I am binning my 500 odd audio cassette collection. Most are of
commercial music that I have now replicated on cd/mp3 now so can be
binned. I have about 50-100 others of my own music or other stuff I
now wish to digitize. I have decided to record all tapes unaltered
into the pc, archive that off to DVDR, so I can bin the tapes while
having a decent unaltered original copy, then over time I can
remaster/edit some/all of them if I wish.

I plan to record via wavelab_live-input as follows

akai_cassette_deck -- DBX386@24bit_88khz -- RME hammerfall --
Wave_L3_Ultra_dither=16bit-normal --

These will be archived as the original tape copy to DVDR's. I will
consider using flac to fit more on disk, BUT as DVDR's are about 20p
these days may not bother to save time. (425mins per DVDR of .wav)

Later I plan to remaster from these .wavs using :-

waves_x-noise -- waves_linear_phase-eq -- waves_l3_ultramaximiser
(rem tape hiss) -- account for tape dullness-- make loud and punchy
These may well be kept on hard drive as flac or backed up as flac and
kept on hd as .mp3 for ipod/portability etc


SO - My issue is that cassette is not a high quality source, my tape
deck isn't great, the original recordings are not great and some are
pretty bad, also I'm doing it for nostalgic reasons, I wish to get rid
of the cassettes but don't know if I'll listen back to them much so
wish to spend the minimum time on this with a simple and fast solution
So I'm asking for help from anyone experienced in this with the
following ques:-

1) I presume it is better to set input at 88.2k rather than 96k if
saving out to 44.1k (as is exactly half freq). I appreciate that this
is way over the possible 15khz frequency response of a cassette but it
doesn't cost me anything to do it like this. Also the DBX386 has a
hard limiter pre the ADAC so it never clips so can be recorded quite
hot if required by the tube preamp.

2) I presume a dithered 88k_24bit-44k_16bit is the best way to record
the tapes and that 16bit_44k is more than adequate as a pre-master for
a cassette. i.e to store them as 88k_24bit would be wasteful, taking 3
times the disk/dvdr space.

I understand music software inside out having used it for years but
come from a computer background and though can audio engineer, don't
think I'm that great at it, so any other tips would be greatly
appreciated. I have googled and read for nearly 2 days, but nobody
talks about 96k or 24 bit in conjunction with cassettes, or solutions
to bulk archiving as a 2 stage process as I have concluded I need. As
this is a one off task before I bin ALL my cassettes for good I wish
to get it right before I begin this arduous task.

Cheers Danny K

I think the most significant link in this chain is the cassette deck.

Scott


WhoRu who@ru wrote:
: I am binning my 500 odd audio cassette collection. Most are of
: commercial music that I have now replicated on cd/mp3 now so can be
: binned. I have about 50-100 others of my own music or other stuff I
: now wish to digitize. I have decided to record all tapes unaltered
: into the pc, archive that off to DVDR, so I can bin the tapes while
: having a decent unaltered original copy, then over time I can
: remaster/edit some/all of them if I wish.

: I plan to record via wavelab_live-input as follows

: akai_cassette_deck -- DBX386@24bit_88khz -- RME hammerfall --
: Wave_L3_Ultra_dither=16bit-normal --

: These will be archived as the original tape copy to DVDR's. I will
: consider using flac to fit more on disk, BUT as DVDR's are about 20p
: these days may not bother to save time. (425mins per DVDR of .wav)

: Later I plan to remaster from these .wavs using :-

: waves_x-noise -- waves_linear_phase-eq -- waves_l3_ultramaximiser
: (rem tape hiss) -- account for tape dullness-- make loud and punchy
: These may well be kept on hard drive as flac or backed up as flac and
: kept on hd as .mp3 for ipod/portability etc


: SO - My issue is that cassette is not a high quality source, my tape
: deck isn't great, the original recordings are not great and some are
: pretty bad, also I'm doing it for nostalgic reasons, I wish to get rid
: of the cassettes but don't know if I'll listen back to them much so
: wish to spend the minimum time on this with a simple and fast solution
: So I'm asking for help from anyone experienced in this with the
: following ques:-

: 1) I presume it is better to set input at 88.2k rather than 96k if
: saving out to 44.1k (as is exactly half freq). I appreciate that this
: is way over the possible 15khz frequency response of a cassette but it
: doesn't cost me anything to do it like this. Also the DBX386 has a
: hard limiter pre the ADAC so it never clips so can be recorded quite
: hot if required by the tube preamp.

