Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Recording from CD or LP in 24 / 32 bit vs 16 bit (Read 9467 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Hi all,

Been reading posts / articles on recording from commercial CDs or ripping from LPs & using various bit rates.

Some advanced? users suggest using possibly 24 bit (or 32).  I guess it depends on what you do w/ the wav files once recorded? 

I have an X-Fi Titanium, so is capable of 24 bit, 96 KHz.  Recording live music at these settings is one thing, but if recording off a CD that is 16 bit, 44.1KHz, what is benefit of recording settings higher than the source?  If there is any benefit, in what circumstances would you be able to utilize the higher settings?

Along same line of thought, lots of encoding prgms will allow encoding any file (say, original source was 16 bit) at 24 or even 32 bit & much higher freq than orig.  Is there any use in that?  I briefly played w/ ripping commercial CDs at 24 & 32 bit & couldn't really tell any difference.

Recording from LPs - I'm not sure there'd be any benefit to higher than 16 bit.


Thanks.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #1 –
Quote from: Jenni28 on
I have an X-Fi Titanium, so is capable of 24 bit, 96 KHz.  Recording live music at these settings is one thing, but if recording off a CD that is 16 bit, 44.1KHz, what is benefit of recording settings higher than the source?  If there is any benefit, in what circumstances would you be able to utilize the higher settings?


Why would you record from a CD?  Just rip it. 

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #2 –
Yes, do DAE from digital sources, don’t record. There is nothing to be gained converting such data to a greater bit depth or sample rate than the source. It only uses up more storage space on your HD. If you do need to record, the same answer applies.

Recording from an analogue source, be it live or from some existing recording such as LP or cassette, might benefit from using a greater bit depth (24 bit integer or 32 or 64 bit floating point). The most likely benefits will be when recording live under good studio conditions with good professional equipment. A good recording is the first step to a good finished product.

You won’t find any difference with LP or cassette. 16/44.1 captures more than they have to offer. Although it is possible for an LP to contain higher frequencies that 44.1 can capture, only a few test disks are likely to do so. You might measure those, but you will not be able to hear them.

Using a greater bit depth is of theoretical advantage, the quantization errors are much smaller with higher bit depths, but an audible advantage is still questionable. Every step of processing after recording increases the quantization errors. With a floating point format those errors are too small to have any audible results. This is also probably true for any processing you are likely to do on 16 bit data, but the possibility at least exists there.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #3 –
Probably recording in floating point is meaningless as there is no widely available equipment with hardware FP ADC. So just capture th audio in the maximum bit depth supported by the hardware. Leave FP processing for temporary storage if you do some adjustments to your recordings.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #4 –
That the hardware isn’t floating point is irrelevant. If one is going to process in floating point format, it is easier and more efficient to record directly into floating point.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #5 –
Quote from: AndyH-ha on
Using a greater bit depth is of theoretical advantage, the quantization errors are much smaller with higher bit depths, but an audible advantage is still questionable. Every step of processing after recording increases the quantization errors. With a floating point format those errors are too small to have any audible results. This is also probably true for any processing you are likely to do on 16 bit data, but the possibility at least exists there.

As Andy says, there is a theoretical advantage to recording at higher bit depths in order to reduce the danger that accumulation of quantisation errors during DSP operations will become significant.

But if you're planning to do any clean-up of the recordings (eg. declicking, noise reduction, etc), then there is also a disadvantage to working at high bit depths. There is a host of potentially useful restoration tools out there, and some of them only work at 16 bit. If you insist on using higher bit depths, you deny yourself access to some potentially useful tools.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #6 –
Quote from: cliveb on
But if you're planning to do any clean-up of the recordings (eg. declicking, noise reduction, etc), then there is also a disadvantage to working at high bit depths. There is a host of potentially useful restoration tools out there, and some of them only work at 16 bit. If you insist on using higher bit depths, you deny yourself access to some potentially useful tools.


