xxxvii

jliat: still life #9 - skiffle: the eternal return cdr

This CD was to be based on a topic from alt.noise about relative values.. that the "same" wave with a dc offset would sound the same. Imagine a wave made from the following data +128 +64 +32 0 -32 -64 -128.. this should sound the same as +228 +164 +132 +100 68 36 -28... although the numbers are different. I experimented with several programs creating such waves and placing them at different offsets. A piece was made by first taking a PCM piano sample (middle C @ 7.5 seconds) and then placing it at offsets from +30000 to -30000, 480 in all, (about an hour) at displacements of 125. In some cases the offset did create a *difference* on certain CD players. Some would not play high and low offsets, one player would play these only at low volumes! Further a *problem* on some players appeared at 56 minutes. One player stopped - another played a burst of noise. I examined the source .wavs which showed no anomalies. I actually created two CDRs from different .wavs - one being mono - both had the problems at 56 minutes. I created another .WAV - examined this in an editor - the samples slowly moving from +30000 dc offset to -30000 - with no glitches. I used another program for burning and still had the anomaly. I then sent a .wav off to a commercial plant for a CDR to be made, this also shows the anomaly. Using software - CD rippers I tried to extract the data from the CDRs. Two rippers crashed at about 98% - I guess near the anomaly, I finally got a .raw image. This shows great bursts of noise around 56 minutes. So I then created a small section - 5 samples - at the offset around 56 minutes - a DC offset of -26000. This short .WAV (which looks fine) produces a CDR that some players cant play - one player after producing a burst of noise would only play a single channel. The data at around this offset seems to confuse the players DACs? Trying to read this back into a .RAW again caused some programs to crash. Creating a CDR based on a wav with no sample - just offsets from around -25875 to 26250 caused bursts of digital noise - only at an offset of zero. The current situation is as follows - a PCM data file of samples at +26000 - with variations of +2 to -2 will play (silence) with no problems. Another using a base of -26000 will not play on some CD players and I cant copy this back to a .RAW - all programs lock or crash. (I added the +2 -2 random variations as I noticed on the piano sample such variations - caused by the samples silence before it begins) Here are the two PCM files - each contains 100,000 samples. With t2.pcm I can burn and play cdrs with no problem - t3.pcm is based at -26000 - and acts it quite peculiar ways. I can burn a CDR - but not copy the data back, and get a variety of actions when playing audio. Mkpcm.exe will allow you to create short .pcm files at any base value. Do samples at minus 26000 to have some special effects on DACs? Please feel free to experiment and contact me.


available now