: 2) I presume a dithered 88k_24bit-44k_16bit is the best way to record
: the tapes and that 16bit_44k is more than adequate as a pre-master for
: a cassette. i.e to store them as 88k_24bit would be wasteful, taking 3
: times the disk/dvdr space.

: I understand music software inside out having used it for years but
: come from a computer background and though can audio engineer, don't
: think I'm that great at it, so any other tips would be greatly
: appreciated. I have googled and read for nearly 2 days, but nobody
: talks about 96k or 24 bit in conjunction with cassettes, or solutions
: to bulk archiving as a 2 stage process as I have concluded I need. As
: this is a one off task before I bin ALL my cassettes for good I wish
: to get it right before I begin this arduous task.

: Cheers Danny K

wrote:
I think the most significant link in this chain is the cassette
deck.

More likely, its the cassettes themselves. But you got the idea.

This guy is cleaning up figurative outhouses using a sanitzed
stainless steel shovel, wearing a fresh set of surgeon's scrubs.

"WhoRu" who@ru wrote in message
...
I am binning my 500 odd audio cassette collection. Most are of
commercial music that I have now replicated on cd/mp3 now so can be
binned. I have about 50-100 others of my own music or other stuff I
now wish to digitize. I have decided to record all tapes unaltered
into the pc, archive that off to DVDR, so I can bin the tapes while
having a decent unaltered original copy, then over time I can
remaster/edit some/all of them if I wish.


I've been through this and found using the higher sample rates (in my case
96k) to be a waste in every aspect and an unneeded headache.

On Thu, 7 Apr 2005 13:15:48 -0400, "Zigakly"
wrote:


You should leave the files at 24-bit until the mastering stage, where L3 can
dither down to 16. I don't see much advantage to processing at 88.2kHz at
any point.

Great reply, thanks. My main query does revolve around the above bit,
that given the source's lowish quality, if 16bit_44k would in this
case, be fine for a pre-master. I understand that NORMALLY you would
only go to 16_44 when 'finished' post-processing but for a tape I
wondered if 16_44 was plenty good enough?

I presume by what you say re 88k that you would just use 44k from the
off as the freq response of cassette doesn't justify any higher? I
wasn't planning to save at 88k just to downsample on the fly.

Re other reply - I understand the tape deck may be a weak link, but I
am wishing to simply do the best possible with the gear I have. These
aren't rare 'Hendrix' bootlegs or anything special, it's just a case
of 'if a jobs worth doing, it's worth doing well', within the bounds
of my gears abilities. I also have a Denon tape deck but it seems to
go a bit weird sometimes.


I may just record at 16_44 then as I realised I can use the dither
option on the DBX386 to do that on the fly, and as the DBX386 can be
run high without fear of clipping th extra 8 bits may be redundant?

PS - I have some tapes recorded using DBX Noise reduction, I don't
suppose there is a software plugin/method to decode this?

"Arny Krueger" wrote

This guy is cleaning up figurative outhouses using a sanitzed
stainless steel shovel, wearing a fresh set of surgeon's scrubs.

I agree with re the 88.2, part but even though you stated a few days ago 14
bit was enough to be inaudible, in this case, wouldn't 20 bit be helpful
since he's scrubbing up the noise floor?

Julian

On Thu, 07 Apr 2005 16:59:02 +0100, WhoRu who@ru wrote:

----Snip all-------

-- While it does have sense to go and stay at 24 bit until preparation
for burning the files, 44,1 kHz sample rate would be fine and would
ease your and your PC's job yielding somewhat smaller files.

