Detecting a dub

[Dan McDonald]

Craig -

I've got some cds that were made from tape masters. I was reading in the forensics section about detecting if tapes were dubs or not.
I am planning to pull the wave files off the cd and run the spectrum analyzer as you suggest to see whether or not there is more than one hum. My question is:
Do successive generations each leave a hum (so that if a tape is 4-5 generations, will there be evidence of that)?
Does conversion to wave format leave any tell-tale residue? How about successive copies of cds? I have assumed that exact copies are exact copies (from CD), but is there actually any evidence relating to the number of dubs?

Thanks

Dan

[Craig Maier]

Re: Detecting a dub

To detect hum and multiple hums on a recording you must use the spectrum analyzer in its highest resolution mode of operation. If there are multiple dubbs, there will be multiple spectra around the 60 Hz area of the graph. If there are a lot of layers to the duplication, there may be a wide "smear" in the 60 Hz region. As for detecting digital copies of CD's, it is not possible to do this, as far as I know, using our software package. I know of no signal fingerprints that one could lock onto in order to make that determination. Maybe someone else has some insight into this. I think that CD clones are just that, perfect replica's of their source material.

[Dan McDonald]

Re: Detecting a dub

Thanks, Craig. This is useful to know. I guess, like everything else, it will take a little time to get used to. One additional question: if multiple copies are made with the same tape recorder (although I'm not certain why that would happen) would you still get a smear, or would you just get one line?

I'm just wondering.
Dan

[Craig Maier]

Re: Detecting a dub

Multiple recordings on the same machine would increasingly "smear" the line frequency signature.

[Dan McDonald]

Re: Detecting a dub

Craig -

Would European recordings have spikes at a different frequency? I am just trying to understand exactly what to look for. Also, I seem to find spikes at 60 and 120 on some recordings. Would the 120 hz spike be some sort of "echo" or would that be something in the source itself?

Dan

[Craig Maier]

Re: Detecting a dub

I think that is only possible if the DAT to CD transfer was performed in analog. If it was digital, no additional noises would have been introduced.

[Craig Maier]

Re: Detecting a dub

European recordings would have the noise spike at 50 Hz. The 120 Hz noise that you are seeing is the power supply full wave rectifier and filter capacitor noise getting into the recording. Low quality US machines will produce this noise at 120 Hz. European machines will produce it at 100 Hz. High quality machines have very little of this particular noise.

[Dan McDonald]

Re: Detecting a dub

Hi Craig -

That's what I was guessing, but wasn't sure. It's actually kind of interesting to check this on recordings where people have indicated the number of generations to see how well the report correlates with the imprints. I'm using reports of known generations to get used to the process so that I can better infer when no information is known.

I was looking at what was supposed to be a recording from analog to DAT to Cd and saw 2 spikes, which indicated to me that the DAT machine may have left an imprint. Would that be possible? I don't know much about how DAT machines work.

Dan

[greg_m]

Re: Detecting a dub

It seems to me that there are various sources of hum which would show up in the spectrum.

For example, if someone originally records mic to analog tape, hum could originate at the mic, mic wiring, or analog electronics; all of that hum would get onto the analog tape. Of course it would all be exactly the same frequency, the frequency of the power line where the recording was made (therefore either 50 or 60Hz). There might also be second harmonic (100 or 120 Hz) and even third harmonic (150 or 180 Hz, from three-phase power wiring) added in there.

Now let's go down the road in time, and play back that tape on some analog deck, patch the analog signal into a DAT machine, and make a DAT copy. If the tape is running at a slightly different speed from when it was recorded (which is likely), then the original hum (let's assume 60 Hz) will be played back at a slightly different frequency; let's say 59.876 Hz. That hum (plus any harmonics that might have been recorded) will be recorded on the DAT tape. In addition, any realtime hum from the playback electronics, connecting cables, and even from the DAT's AC power supply, will be recorded onto the DAT tape... these will be at the present power line frequency, which is (let's say) 60.0 Hz.

So in the example Dan asked about, it seems to me there might be two different hum prints: one that was recorded on the original tape, and one from the analog>DAT transfer. Both of these would be recorded on the DAT tape, and then would get transferred into the computer at a later time.

If the DAT>computer transfer was analog, AND at a different power line frequency from the DAT recording frequency, you might see a third hum signature from this final transfer; otherwise, not.

Incidentally, the power line frequency is not as stable as you might think. During the summer, when there are very heavy loads on the US power grid during the day (largely from air conditioning loads), the generators actually run a tiny bit slow, and the frequency actually is a little lower than 60.000 Hz! You can observe this if you compare an old AC-run electric clock (with a synchronous motor) to a newer clock with a crystal timebase. You will see the older AC clock eventually becomes a few seconds slow in the course of the day! However, the utility companies speed up their generators during the night, when the load is light, so that the _cumulative_ error is very nearly zero... by the next morning, the AC clock will be back nearly in sync with the crystal clock. Sounds amazing, I know, but I have observed it and it truly happens.

I haven't played with the DC-Art spectrum analyzer, so I don't know how small a frequency difference it can resolve; you might or might not see different spikes on a "summer day" and a "summer night" recording.


[This message has been edited by greg_m (edited 09-08-2002).]