If your levels are right, 16 bit lsb errors will be a good 20 dB below anything you get out of a vinyl record.  So for that application I wouldn't worry about needing more bits except that more bits/FP makes the initial recording levels less critical.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #7 –
Quote
If your levels are right, 16 bit lsb errors will be a good 20 dB below anything you get out of a vinyl record.  So for that application I wouldn't worry about needing more bits except that more bits/FP makes the initial recording levels less critical[/b].

Yes, but don't take this to extremes either. If your recording level is too far below full scale then you could start to run into the inherent noise in your ADC.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #8 –
You've all been very helpful.  As usual, pros & cons.  But, most recording software  default to 16 bit / 44.1 or 48 KHz (espec. if has a "record from LP / cassette" setting).

Temporarily, a moot point as it appears the line in jack on X-Fi card isn't working.  Mic in jack on frt of pc will record, so I'll put that question in another post.  Look for it in Audio Hardware.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #9 –
Hi,

You've gotten some good advice already.  I would like to say there may be a case for recording LPs at 32 bit.  If you plan to do any removal of clicks and pops there is (reportedly) some advantage to working with a 32 bit depth file.  Here's a discussion that may give you some additional insights:

http://www.hydrogenaudio.org/forums/lofive...php/t10969.html

cheers

Kerkula

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #10 –
If you mean my post in that thread, I'm sure there was some specific context to the advice (it refers to another thread), or I was being over cautious.

Digitising and restoring/processing of LPs is fine in 16-bits throughout.

Cheers,
David.
 

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #11 –
Quote from: 2Bdecided on
Digitising and restoring/processing of LPs is fine in 16-bits throughout.

I thought 32 bit float was sensible for a couple of reasons:
1) You get some additional headroom on recording for unexpectedly loud peaks, and
2) This thread convinced me that since CEP and Adobe Audition do all their processing in 32 bit, then ...

Quote
"that suggests I am better off converting the file initially to 32 bit float since that would prevent Cool Edit Pro requantizing output back to 16 bits after each process"

So if you are doing any processing, at least in CEP or Audition, then you're going to be going 16 bit > 32 bit (anyway) then > 16 bit for every process. So why not start in 32 bit and then you only convert to 16 bit once, when you're done.

C.

EDIT: Grammar
PC = TAK + LossyWAV  ::  Portable = Opus (130)

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #12 –
Quote
Temporarily, a moot point as it appears the line in jack on X-Fi card isn't working. Mic in jack on frt of pc will record, so I'll put that question in another post. Look for it in Audio Hardware.


Could it be that you don't know exactly how to set the line-in as recording source (Audio Creation Mode) ... anyway, by using ASIO, you be able to access all I/O ports found in your card directly through recording software (though, you need to enable those ports through X-Fi mixer (Audio Creation Mode) or through system recording/playback options (on 32-bit OS (Windows) Start menu Run... option and sndvol32 /r (recording options) and without /r (playback)).

Juha

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #13 –
Thanks again.  I probably won't be using Cool Edit Pro - for now, anyway.  AFA adobe audition - don't know.  Not seen it.
For now, the features in Creative X-Fi titanium's software & Nero are fairly advanced - until get into some heavy technical stuff.  There's a lot to absorb - large learning curve.

But, I may just record a couple tracks in 16 bit & 24 and / or 32, then run some cleanup filters and see if * I * can tell a difference. 

I have used Nero in past on a few LPs & used Wave a bit, & results were pretty good - though LPs were NOT in bad shape.

Juha, there's a 100% chance I don't fully know what I'm doing.    This is 1st time I've messed w/ recording w/ the X-Fi titanium or looked at its settings / audio console.  I did finally get it to record.

Everyone, please see my other question on live monitoring:  http://www.hydrogenaudio.org/forums/index....showtopic=86213

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #14 –
Quote from: carpman on
So if you are doing any processing, at least in CEP or Audition, then you're going to be going 16 bit > 32 bit (anyway) then > 16 bit for every process. So why not start in 32 bit and then you only convert to 16 bit once, when you're done.


If whatever software you have is stupid enough to do this, you should probably just uninstall it and get something that isn't completely worthless.  I'm pretty skeptical it could be that stupid though.  It probably just uses floats for everything internally.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #15 –
One has the option to convert an integer file to floating point upon opening. If one chooses to use the 16 bit format, there is no other choice than to convert back to 16 bit after every operation. Here you have the option of simple truncation or dithering first.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #16 –
Quote from: saratoga on
If whatever software you have is stupid enough to do this, you should probably just uninstall it and get something that isn't completely worthless.  I'm pretty skeptical it could be that stupid though.  It probably just uses floats for everything internally.

That was my assumption, that EVERY operation would be 32 bit float until final save option, but read this thread, if you're skeptical.

All it means is that it's best to choose to work in 32 bit float to begin with, which is what I'd always done (unwittingly) anyway, then there's no issue. I don't think this renders Cool Edit Pro or Adobe Audition "completely worthless".

See this post and JAZ's reply; it does actually make sense.

Jenni28, re. live monitoring, AFAIR (bit sketchy) the routing went something like this:
Phono > Tapedeck in (record) > Tapedeck out [use tapedeck record level to boost the signal] > Soundcard in > Recording Software > Soundcard out > Amp > Headphones.
If I've got this a little wrong, I'll be happy to be corrected.
I don't know the Creative X-Fi titanium, so can't be any more specific on what settings you need.

C.
PC = TAK + LossyWAV  ::  Portable = Opus (130)

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #17 –
Quote from: saratoga on
Quote from: carpman on
So if you are doing any processing, at least in CEP or Audition, then you're going to be going 16 bit > 32 bit (anyway) then > 16 bit for every process. So why not start in 32 bit and then you only convert to 16 bit once, when you're done.


If whatever software you have is stupid enough to do this, you should probably just uninstall it and get something that isn't completely worthless.  I'm pretty skeptical it could be that stupid though.  It probably just uses floats for everything internally.
It's not stupid. If you insist on working in 16-bit, then for every operation that increases wordlength, it uses floats, and converts back to 16-bits afterwards (with or without dither - you have the choice).

It's clever, not stupid. It's very clever when it comes to things like reverb, if you think about how much multiplication and addition that entails for each and every sample.

Cheers,
David.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #18 –
Quote from: carpman on
Quote from: 2Bdecided on
Digitising and restoring/processing of LPs is fine in 16-bits throughout.

I thought 32 bit float was sensible for a couple of reasons:
1) You get some additional headroom on recording for unexpectedly loud peaks, and
2) This thread convinced me that since CEP and Adobe Audition do all their processing in 32 bit, then ...

Quote
"that suggests I am better off converting the file initially to 32 bit float since that would prevent Cool Edit Pro requantizing output back to 16 bits after each process"

So if you are doing any processing, at least in CEP or Audition, then you're going to be going 16 bit > 32 bit (anyway) then > 16 bit for every process. So why not start in 32 bit and then you only convert to 16 bit once, when you're done.
There's no significant reason not to, apart from the fact that it's slower, uses more disc space, and I'm sure at least some DirectX plug-ins just truncate to 16-bits anyway.

The point is, you can't get unexpectedly loud peaks on an LP. You know how loud it can go before the needle jumps out of the groove. Use a test record if you don't. If you set the recording level once in your whole life(!) so that the loudest possible LP can't clip, then the LP background noise is still well above the digital noise floor - even the ADC's noise floor.

You've got to do a stupid amount of processing for the repeated truncation or dither noise to get anywhere near the noise from the LP. If you add dither noise at -90dB ten times over the result is noise at -80dB (that's how uncorrelated noise adds - after a couple of iterations it only increases in level very slowly!). Given the LP noise floor is at -70dB, how are you going to hear the dither noise 10dB below it? Not that I've ever needed 10 separate processes on an LP transfer.
1 declick
2 decrackle (rarely!)
3 denoise
4 manual touch-up
5 EQ
6 reverb (no, I don't, but for the sake of argument!)
7 adjust gain
8 fade in/out (doesn't affect the music itself!)
9 split into tracks (no dithering needed)


I agree that it does help if you want to be very careless with the levels, either when recording, or when applying EQ. I use it in exactly these circumstances.
I agree that it can't hurt (unless a Direct-X or other plug-in truncates - or worse - without telling you).