What I would care of is the way the cassettes are played back at the
time of recording. It has been discussed many times here at the Board.
Mint clean heads and pinch roller would assure that the reproduction
is clean and crisp as it goes and there are no software solutions for
correcting the azimuth that are as good, as your screwdriver is for
doing it before recording for every cassete and its side. Also if the
recordings are Dolby, match it.

A huge amount of a doubtful salvage work could be spared by a good
playback.


Edi Zubovic, Crikvenica, Croatia

Julian Adamaitis wrote:
"Arny Krueger" wrote

This guy is cleaning up figurative outhouses using a sanitzed
stainless steel shovel, wearing a fresh set of surgeon's scrubs.

I agree with re the 88.2, part but even though you stated a few days ago 14
bit was enough to be inaudible, in this case, wouldn't 20 bit be helpful
since he's scrubbing up the noise floor?

It wouldn't hurt, but the noise floor is going to be WAY higher than
16 bits will handle anyway. Way higher than 14, even. These are
cassettes after all.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

OK this is all very useful, I will try and sum things up when the
thread completes for future googlers as I couldn't find other
discussions re 96k or 24 bit with regard to cassettes. 96/88k seems
deffo off the menu so far.

With regard to 24bit, and considering the original ain't going to be
great, is there much loss if it is input at 24 bit, but dithered down
to 16bit and saved for later remastering? (BTW is it OK to remaster
from a file that has been dithered down and maybe mildly limited using
ultramax?) It's just I'm thinking that maybe 24 bits is overkill for a
cassette, and it is after all a 50% larger file size.

Given that it's a 2 pass approach, firstly to just get all tapes on pc
quickly, then to later remaster them a bit, I wonder if in stage1 the
raw recordings should be recorded with absolutely zero processing? or
is ultramaximiser_16-bit premaster (edit-ready)? OK. I intend using
the exact same recording path for ALL tapes as I don't want to have to
listen to each one in detail prior to initially recording them. I
won't be keeping the tapes either so want like to get it right, but
also keep it simple.

For most tapes I hope to spend no more than 20 mins or so on later
remastering, just learn the noise_x footprint, apply a bit of eq and
ultramax nice and punchy. I 'will' archive the phase-1 recordings just
in case, but doubt I will ever return to them after.

Just trying to finalise the approach I will take before spending the
hours it will take to do it. Don't wanna get to the end and think 'Oh
No I shoulda done it this way' - doh!

"WhoRu" who@ru wrote

With regard to 24bit, and considering the original ain't going to be
great, is there much loss if it is input at 24 bit, but dithered down
to 16bit and saved for later remastering? (BTW is it OK to remaster
from a file that has been dithered down and maybe mildly limited using
ultramax?) It's just I'm thinking that maybe 24 bits is overkill for a
cassette, and it is after all a 50% larger file size.

I don't get what you mean here or what you meant earlier by converting 88 to
44 "on the fly" I'd say just do it 44/16.


Given that it's a 2 pass approach, firstly to just get all tapes on pc
quickly, then to later remaster them a bit, I wonder if in stage1 the
raw recordings should be recorded with absolutely zero processing?

ABSOLUTELY!

is ultramaximiser_16-bit premaster (edit-ready)? OK. I intend using
the exact same recording path for ALL tapes as I don't want to have to
listen to each one in detail prior to initially recording them

Except you SHOULD adjust the azimuth of your cassette player for every tape.


Just trying to finalise the approach I will take before spending the
hours it will take to do it. Don't wanna get to the end and think 'Oh
No I shoulda done it this way' - doh!

If you tweak the azimuth for each recording and optimize the Dolby dbx, etc
and archive the phase one recordings, you will never have to go back to the
cassettes, no matter what doh!, you might discover later. I recommend you
archive to 44/16 wav filed DATA not CDA format. OK?

