in general. It serves as a great way to sample audio files downloaded from the internet. But it shouldn't be anywhere near the equation where live shows are concerned.
The artists go to great lengths to provide us with a quality sound >
The tapers go to great lengths to provide a quality sounding tape >
The traders go to great length to preserve this quality >
The listeners appreciate this

When a wave file is encoded to mP3, the higher frequencies are discarded to achieve a file size up to ~1/10th the size of the original. This smaller file size results in less time downloading. The different bit rates refer to the amount of data required per second. That is, 96 kilobits per second for a 61 second song is 5856000 bits, which is equal to 732000 bytes.

Now, there are many many different encoders out there, each claiming to 'sound' better, or encode quicker etc. Using a poor DAC, using a sound card that resamples, analog transfers et al can all effect the 'sound' of an audio file. This page will be used to compare the visual representation of these audio files.

The following graphs are pretty self explanatory. I've used the song Lukin by Pearl Jam as my test file. It was encoded to 96kbps, 128kbps, 160kbps, 192kbps, 256kbps and 320kbps constant bit-rate mp3 files. Variable bit-rate mp3 files were also created at 10, 50 and 100 values (where 10 is the lowest quality setting and 100 is the highest). These were created using MP3/mp3Pro fraunhofer IIs encoder via CoolEdit Pro 2.
I also transferred the song to mini disc and recorded at SP and LP2 modes using a Sharp MD-MT99W with ATRAC 3 compression.

The following graphs are the visual representations of the resulting audio files. Click on the frequency analysis to view the spectral view.

96kbps     0.732MB


Notice sharp decline after 11KHz and not much after that

 

128kbps     0.976MB


Notice sharp decline after 16KHz and not much after that

 

160kbps     1.22MB


Notice sharp decline after 16KHz and not much after that

 

192kbps     1.464MB


Notice sharp decline after 16KHz and not much after that

 

256kbps     1.952MB


Notice sharp decline after 16KHz and not much after that

 

320kbps     2.440 MB


Notice sharp decline after 16KHz and not much after that

 

VBR10     0.724MB


Notice sharp decline after 13.5KHz and not much after that

 

VBR50     1.003MB


Notice sharp decline after 15KHz and not much after that

 

VBR100     1.676MB


Notice sharp decline after 16KHz and not much after that

 

MD


Notice sharpish decline after 16KHz yet not full cutoff

 

MDLP2


Notice bluntish decline after 15KHz yet not full cutoff as well as lower db from 20-15000Hz

 

Uncompressed wave file     10.750MB

 

File
Frequency cut-off
Data after cutoff
96kbps
11000Hz
< -108dB
128kbps
16000Hz
< -108dB
160kbps
16000Hz
< -108dB
192kbps
16000Hz
< -108dB
256kbps
16000Hz
< -108dB
320kbps
16000Hz
< -108dB
VBR10
13500Hz
< -108dB
VBR50
15000Hz
< -108dB
VBR100
16000Hz
< -108dB
MD
16000Hz
~ -96dB
MDLP2
15000Hz
< -96dB       > -108dB
Uncompressed
none
> -84dB

Observing the frequency analysis of a song whilst playing, even an mp3, the values after the 'cutoff' may fluctuate, however these values are not a true representation of the file. In these images, I have chosen a specific point in the song where I can best determine the true file properties.

This is not a comprehensive analysis of mp3 frequencies and should not be viewed as such. Maybe someday someone will attempt to conduct a complete study. This would include testing involving many different encoders, different music files, even different data types, (ogg vorbis, VOF), however futile it may be.

Hopefully, in the not too distant future, with the introduction of widespread broadband access, increased hard drive capacities, and better lossless compressions, we may rid the world of the mP3 bug!