advertisement


Digitising cassettes: Is 16 bit / 44.1 kHz enough ?

serendipitydawg

Dag nabbit!
I've started to digitise my stash of mainly off-air cassette recordings. Is any bit-depth or sampling frequency in excess of 16/44.1 a waste of disk space?
 
Probably, yes. Cassettes tend not to have much in the way of high-end or dynamic range, so you’ll be fine. It’s all I used when digitising old band demos etc, and some of those likely have more hf than FM radio which is gone by about 14kHz.

PS 16/44 is actually really capable. A lot of music labels and mastering engineers spend a lot of time and money getting it to sound like shit. I honestly couldn’t tell a really well mastered CD from an SACD. As ever mastering is everything.
 
Most certainly, there is no need for any more than 16/44.1. You could easily argue even that was excessive - esp. for off-air recordings. The frequency and dynamic ranges, along with a high-ish noise level would suggest that 16/44.1 is technically more than needed.
 
The only reason you might want to do a higher sampling rate is if you plan to process them in any way. Otherwise 44.1 is sufficient as already discussed.
 
16/44.1 is orders of magnitude more than cassette or vinyl technically. More than sufficient.

I'll disagree for vinyl. You can get quite high frequencies out of a record, surprisingly high. For example, there were quadraphonic records with 4 channels encoded onto the 2 left/right channels using FM with a 30Khz carrier, so basically the 'rear' channels were in the range 20-40Khz, which would suggest it is possible there are records out there with real information above 20Khz. This would suggest an 88.2k sample rate might be necessary.

I agree that 16 bits is more than enough to capture the dynamic range.
 
I'll disagree for vinyl. You can get quite high frequencies out of a record, surprisingly high. For example, there were quadraphonic records with 4 channels encoded onto the 2 left/right channels using FM with a 30Khz carrier, so basically the 'rear' channels were in the range 20-40Khz, which would suggest it is possible there are records out there with real information above 20Khz. This would suggest an 88.2k sample rate might be necessary.

I agree that 16 bits is more than enough to capture the dynamic range.

Why do we need go past 20khz, when most adults can’t hear past 15khz and hearing declines with age?
 
Why do we need 20khz plus when most adults can’t hear past 15khz and declines with age.

We don't, but i'm not sure that is relevant to the discussion. I'm interpreting the original question as 'how can I made a copy of the contents of my tapes', and understanding the limits of the media allow you to extract the information from it. Whether it's audible or not is a different question entirely. As for 15Khz hearing, i'm lucky if I can hear 12Khz (one ear sometimes, not if i've got a cold, the other ear, nah)!
 
I'll disagree for vinyl. You can get quite high frequencies out of a record, surprisingly high. For example, there were quadraphonic records with 4 channels encoded onto the 2 left/right channels using FM with a 30Khz carrier, so basically the 'rear' channels were in the range 20-40Khz, which would suggest it is possible there are records out there with real information above 20Khz. This would suggest an 88.2k sample rate might be necessary.

I agree that 16 bits is more than enough to capture the dynamic range.

Except on most albums they had a filter at 16khz to prevent the cutting head from overheating. The Scully Westrex combo had an even lower filter point.
 
Except on most albums they had a filter at 16khz to prevent the cutting head from overheating. The Scully Westrex combo had an even lower filter point.

Are you sure this is right?

Looking here - https://www.stereophile.com/content/cut-and-thrust-riaa-lp-equalization-page-2

There is mention of adding attenuation in the pre-emphasis curve at 50Khz to avoid damage to cutting heads, in fact the article plots the response of the neumann cutter amplifiers out showing -12db at 100Khz (2nd order butterworth with a cutoff at 50Khz).

Anyhow, as has been pointed out, this is fairly irrelevant in practice, certainly there is no musical content up there, quality mics tend to give up before 20Khz, but i'd be interested in more info on the cutting head you are on about and the filtering that was applied, always interested in learning more about the process that was used to produce the stuff we all love.
 
1) The famous large diaphragm condensors of the golden age rolled off before 20kHz.

2) The 50kHz ‘Neumann pole’ is actually a myth. I have the circuit diagrams of several cutters somewhere …

3) From the late seventies on an awful lot of LPs were cut through a digital delay line, sampling at 44.1, 48, or 50kHz. Nothing above 20kHz from them except distortion. There exist LPs with ‘meaningful’ content up to 25 or even 30kHz, but they are very rare.


Above 10kHz LP replay is a mess. Why bother with the technicalities. Just find something that your ears like.
 
Why do you say that Cesare? Lots of dynamics and ribbons certainly but most decent condensers will have a reasonably constant frequency response up to 20Khz.

Here's a KM184 for example: http://recordinghacks.com/microphones/Neumann/KM-184

Yes, but something ubiqutous like a 414 is peaking around 15k and it's down probably 6db by 20Khz. Oh hang on, that site has the graphs: http://recordinghacks.com/microphones/AKG-Acoustics/C-414-XL-II

We have wondered off topic a bit though. The point is, records can record up to at least 40Khz, as there are quadraphonic records out there with audio in this region. It's of good enough quality that you can FM demodulate it and play it through rear speakers (not that i've ever heard this!), and this isn't fancy digital trickery, this is straight analog demodulation like an FM stereo broadcast. The other points are I believe discussing whether anyone bothered to utilise the bandwidth when cutting records, and it sounds like the consensus is no, they did not (with some exceptions of course!).

I've no idea of the bandwidth on analog multitrack, it's probably a bit rubbish, as they tended to squeeze more channels in rather than go for absolute quality, as that's what studios were screaming out for. The only large multitrack i've used was a tascam 16 track, and that was clearly made for project/semi-pro markets, and that market was quickly swallowed by ADAT and the digital revolution.


I've now wanting to track down a quadraphonic record, capture it and demodulate and see what's up there!
 
1)
3) From the late seventies on an awful lot of LPs were cut through a digital delay line, sampling at 44.1, 48, or 50kHz. Nothing above 20kHz from them except distortion. There exist LPs with ‘meaningful’ content up to 25 or even 30kHz, but they are very rare.

I'm guessing this is to enable some sort of dynamics analysis to automatically modulate the groove pitch? Sounds interesting
 
Yes, but something ubiqutous like a 414 is peaking around 15k and it's down probably 6db by 20Khz. Oh hang on, that site has the graphs: http://recordinghacks.com/microphones/AKG-Acoustics/C-414-XL-II

I'd have said a 184 was pretty standard kit too tbh but fair enough. And yes, OT and I certainly agree that the amount of musical information up there is pretty limited. I can't hear anything past 14Khz but don't feel hard done by.
 
I've no idea of the bandwidth on analog multitrack, it's probably a bit rubbish, as they tended to squeeze more channels in rather than go for absolute quality, as that's what studios were screaming out for. The only large multitrack i've used was a tascam 16 track, and that was clearly made for project/semi-pro markets, and that market was quickly swallowed by ADAT and the digital revolution.

A decent 2" tape running at 30ips will get you arguably upto 40khz, I certainly recall proving you could get our soundcraft to 30khz pretty flat.

Edit: and fwiw, I would be very happy at 16bit, 44.1khz for digitising regular compact cassettes. A 1/4" reel to reel, I might consider going up to 96khz.
 
Last edited:
I record analogue sources at 24 bit, that gives me a bit of margin for level/EQ/(insert your favorite post processing) adjust, then saves to 16 bit in Audacity. 16/44 is perfectly OK in itself.
 
Thanks for all the comments so far.

I'm contemplating using foobar2000s ABX function to compare differing bit rates & sampling frequencies
 


advertisement


Back
Top