Also, I am not aware of any software DBX, although you are teh second person
to ask recently - it would be nice. That's why I have an old DBX box
hanging out in the basement. You could rent it from me ;-)

Julian

WhoRu wrote:
OK this is all very useful, I will try and sum things up when the
thread completes for future googlers as I couldn't find other
discussions re 96k or 24 bit with regard to cassettes. 96/88k seems
deffo off the menu so far.

With regard to 24bit, and considering the original ain't going to be
great, is there much loss if it is input at 24 bit, but dithered down
to 16bit and saved for later remastering? (BTW is it OK to remaster
from a file that has been dithered down and maybe mildly limited using
ultramax?) It's just I'm thinking that maybe 24 bits is overkill for a
cassette, and it is after all a 50% larger file size.

My own personal reaction would be that anything higher than 44.1 kHz
would be overkill on a cassette. However, 24-bit vs. 16-bit is not
that big a space difference. Yes, it is 50%, but even if you have a
full 90-minute cassette, that's only about 240MB. If you're archiving
to DVD-R, those can hold something like 4 GB conservatively, so it will
take close to 20 cassettes to require an extra blank DVD-R.

Since you are talking about de-noising things and doing other
processing, personally I think it'd be nice to have the extra
bits of precision to work with and just dither at the end. It
might not be necessary, but since the cost is so low and since
you can't go back, I would just use the 50% extra space and not
have to worry about it.

- Logan

Julian Adamaitis wrote:
"Arny Krueger" wrote

This guy is cleaning up figurative outhouses using a sanitzed
stainless steel shovel, wearing a fresh set of surgeon's scrubs.

I agree with re the 88.2, part but even though you stated a few days
ago 14 bit was enough to be inaudible, in this case, wouldn't 20 bit
be helpful since he's scrubbing up the noise floor?

Common cassette tape recording is a really sorry format. This includes
both commercial pre-recorded tapes, and almost all homemade tapes. I
just moved our church from making HX cassettes on type 2 tape to CD
recording. It made a TREMENDOUS difference.

Cassettes have dynamic range of 8-10 bits. Therefore, 16 bit
resolution in the digital domain is more than enough for processing
it.

On Thu, 7 Apr 2005 14:57:18 -0700, "Julian Adamaitis"
wrote:

I don't get what you mean here or what you meant earlier by converting 88 to
44 "on the fly" I'd say just do it 44/16.

well i meant setting the live-input at 88k with it being resampled and
saved out at 44k in realtime. I thought wavelab could do this on the
fly but on testing it bombed out (even with diff sound card for output
set to 44k). So as a live sample rate convertor wavelab doesn't appear
to work. My thinking was to get the ADAC's inputs to the best signal
to get as much info sampled and into the digital domain, then use
software to resample/dither down to 16_44, hopefully giving a better
result than if merely sampled at 16_44 in the first place as the pc
has more info to play with. I also initially wondered if the noise-x
plugin would give better results with an 88/96k input as it has a more
accurate noise fingerprint to work with - (or not?)

is ultramaximiser_16-bit premaster (edit-ready)? OK. I intend using
the exact same recording path for ALL tapes as I don't want to have to
listen to each one in detail prior to initially recording them

Except you SHOULD adjust the azimuth of your cassette player for every tape.

so that means listening to each tape initially as I adjust azimuth to
get best sound.

so i shouldn't mildly ultramax-limit the original prior to saving,
though as the input is at 24bit i guess i will need to dither down to
16bit prior to it being saved using it anyway.



Just trying to finalise the approach I will take before spending the
hours it will take to do it. Don't wanna get to the end and think 'Oh
No I shoulda done it this way' - doh!

If you tweak the azimuth for each recording and optimize the Dolby dbx, etc
and archive the phase one recordings, you will never have to go back to the
cassettes, no matter what doh!, you might discover later. I recommend you
archive to 44/16 wav filed DATA not CDA format. OK?

Also, I am not aware of any software DBX, although you are teh second person
to ask recently - it would be nice. That's why I have an old DBX box
hanging out in the basement. You could rent it from me ;-)


are the frequency and compression changes it makes not known? and
therefore recreatable in a plugin? or at least approximated.

