Your message stated 1000+ and I might be up for that, though the
original poster I think said that this was unfunded and was a labour of
love. I have stopped my labour of love at the church because after five
years my paying work was suffering. HOWEVER, 10,000+ (mentioned in the
thread subject) is not something that I would be up for at the moment.
Thanks for the vote of confidence!
My setup is really aimed at high-end transfers using six Nakamichi
Dragons which can run in parallel into two RME Multiface II units. For
many of the high-end reel projects, I use an RME Fireface UFX, but it
only provides 8 line inputs and four mic/instrument inputs. I have a
MOTU 828 MK II and it was reasonably close to the RME Multiface (not
II). There was an improvement in the Multiface II and the Fireface UFX
is a further improvement, but the standout on the UFX is the mic preamp
(of which there are four). I MUCH prefer the RME driver software
including the fact that at least with the FF you can use it for direct
recording to the PC without using a DAW.
I will sometimes run processing upon ingest, tailored to the original
cassette quality (or more aptly, lack thereof) and generate twelve
stereo files: six raw and six with preliminary processing. The
preliminary processing varies but is generally a combination of
compression and filtering. Multiband compression can actually improve
intelligibility, especially if one party to a conversation is recorded
with a very boomy voice and the other is thin. The boominess gets
attenuated a bit if the wind is blowing in the right direction.
It is quite time consuming to do further cleanup in iZotope as even with
a quadcore i7 930 processor, iZotope in many noise-reduction modes will
run up all four processors to about 80 % and still take a significant
time to process. In the most processor-intensive configuration I've
found, a stereo 96/24 file takes approximately real time to process.
Click removal is perhaps the next slowest, again depending on settings,
running at 2-4 times real time. A simple task runs at 30-40 times real
time on this system.
There is an alternate option for the original poster that is probably
adequate quality. One dealer (http://www.avcom.co.uk/digitizer.htm) says
> How Long Will It Take To Digitize Your Tape Library?
> The Fast Way - Add a Digitizer to your computer, play your cassette at
> high speed and digitize up to 50, yes fifty, C90 cassettes in an
> eight-hour day.
> The Slow Way - Take at least 1 hour and 30 minutes to playback each
> C90 cassette into your computer's sound card. In an 8-hour day you
> could digitize a maximum of five C90 cassettes.
To which I add that with eight parallel players (a reasonable maximum)
you could, using the same assumptions, digitize 40, yes forty, C90
cassettes in an eight-hour day. That is assuming a very fast reload
cycle that may not be practical...
The manufacturer link for these ingest products is:
digitizes at 8x speed and produces 22050 sample/second WAV files,
limiting the upper bandwidth to about 10 kHz and is stereo.
offers a 4x speed and 44100 s/s WAV files.
These appear to do both sides in one pass so at 4 x the run time is 12
minutes for a C90 and 6 minutes at 8 x.
If you went with the high-speed model, I would still consider two or
three units running in parallel and get a spare transport module.
If you opted to run 8 x with two units, that would be 100 C90s per day,
so you could conceivably finish the job in 100 days. With 8 cassette
recorders I doubt you could keep the 40 tapes/day throughput but, if you
could, then it would take 250 days.
Compare that to doing one at a time in real time...that would be 2000
days...8.3 years based on a 240-day work year.
If you digitized MONO at 22050 s/s for 10,000 tapes at 16 bits that
would be about 250 MB per tape. Go to 44100 s/s and it would be about
500 MB per tape. Stereo, of course, would double that. Let's figure 500
MB per tape so that would be about 5000 GB for one copy of the project
(5 TB)...in very round numbers.
Any problematic tapes should be immediately reshelled and take it from
As a further note, I came across this as a useful bit of information
There might be additional useful information in TB30 which I co-authored
The reference for this type of work is IASA TC-04
IASA's TC04 on audio preservation is available online (and for a fee in
a print edition)
On 2013-02-21 6:19 AM, Tom Fine wrote:
> The real efficient way is to call up Richard Hess. This is the kind of
> cassette job he's set up to do. Trying to do 1000+ cassettes yourself,
> unless you are an expert with lots of half-decent audio gear at your
> disposal, is a fool's errand.
> -- Tom Fine
Richard L. Hess email: [log in to unmask]
Aurora, Ontario, Canada 647 479 2800
Quality tape transfers -- even from hard-to-play tapes.