I bought a 2nd hand copy of “Elements” by Atheist (printed in 1993) from an online vendor (a risky move, I know) and all seemed fine until I went to play it.

There seems to be some crackling distortion during louder sections of music which is annoying to say the least. At first I thought it was just poor production but I decided to download some mp3s from the same recording to compare. The downloaded MP3s were fine in comparison. I then reasoned that the crackling was being diminished through the encoding process and tried encoding at the same bit rate as the downloaded file (128kbps) - but this also did not help. I’ve included some samples so you can hear this distortion for yourselves.

Note: These samples differ greatly in volume so try replaygaining them before you compare.

Sample 1:
Original WAV
Downloaded MP3

Sample 2:
Original WAV
Downloaded MP3

I noticed that the downloaded MP3’s waveform has much more defined peaks/troughs than the extracted WAV (EAC Secure Mode > 44.100 kHz 16 Bit Stereo PCM) and I wounder what is causing this – could the CD have simply deteriorated over time or could this be the result of some kind of CD resurfacing?



Physically the data surface of the CD looks perfect except for some very light scratches that are invisible unless held up to a light. Certainly there are no signs of oxidisation or fungal infection. The label side of the CD is in mint condition.

This may be nothing, but it seems the data area of the CD takes up slightly less space than on my other CDs – meaning there is slightly more clear plastic slowing on the very outer edge of the disk. Also when held up to the light, extremely fine circular tracks are visible all along the CD. I guess these tracks are made visible due to an older manufacturing method.

I’ve played it in a variety of players including my iRiver iMP-400, Sony CDP-597, Lite-On LTR-40125S and Pioneer DVD-117 using Sennheiser HD-265 headphones, Pioneer speakers and Logitech speakers all with the same results. PC is set up using foobar2k v0.8.1: ASIO output; track gain @ 89.0dB; no dithering; scale using peak info for tracks that still clip.

As this CD is no longer in print and somewhat rare I’m fairly keen to get it working properly (if this means I have to extract/equalize/re-burn/etc that’s fine). Any help would be greatly appreciated!

Thanks,
David Horgan.

The CD samples are difinitely distorded. They suffer from bad digital clipping. The waveform is truncated, and it causes annoying clicks. There is no way that a physical damage of your CD causes this.
These clicks are not audible in the MP3s. However, a close inspection of the MP3 waveform shows that it comes from the clipped version. If the MP3 is a copy of the CD, and the encoding removed the clicks, then why is it quieter ?
The MP3 might be a copy of a vinyl. The comparison of the frequency spectrums shows that they are identical until 7 kHz, above which the CD is louder than the MP3 (about 1.5 db louder).
The MP3 can also be a processed version of the CD, with clipping restoration.

I don't know if the loss of treble is caused by the MP3 encoding, the clipping restoration, the possible response of a vinyl reocording, or the possible noise removal process that would have been applied to a vinyl copy. Same thing for the fact the the clipping is no longer audible on the MP3.

All that we can say is that the MP3 comes from an original that clipped exactly the same way as your CD, and that it was intentionally processed (because of the volume difference).

Thanks for your reply Pio2001, great information.

I’ve been playing with Adobe Audition’s “Clip/Pop Eliminator” using the following settings:



I select the parts with audible distortion then click “Auto find all levels” using the most aggressive sensitivity/discrimination settings. While this does remove some of the crackling there is still a substantial amount left over, and if I run the filter again the loss in audio quality is worse than the original distortion.

Due to the limited nature of this method I’m wondering if this is the best course of action to take and if there are better settings and/or filters to use. Once again, thanks for your help!

Poseph:
Your samples have DC offset of -56db (somewhat high).
Trial 1. Do a high-pass filter 20hz on the WAV.There is some clipping still there, but improved.

If the sound is still not acceptable you can try this,

For Sample 1:
1.Using a Compressor module (settings for WAVE Rcomp)
Threshold: -6.9
Attack: 7
Release: 66
Ratio: 0.69
Gain: -4 (you will need to play with this value, it works for Sample 1)
2. Do Trial 1; check for clipping, if occurs then reduce Gain above (start from the original WAV)

The lost data can't be recovered but the mid + low frequencies are a closer shape.
The down side is that the dynamic range is increased which may not be to your taste.
Sample 2 has the same high DC offset.