On Thu, 7 Apr 2005 19:29:58 -0400, "Arny Krueger"
wrote:


Cassettes have dynamic range of 8-10 bits. Therefore, 16 bit
resolution in the digital domain is more than enough for processing
it.


Is it worth the sampling being done at 24 bit but saved at 16 bit via
dithering or is this all pointless as 16bit-44k is more than enough. I
could dither either via wavelab or the dbx386 preamp which has it
built in prior to the AD. Maybe it is my own dithering that is the
issue here though .

It does seem crazy to worry too much about crappy old cassettes,
though my initial tests at remastering some old tapes I was able to
easily make them sound a lot better, by simply removing the hiss.
eq'ing and then ultramax hard limiting them for punch. There is
surprisingly little on the web regarding this, more for vinyl but that
is very different I guess and can perhaps justify 96k. I couldn't even
find out what freqencies would most likely need boost/sub to account
for the average cassette's freq response.

WhoRu wrote:

OK this is all very useful, I will try and sum things up when the
thread completes for future googlers as I couldn't find other
discussions re 96k or 24 bit with regard to cassettes. 96/88k seems
deffo off the menu so far.

Good choice. The power response of the cassette medium makes it sound
like much of the time its bandpass is like about 8 KHz.

With regard to 24bit, and considering the original ain't going to be
great, is there much loss if it is input at 24 bit, but dithered
down
to 16bit and saved for later remastering?

Remember, the basic dynamic range of garden-variety cassettes is 8-10
bits.

(BTW is it OK to remaster
from a file that has been dithered down and maybe mildly limited
using
ultramax?)

I don't know why you would need to limit a 16 bit transcription of a
cassette.

It's just I'm thinking that maybe 24 bits is overkill for a
cassette, and it is after all a 50% larger file size.

File size means nothing to me, because disk space is so cheap and
plentiful.

Given that it's a 2 pass approach, firstly to just get all tapes on
pc
quickly, then to later remaster them a bit,

One of the major audible changes in the remastering step will be
headroom optimization. You probably want to make your recordings with
up to 10 dB of headroom. Your final work product will probably end up

I wonder if in stage1 the
raw recordings should be recorded with absolutely zero processing?

My approach is to record as clean as possible. That makes all
processing reversable.

"WhoRu" who@ru wrote

well i meant setting the live-input at 88k with it being resampled and
saved out at 44k in realtime.

I am willing to be corrected, but I dont think that would give you any
different result from 44/16 bit to start with.

so that means listening to each tape initially as I adjust azimuth to
get best sound.

Oh yes. If oyu don't you won't be gettign all teh high freq information
from your cassette and that's more way important than what youdo with it
later.

so i shouldn't mildly ultramax-limit the original prior to saving,
though as the input is at 24bit i guess i will need to dither down to
16bit prior to it being saved using it anyway.

I wouldn't record any processing or you will have an opportunity to go doh!
later.

Also, I am not aware of any software DBX, although you are teh second
person
to ask recently - it would be nice. That's why I have an old DBX box
hanging out in the basement. You could rent it from me ;-)


are the frequency and compression changes it makes not known? and
therefore recreatable in a plugin? or at least approximated.

They are well known and someone could easily write a plug-in for it. I just
don know of anyone who has. You are welcome to search the web for it.
There were at least 3 or 4 different specs on DBX, so you have to make sure
you got the cassette tape one.

Julian

I would suggest that the single best thing you could do for this project
would be to find a Nakamichi 3-head deck in good condition, one which has
easy-to-adjust azimuth. You'd be surprised how much better cassettes sound
without the scrape flutter from the pressure pad inside the cassette. And
how much better still with the azimuth adjusted right.

Compared to those two improvements, the difference between 44.1kHz and
88.2kHz, or for that matter between 24 and 16 bits, is trivial.

Peace,
Paul

WhoRu wrote:
On Thu, 7 Apr 2005 19:29:58 -0400, "Arny Krueger"
wrote:


Cassettes have dynamic range of 8-10 bits. Therefore, 16 bit
resolution in the digital domain is more than enough for processing
it.


