At 01:58 AM 2/20/2006, Geeta Jatania wrote:
>Can you explain why it wouldn't be worth digitising cassette to 96/24?  We
>have some music tracks on cassette that we are currently capturing at this
>sample rate.

Let's look at it one way: Philips and Sony managed to convince most 
of us that the CD sounded better than cassettes or vinyl - then 
record companies went on to remaster many albums in an inferior 
manner, not the fault of the medium. Forgetting the vinyl issue, 
except for people on the Nakamichi mailing list, most of us think CDs 
sound better than cassettes. So, if the CD offers wider bandwidth, 
lower noise, and nonexistent wow and flutter, why digitize at any 
higher a rate than a CD?

To look at it another way: Unless you have recordings made on 
high-end Nakamichi (and perhaps a handful of other) machines, there 
will be little or no information past 20 kHz. The noise floor will be 
about 75 dB below peaks. CD quality provides a 20kHz passband and 
96dB S/N ratio (44.1/16).

Certainly if you have a high-end cassette made with excellent 
microphones of a musical performance, it doesn't hurt to digitize at 
88.2 or 96 ks/s and 24 bits. But, even there, it's overkill.

If you are hearing significant differences between 44.1 and 96 ks/s 
and between 16 and 24 bit digitization I would look at how the 
comparison was being done. There have been great audible differences 
even among various 44.1/16 converters. This is mostly ancient 
history, but if the upgrade from 44.1 to 96 included swapping to a 
different converter, then the audible difference may be in the 
converter. Try running your 96 ks/s converter at 44.1. Obviously, 
high-quality reels will show a difference. Most cassettes won't, I contend.

One of the reasons I keep bringing this up is that storage is cheap, 
but it's not free. If you're digitizing oral histories made on 
run-of-the-mill cassette recorders with run-of-the-mill microphones, 
there is nothing that you'll get at 96/24 that you won't get at 
44.1/16 except more detail of the noise. You won't get bias (which 
might be useful for removing wow and flutter).

96/24 takes more than 3x the storage as 44.1/16.

There will be far more improvement in high-frequency response if you 
manually (or automatically) adjust the play azimuth to match the 
recording than going from 44.1 to 96 ks/s. Even with totally tweaked 
azimuth, the response of a cassette on most decks (there were some 
exceptions with Nakamichi) plummets at 20 kHz.

It's easy to say 96/24 for everything, but if you can put 3x the 
amount of stuff in the same bucket at 44.1/16 and you haven't lost 
anything, why wouldn't you do that?

It's a balancing act. The cassette is the weakest link in the chain.

With that said, I do make a point in my lectures and seminars that 
cassettes shouldn't be dismissed as a high-fidelity medium out of 
hand. The example I play is a recording I made in a NYC church of an 
organ recital. I used a Nakamichi 550 portable and a pair of AKG 
C-451 microphones. I transferred it on a Dragon. It sounds good, but 
there is a hint of flutter. It was on a Maxell XL-II cassette. The 
recording was made in 1976, it was transferred in 2002 or so. It had 
only been played once or twice in the interim. I considered it a 
master and wouldn't play it regularly (another thing that causes 

I will also say, that I have yet to receive a cassette to transfer 
that came close to this one that I made.

I hope that helps.



Tape Restoration Seminar:    MAY 9-12, 2006; details at Web site.
Richard L. Hess                   email: [log in to unmask]
Aurora, Ontario, Canada       (905) 713 6733     1-877-TAPE-FIX
Detailed contact information: