tape to CD (Sample rate)

[ladiki]

Hi all,

I have a question concerning a transfer from a tape to CD, and to do a restoration.

My questions is:

i will use my Z-Link with Luna I/O / Audition 3.0 to do the transfer, should i use the higher sample rate, doing the restoration then down sampling to make the cd's

or it doesn't make a big difference.

Regards,
Ladiki

[chriskorff]

Yep, that's what I'd do!

Record at 24bit/96kHz, then do any noise reduction and EQ processing, and then do the sample rate conversion and bit depth truncation (with dither!) for the file you plan to burn.

HTH

Chris

Oh,

It just occurred that 24bit might be overkill if you're recording from a cassette — the tape/tape player's own noise floor is likely to mask any benefits of 24bit recording, rendering it a bit useless. Record at 16bit, then you won't have to worry about dither.

Cheers!

[Lima]

It depends on the kind of the tape and the quality of the registration IMO. If it's an audio cassette then its response don't go over 15-17kHz in most of the cases, so working with frequences above 44.1kHz is quite unuseful.
Otherwise if you're working on a DAT tape, or similar, the use of higher samplerate is worthy.

Same for the bit dept: if the recording is very low you should go with 24 bit, in this way you can preserve the dynamics, but if the recording is loud or (worst) distorted using higher bitrate is only a waste of disk space and processing time i my opinion.

[eliam]

I think 24 bit is worth it, but higher than 44.1... maybe you should run a couple of tests and hear if it's worth the extra hassle... Since the luna box is by no means a hi end converter box, I'm not so sure that you'll benefit much from higher SR. Don't take my word for it though, and let us know which road you pick.

[chriskorff]

Ah,

See, I assumed the OP would be doing some noise removal and possibly EQ, which would probably benefit from being done at higher SRs.

Bit depth - cassettes only have a dynamic range of 70dB or so (and that's pushing it! Also assuming a very high quality playback system). 16bit (CD quality) has a theoretical dynamic range of 96dB IIRC, so there is absolutely NO point in recording your cassette with a bit depth greater than 16 bits. The noise floor of the 24bit recording would still be -70dBFS (minimum!), rendering the other 74-odd decibels a complete waste of hard drive space.

About the sample rate thing though - if you wanted the easiest way to digitise a cassette, without bothering with noise reduction, you might be better off recording it at 16 bit, 32kHz! That would most likely remove a lot of the hiss in a not-too-destructive way (hiss which would be quite noticeable at CD quality).

JMHO

Chris

[garyb]

24bit 44.1k should be sufficient.

[chriskorff]

24 bit is overkill, I reckon - record at 16 bit. Not only does 16bit digital audio have a lower noise floor than the cassette, it will also eliminate the need to dither and truncate to 16 bit when you want to burn the file to CD.

Cheers,

Chris

[chriskorff]

Oops - I'm assuming it's cassette aren't I...

Which is it dude - 1/4-inch? 1/2-inch? DAT? ADAT? Cassette? If reel-to-reel, 30ips? 15ips?

Bugger. That's a lot of variables!

[Shroomz~>]

It probably depends on the particular model of tape machine as well (ie. it's conversion & output stage).

[Shroomz~>]

I guess the Luna conversion Ladiki's using could be a deciding factor on the bit depth of the recording as well. As for samplerate, I'd personally record at whatever samplerate the Scope system's running at.

[garyb]

go with the higher bit depth and the lower samplerate.

samplerate conversions are very violent while truncating from a higher to lower bit depth is relatively benign. dither isn't necessary, although you can use it if you like.

[Shroomz~>]

I think the most important thing for me would be to limit the number of conversions (if possible).

[chriskorff]

Which is why I say, go with 16bit.

While the 24-to-16bit dither & truncation might not be very audible, it is completely unneccessary to record at higher than 16bit if the delivery medium is CD (assuming the tape, whatever type of tape that is, has a dynamic range <96dB - which it probably does).

The higher sample rate might be advantageous if performing noise reduction and EQ on the digitised tape - particularly if the EQ is used at the higher end of the audible spectrum to remove hiss, which, in this case, is its most likely function.

Cheers,

Chris

[Shroomz~>]

Totally agree, but I'm wondering whether that approach might be better suited to using the card's analogue inputs instead of the Luna I/O box?