Is it worth the sampling being done at 24 bit but saved at 16 bit
via
dithering or is this all pointless as 16bit-44k is more than enough.

That's the question. I weigh in on the 16 bit side. But, to tell the
truth this is for all practical purposes a tempest in a teapot. It
will do little or no harm to do the processing in 24 or 32 bit
arithmetic. It then becomes yet another example of the kind of
technical overkill that is practical these days because of the immense
price/performance of modern digital hardware and computers.

I could dither either via wavelab or the dbx386 preamp which has it
built in prior to the AD. Maybe it is my own dithering that is the
issue here though .

Dithering will be no issue as long as we operate at 16 bits or more.
There's so much noise in the cassette medium that every reasonble
operation will be dithered by it, if only that.

It does seem crazy to worry too much about crappy old cassettes,
though my initial tests at remastering some old tapes I was able to
easily make them sound a lot better, by simply removing the hiss.

That would be reducing the hiss, not removing it, right? ;-)

eq'ing and then ultramax hard limiting them for punch.

You like it this way this week, what about next year?

There is surprisingly little on the web regarding this, more for
vinyl but that
is very different I guess and can perhaps justify 96k.

There's really no way to justify 96K over 44.1 K as long as human ears
are involved.

I couldn't even
find out what freqencies would most likely need boost/sub to account
for the average cassette's freq response.

If you look at the technical performance of the cassette medium, it
starts running out of power response at amazingly low frequencies -
like maybe 2-4 KHz. Boost above those frequencies is easily within the
limits of modern digital. Restoring spectral balance can improve sound
quality very nicely, thank you.

On Thu, 07 Apr 2005 16:59:02 +0100, WhoRu who@ru wrote:

I am binning my 500 odd audio cassette collection. Most are of
commercial music that I have now replicated on cd/mp3 now so can be
binned. I have about 50-100 others of my own music or other stuff I
now wish to digitize. I have decided to record all tapes unaltered
into the pc, archive that off to DVDR, so I can bin the tapes while
having a decent unaltered original copy, then over time I can
remaster/edit some/all of them if I wish.

How long do you want your digitised version to last? Another 10 years, 20
years or 50 years? DVD will probably be OK for 10 years but the rated
lifetime on current media is only 25 years at best. If you don't want to
have to salvage files from deteriorating DVD's in a few years time then
you are probably better off copying to decent archive quality CD-R's which
have a much longer predicted lifespan.

If you really value what is on the tapes then keep the original - we know
that most cassettes will last 30 years whereas no-one has a 30 year old
CD, let alone a DVD-R.

Cheers.

James.

James Perrett wrote:
How long do you want your digitised version to last?

Analog degrades over time (print-thru is one obvious symptom), and loses
something every time you copy it. And some batches of analog tape have
self-destructed. So the real answer is to focus on the data rather than
the medium.

Since digital media can be copied losslessly, the proper response to
this concern is to copy the data onto new media whenever you start to
distrust it. You'll probably want to do so anyway as denser media
continue to become available.

Of course there's absolutely nothing wrong with preserving _both_ media.
For irreplacable stuff I pull a pure digital copy (data CDR, with error
correcting codes for absolute accuracy), AND a digital audio copy (audio
CDR, which allows players to interpolate past failure points and thus
may be playable even after it has started to degrade), AND retain the
original media.

OK well then it seems the best option is to record all the cassettes
in at 44k 16 bit with no processing of any kind (no dither) and back
these up as the original raw backup. 88/96k or 24 bit seem overkill
for cassettes given their dynamic range.

Then over time these can be edited. Obviously when later remastering,
the processing in wavelab will all be done at 32bit_float internally,
as that's how it operates, despite the source being 16bit.

I had considered whether to record in at 24bit, dithering down to save
at 16bit via ultramax purely to use the full potential of my A/D's so
possibly getting a better end result than purely using the A/D's at 16
bit. But to be honest all you ever read is that dither MIST be the
last thing you do. This seems to mean that 24bit--dither--16bit will
be worse than direct 16 bit with no dither? hmmm don't really get
that.

