Track Detection Problem

[Nivlem]

I am having a problem with the track scan feature of Wave Corrector. I have used the software for several years and have had no problems til now. I have had to manually locate/split the tracks on the last 12-15 albums and Cassettes that I converted to CDs.

At first I thought it was something peculiar to those Albums, but more albums tracks it failed top detect, the more concerned I became. It does not seem likely that it would be unable to detect tracks on every album or Cassette I rip. I tried the re-scan feature and it did not help. I even input the expected number of tracks, but it still did not find/split the tracks.

Eventually, I rescanned some archived raw files that had been split previously automatically. The software failed to find/split those tracks as well.

Visually, the “quite” sections just jump out at you. The Waveform is almost non-existent (amplitude wise) when compared to the rest of the track. Based upon my analysis of the waveform displayed on my computer, the quietest sections range from about 20-60 mV Peak-to-Peak in amplitude and last from 2-8 Seconds. This was followed by an additional period where the amplitude may vary from 80-100 mV Peak-to-Peak. All other portions of the waveforms in question are about 2500 mV in amplitude Peak-to-Peak.

If anyone had had this problem I would appreciate info on how you fixed it.

I e-mailed the support staff at Wave Corrector but I thought I would try this forum as well

Thanks

[Derek]

I'm not aware of any problem that might be causing this. When you put in a target number of tracks, does the program find no tracks at all or does it find track boundaries in the wrong places?

One thing that can cause the track detector to malfunction is the presence of inaudible noise (eg rumble) or if there is a DC offset on the D/A converter in your soundcard. But you description seems to preclude this because your archived recordings (which previously track-detected OK) are also affected.

I can only suggest that you email me a short section off an affected recording. Start the section a few seconds before an intertrack gap and finish it a few seconds after. I may then be able to discern some peculiarity of your recording that might be the cause.

If you have a broadband Internet connection, I can give you a URL where you can upload a longer section (eg an entire LP side). This would enable me to reproduce the problem and hopefully find a cause/fix.

[Nivlem]

Thanks for your help. I will try to clarify my problem a little; I have one problem with winyl media and another with cassette tapes.

(1) Vinyl: The Problem is poor Performance on the first Scan and incomplete track separation on subsequent scans.

In most instances the software only picks up the two sides of an album of the first scan. In those rare instances when it does identify additonal boundaries, tracks are merged (tracks 1-4 as a single track for expample).

The software miss identifies tracks about 60% of the time, even after several re-scams. It identifies all the tracks correctly about 40% of the time, after re-scan. (NOTE: This result is better than I originally reported. It appears that I may not have been entering the number of tracks, correctly).

(2) Cassttee Tapes: Complete Failure to Split Tracks: The software does not find any track boundaries.

I have a high speed conncetion and I would love to send you an extended version of one of my cassttee tape based files. If you will provide me the information how to do that, I will send it to you.

I seem to remember that the pervious version did a better job of identifing vinyl tracks than the current version. Is it unreasonalbe to expect this software to perform better on the first scan?

Again, Thank You

[Derek]

Thanks for the further information. I'll refrain from commenting further until I receive your problem files. I've sent you a pm with the upload instructions.

[Derek]

Thanks for the sample cassette and vinyl recordings.

Wave Corrector was unable to find any tracks on the cassette recording. I thinks this was mainly because the tracks all ended with a long slow fade out and that this was preventing the program from identifying the gaps properly. I'll see if I can tweak the algorithm so that it copes with recordngs like this one. I must admit I was rather surprised by this result and I think it should be possible to improve the performance.

The vinyl recording only found 2 tracks on the initial auto-scan. This was mainly because of the presence of significant noise (clicks and crackles) between the tracks. Rescanning with a target track count of 11 (the number of tracks on the recording), resulted in all the tracks being found correctly. A couple of the track boundaries needed slight adjustment but this is fairly normal.

I should explain that when track scanning is set to auto (as the initial scan always is) it depends on the consistency in the wave file. The program gradually increases the 'silence detection' threhold, and aims for a stable track count. If the track count keeps changing as the threshold varies, then the program gives up and you have to give it a manual target.