[Micha]

I do all the recording from tape 24bit/44.1. Much less flanging effects caused by noiseprint with 24 bit.
Do you use Osiris? Very practical, all I do after recording disk or tape is converting to 16bit.
Anyway, if no Osiris: Do you know the younglove script? Good values to check out inside.
Do you have Optimaster and/or PsyQ? These can do miracles. Sometimes, not always.
And always: the last step is converting down to 16bit.

[husker]

Which is why I say, go with 16bit.

While the 24-to-16bit dither & truncation might not be very audible, it is completely unneccessary to record at higher than 16bit if the delivery medium is CD (assuming the tape, whatever type of tape that is, has a dynamic range <96dB - which it probably does).

That might be true if you were going straight to CD and your record levels are exactly right and you're doing no post processing whatsover....otherwise 24 bit is a much better idea

[chriskorff]

Well, I suppose so...

[/16bit high horse]

But, if you're just digitising something that's already recorded to another medium, you have the luxury of being able to go to the loudest point and setting your level from there so that it doesn't clip.

From then onwards, all that's needed is a few dB headroom so that any noise reduction/EQ'ing you do in the digital domain doesn't clip the final result (and given that there's a good 20dB differential between the noise floor of, say, a cassette and 16 bit digital audio, there's plenty of room for manouvre).

So... when recording a tape at 16 bit, even if the peaks only reach -20dBFS, you'll still be well clear of the noise floor of the original medium, and there won't be any deterioration in quality at all, while still giving you ample headroom for digital processing. In all other cases though, I would absolutely advocate 24bit recording - not least because it allows you to concentrate on what you're hearing, rather than staring at a peak meter!

I am assuming cassette though, which might be completely wrong... in the case of a digital tape, I'd skip the extra conversion step and go via AES/EBU, ADAT, S/PDIF or whatever. I don't know the dynamic range of 2inch mastering tape though... that might be enough to warrant greater bit depths?

[Fluxpod]

16 bit makes 0 sense to use.Every daw available atm. has 24 bit.And i would like to hear what "tape" means? Cassette,1/4,1/2,2??? More info plz..
And capturing the source at the highest possible bit and samplerate is a BIG+ .
I always do transfers of cassette or "baked" tape on 24/96.Its mostly 2 track anyways so if you have the tools and the possibilities why not use them? Dithering noice and src´s aside.Capture at best available is my tip.

[chriskorff]

And capturing the source at the highest possible bit and samplerate is a BIG+

Absolutely! But, where the source has a definite peak (eg. an already recorded tape), and the source has a noise floor significantly greater than 16bit (eg. cassette), recording at a greater bit depth is a waste of hard disk space and will not benefit the end result.

Please remember that I'm ONLY talking about noise floor here, NOT 'accuracy' (either of the recording, or the digital processing done after the recording). Assuming a source with SNR <96dB, the noise floor of the source will MASK the inherent noise floor of 16 bit audio.

Live recording is, of course, another matter, requiring headroom for unpredictable peaks. Digitisation of analogue recordings, however, is entirely predictable.

Cheers,

Chris

[Fluxpod]

And capturing the source at the highest possible bit and samplerate is a BIG+

Absolutely! But, where the source has a definite peak (eg. an already recorded tape), and the source has a noise floor significantly greater than 16bit (eg. cassette), recording at a greater bit depth is a waste of hard disk space and will not benefit the end result.

Please remember that I'm ONLY talking about noise floor here, NOT 'accuracy' (either of the recording, or the digital processing done after the recording). Assuming a source with SNR <96dB, the noise floor of the source will MASK the inherent noise floor of 16 bit audio.

Live recording is, of course, another matter, requiring headroom for unpredictable peaks. Digitisation of analogue recordings, however, is entirely predictable.

Cheers,

Chris

I disagree 100% capturing of "tape" is never predictable...without all that noisefloor etc talk.It is better to use 24 bits..harddiskspace isnt anything to worry about.And nobody here knows how hot this"tape" was recorded..so it could very well be needed to capture it very hot or very soft.Its not as easy as some peps think...there are always thing that come by and get or can get you in trouble..and i dont like that so i just go 24/96 and never worry about what could have saved me some mb on the hd or how i could have avoided dither....or an src.This goes for stereo,if its 8 or more tracks..i do that only at work due no machines here at home exept a 1/2 studer.And i LOVE tape..i just bought 4 reels of 1/2 ATR Magnetics, brand new and it is the shit!!! I record everything in Cubase out of scope> tape and back into sx..it gives the glue to a mix you have to hear it ...and the smell of new tape gets me everytime i buy a reel .