Anyway I'll tweak the azimuth before each tape and will begin grabbing
them, (unless anyone has any last comments to make?) e.g "STOP no use
192khz the oberon theory says that the nyquist squared noise floor
threshold must be quadrupled to absorb the ambient affects of as yet
undecoded alien transmissions from ...."

On Fri, 08 Apr 2005 05:55:52 GMT, "Paul Stamler"
wrote:

I would suggest that the single best thing you could do for this project
would be to find a Nakamichi 3-head deck in good condition, one which has
easy-to-adjust azimuth. You'd be surprised how much better cassettes sound
without the scrape flutter from the pressure pad inside the cassette. And
how much better still with the azimuth adjusted right.

Compared to those two improvements, the difference between 44.1kHz and
88.2kHz, or for that matter between 24 and 16 bits, is trivial.

Only if you're trying to make sense.

Chris Hornbeck
6x9=42

On Fri, 08 Apr 2005 21:38:19 +0100, WhoRu who@ru wrote:

OK well then it seems the best option is to record all the cassettes
in at 44k 16 bit with no processing of any kind (no dither)

There's an error here.

I had considered whether to record in at 24bit, dithering down to save
at 16bit via ultramax purely to use the full potential of my A/D's so
possibly getting a better end result than purely using the A/D's at 16
bit. But to be honest all you ever read is that dither MIST be the
last thing you do. This seems to mean that 24bit--dither--16bit will
be worse than direct 16 bit with no dither?

*All* word length reductions must be dithered (to avoid yada yada).
This includes any analog to digital conversion. This whole thread
has been based on the assumption (good one too) that any real tape
will make enough noise so as to be much bigger than even a 16 bit
converter's LSB quantization errors.

IOW, the tape's noise will (possibly/ probably) dither the conversion
pretty well no matter what. But still:

You should still dither any word length reduction, including A/D
conversion. It's The Cowboy Way. And years later, you won't be
staying up late at night wishing, and wondering...

Good fortune,

Chris Hornbeck
6x9=42

"WhoRu" who@ru wrote in message
...
OK well then it seems the best option is to record all the cassettes
in at 44k 16 bit with no processing of any kind (no dither) and back
these up as the original raw backup. 88/96k or 24 bit seem overkill
for cassettes given their dynamic range.

Then over time these can be edited. Obviously when later remastering,
the processing in wavelab will all be done at 32bit_float internally,
as that's how it operates, despite the source being 16bit.

I had considered whether to record in at 24bit, dithering down to save
at 16bit via ultramax purely to use the full potential of my A/D's so
possibly getting a better end result than purely using the A/D's at 16
bit. But to be honest all you ever read is that dither MIST be the
last thing you do. This seems to mean that 24bit--dither--16bit will
be worse than direct 16 bit with no dither? hmmm don't really get
that.

No, it won't; among other things, because these days if you record 16 bit
you're just recording 24 bit with automatic dithering down to 16 bit (if
they're doing it right) or truncation (if they're not).

Personally, since you're planning to run these through Wavelab, I'd suggest
storing the files in 32-bit float, so it can have the format it likes. I
agree that the cassette itself will be much worse than that, but I like to
start with a format that's overkill if I'm going to mess with it. Dither at
the end of your processing. Oh, and don't throw away the 32-bit float discs
once you're done squeezing the crap out of the files. Someday you may decide
you want to listen to the music without the squeeze.

Peace,
Paul

On Fri, 08 Apr 2005 21:38:19 +0100, WhoRu who@ru wrote:
---------------8-----------------------------

Anyway I'll tweak the azimuth before each tape and will begin grabbing
them, (unless anyone has any last comments to make?)