Cheers,
David.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #19 –
Quote from: 2Bdecided on
Not that I've ever needed 10 separate processes on an LP transfer.
1 declick
2 decrackle (rarely!)
3 denoise
4 manual touch-up
5 EQ
6 reverb (no, I don't, but for the sake of argument!)
7 adjust gain
8 fade in/out (doesn't affect the music itself!)
9 split into tracks (no dithering needed)

I agree with David - even very enthusiastic audio restoration of LP recordings at 16 bit resolution has no realistic prospect of accumulating sufficient quantisation errors to bring it anywhere near the vinyl noise floor.

Of the operations David mentions, the only ones where added quantisation errors might conceivably concern us are EQ, reverb and gain adjustment. One other global operation that David didn't mention is filtering (eg. notch removal of hum). Perhaps he regards that as part of the EQ pass, but in my experience it's common to do gross filtering as a separate operation to subtle EQ tweaks. (FWIW, of these operations my typical LP transfer procedure involves only normalisation, and on odd occasions a little bit of EQ).

Broadband noise reduction typically uses spectral subtraction, and the artefacts of that are going swamp any additional quantisation noise.

Declicking, decrackling and manual touch-up make localised changes that are vastly greater in effect and have pretty low precision compared to any added quantisation noise.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #20 –
Quote from: saratoga on
I'm pretty skeptical it could be that stupid though.

It's about knowing how to configure the software in order to suit your needs.  Loading everything up as 32-bit and automatically applying dither when saving to 16-bit might not be what you want your wave editor to do by default either.

http://www.hydrogenaudio.org/forums/index....mp;#entry701712

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #21 –
This is all very informative.  I appreciate it.  Thanks everyone for the helpful links.

David, on your steps
5 EQ
7 adjust gain

After you've done steps 1 - 4 (when apply), what program are you using (vs one used to record?) for EQ & to adjust gain?

EQ:  are you trying to restore frequencies diminished by other processing, or adjusting it to your own taste?

Gain:  again, are you trying to restore some lost during processing or...  I assume the levels set during initial recording were acceptable?  Unless didn't listen to a short test recording?

In my limited experience, the denoise / noise reduction filters of some prgms reduce the volume noticeably.  But that depends on the threshold used and even the software.  I don't yet have a good feel for different softwares' default levels for any of the noise reduction filters - if they're generally way too much or just about right.

Recording from CD or LP in 24 / 32 bit vs 16 bit

Reply #22 –
Quote from: Jenni28 on
David, on your steps
5 EQ
7 adjust gain

After you've done steps 1 - 4 (when apply), what program are you using (vs one used to record?) for EQ & to adjust gain?
I use Cool Edit Pro because that's what I've been using for over a decade. You can't buy it any more (it's now Adobe Audition - more expensive and unecessarily complicated for this task). For declicking I mostly use the DirectX NR-2 plug-in from Sony - that's quite old too - not sure if it's still available. Sorry - that's not much help, is it?!

Quote
EQ:  are you trying to restore frequencies diminished by other processing, or adjusting it to your own taste?

Gain:  again, are you trying to restore some lost during processing or...  I assume the levels set during initial recording were acceptable?  Unless didn't listen to a short test recording?
I don't use all the steps I listed. I was just trying to think of all the steps I could, to see if it came to ten. It didn't. Hence 16-bits really is more than enough.

For gain, I'd peak normalise the whole album, or wavegain the result if I was mixing tracks from different albums onto one CD.

If you haven't read this excellent page already, then you should...
http://www.delback.co.uk/lp-cdr.htm

Quote
In my limited experience, the denoise / noise reduction filters of some prgms reduce the volume noticeably.
I have heard that, but it's usually when the settings are completely wrong - e.g. it's removing/reducing everything (so the threshold is far too high or the noise print is wrong), but it's only reducing the unwanted content by 10dB. Or else it's removing far too much. Be careful. You can do far more harm than good to the sound.

Cheers,
David.