Yes, I forgot to mention that it could be necessary to demagnetize
your cassette tape deck heads by a hand demagnetizer. The procedures
are discussed at various places at the Internet. This in some
circumstances could have a "wow" effect. This, plus cleaned heads and
tape path, properly adjusted azimuth for every cassette and its side
as well as matching noise reduction setting (whatever the tape has
been recorded with, if any) will ensure you optimal reproduction from
the analog part. And then you could choose bit depth, sampling rate
and so on (some editor programs have an option to work and save
temporarily files 32 bit, it can add to "internal" precision during
work no matter what bit depth you're working with. It also adds to
working payload though).

Edi Zubovic, Crikvenica, Croatia

On Sat, 09 Apr 2005 06:34:52 GMT, "Paul Stamler"
wrote:

Personally, since you're planning to run these through Wavelab, I'd suggest
storing the files in 32-bit float, so it can have the format it likes. I

I don't see the point of storing files at 32 bit, that doubles the
file size for no gain whatsoever, from what I've ascertained.
Obviously when I later come to remaster it wavelab will start with a
16bit file and internally as it runs through plugins do so at 32 bit
float regardless and dither back down to 16 bits at the end after
processing. Whather the source is 16 or 32 bit will make no difference
to wavelab, 32 bit would just mean extra dynamic range resolution that
won't exist on a cassette - correct?

Part of my confusion about dither was everything you read says to
dither down only as the last stage, BUT in this case I am preparing
raw copies of audio cassettes that it seems will not benefit from
going above 16bit/44k, then dither is OK I presume, though possibly
unneccessary as the inherant noise of a cassette will mean the high
bits are in a state of flux anyway. The main thrust of the thread was
to determine if anything greater than 16/44k was overkill which it
seems it is. I certainly dont wish to create larger files
unnecessarily, especially as when I tested flac on a 24 bit file it
made no difference but on a 16bit reduced it by about 30%, that's if I
decide to use it.

Bottom line is as it seems 16/44k is more than adequate to capture the
freq response of a cassette, there is no point using any higher.
Regarding dither there is no problem recording at 24 bit if I wish
with a dither down to 16 bit on the output and saving at 16 bit (on
the fly as I can do in wavelab).

On Sat, 09 Apr 2005 09:31:15 +0200, Edi Zubovic edi.zubovic[rem
wrote:

the analog part. And then you could choose bit depth, sampling rate
and so on (some editor programs have an option to work and save
temporarily files 32 bit, it can add to "internal" precision during
work no matter what bit depth you're working with. It also adds to
working payload though).

the chip in my pc is easily powerfull enough to run the 4 or 5 plugins
I need at 32bit res in real time when remastering, so there will be no
temporary files and no payload. I just dont want to being with higher
res files than I need as there may be 100+ tapes to do and don't want
to bunr more dvd's than necessary. Some people seem to say "no harm
storing at 24 bit" but if there is no gain then there is no point.
Arny says the res of a tape is 8-10 bits so... Also as I am recording
in via my dbx386 (dbx4 = no fear of clipping) and boosting the input
to -8 to -2 db range its pretty hot so is already using all the bits.
Wavelab bit meter shows the first 2 bits permanantly on anyway even
when there's no signal.

"WhoRu" who@ru wrote ...
the chip in my pc is easily powerfull enough to run the
4 or 5 plugins I need at 32bit res in real time when
remastering, so there will be no temporary files and
no payload. I just dont want to being with higher res
files than I need as there may be 100+ tapes to do and
don't want to bunr more dvd's than necessary. Some
people seem to say "no arm storing at 24 bit" but if
there is no gain then there is no point. Arny says the
res of a tape is 8-10 bits so... Also as I am recording
in via my dbx386 (dbx4 = no fear of clipping) and
boosting the input to -8 to -2 db range its pretty hot so
is already using all the bits. Wavelab bit meter shows
the first 2 bits permanantly on anyway even when
there's no signal.

I'd have to agree with Arny. You would be really lucky to
get 16 bits worth of dynamic range our of your old cassettes.
Storing 32 (or even 24) is just silly and wasteful, particularly
if you have "100+ tapes". 88K or 96K sampling rate also
seems optimistic at best. Stick with 44K x 16bit and get
on with your life.