Bit by Bit

[HapiCmpur]

My Yamaha digital workstation generously offers me the option of recording in either 16 or 24 bits. However, the manual says that if I record in 24 bits, "the lower 8 bits will be discarded" when I do the mixdown to stereo. Three questions:

1. The lower 8 bits? Lower than what? The lower end of the frequency spectrum?

2. Why would Yamaha create a device that records in 24 bits if the machine's output is limited to 16?

3. Given this limitation, is it better to record in 24 bits and let the machine discard the "lower" 8, or would I get better results recording in 16 bits to start with?

 

[guitar zero]

The mixdown to stereo is the final step before putting it to CD. CD's are 16 bit recordings, so I think it's just making it compatable with the format used on CDs (correct me if I'm wrong) . It's just lowering the resolution, not changing any frequencies. Think of a picture and taking every third pixel out. You'll have the same picture, just not as clear. (gross over-simplification).

 

[Ripthorn]

Okay, hope this isn't too in depth. The number of bits relates to how many steps audio is broken up into in the digital domain. 24 bits means that your sound card is capable of recording down to extremely low levels and that it can accurately represent very small changes in amplitude. What they imply by the lower 8 bits is the 8 quietest bits. The way to prevent this (and the noise it can bring) is to dither your tracks down to 16 bits. This is a clever way of getting the benefits of 24 bit recording onto a 16 bit cd. You can do 24 bit mixdowns, you just can't put them on a cd.

 

[mixsit]

3. Given this limitation, is it better to record in 24 bits and let the machine discard the "lower" 8, or would I get better results recording in 16 bits to start with?

Put another way, it means doing recording with the wider dynamic range (lower noise floor) of 24 bit. Likely the only down side is bigger file size (less songs per drive size.
Unless there's other reasons on that particular machine not to, working at the higher bit depth would be the way to go.

 

[HapiCmpur]

What they imply by the lower 8 bits is the 8 quietest bits. The way to prevent this (and the noise it can bring) is to dither your tracks down to 16 bits.

If my workstation truly discards the quietest 8 bits during the mixdown, why would that create noise? The process of removing data is noisy?

I just checked the manual for my Yamaha. It says that the machine has a built-in dithering function, but the way the passage is worded, it seems that dithering can only be done if I'm sending the signal out to a DAT recorder via the digital stereo out jack. It's not clear whether I could also apply dithering during mixdown for the purpose of burning to CD-R. Yamaha manuals are notoriously inscrutable.

At any rate, it looks like dithering might not be an option for me unless I decide to start recording to DAT. (Or unless I get a more up-to-date workstation.) I guess guys who record on computers can buy plug-ins for this kind of thing, right?

 

[mixsit]

It seem that if they dither to the digital output it would also do so for the CD.

If my workstation truly discards the quietest 8 bits during the mixdown, why would that create noise? The process of removing data is noisy?

No, it's just that the very low noise floor and low end of the dynamic range (-144?) comes up to the neighborhood of -96 -still in most cases well below the noise you will have recorded along with your desired signals -room, mic, amp noise etc. Dither is used to smooth and make more benign the noise that will be down there at -96'ish.

 

[HapiCmpur]

it's just that the very low noise floor and low end of the dynamic range (-144?) comes up to the neighborhood of -96 -still in most cases well below the noise you will have recorded along with your desired signals -room, mic, amp noise etc. Dither is used to smooth and make more benign the noise that will be down there at -96'ish.

Thanks, mixsit. So if I understand this correctly, you're saying that certain low-level noises that would be inaudible in 24 bit resolution become audible (or more likely to be audible) when the tracks are converted to 16 bits. Have I got that right?

If so, and if it turns out that my workstation really doesn't have a built-in dither option, could I perhaps solve the problem by applying a noise gate to all my tracks during mixdown? I'm pretty sure my workstation has an on-board noise gate. Also, isn't it likely that a lack of dithering might not even be noticable as long as I stick to recording pop and rock tracks and avoid quiet string ensembles?

 

[mixsit]

Hey Hapi', I went back and dug up this info from RME on dither, relative noise and the effects of bit reduction. It's pretty old (from 99') but still relevant. I found it a good primer for laying this stuff out in a clear way.
http://www.rme-audio.de/english/techinfo/dither.htm

From 'Dither in the ADI-1 / ADI-8 PRO 'Remarks about the Need for Dither'

'.. However, their purpose is often to prove that A/D converters without dither would produce horrendous distortion, and therefore sound terrible. This is actually (willful?) deceit, as truncation of a digital test signal has little or nothing in common with truncation of an A/D converter's signal, and certainly doesn't lead to such 'catastrophic' results as seen in fig. 2. This is mainly thanks to inherent noise in the converter, as can be seen in the following measurements using an ADI-1. '

Thanks, mixsit. So if I understand this correctly, you're saying that certain low-level noises that would be inaudible in 24 bit resolution become audible (or more likely to be audible) when the tracks are converted to 16 bits. Have I got that right?

Yes. There was a nasty thread a while back here at HomeRecr's where fur was flying as to whether +/- dither could be proved to be heard at all. (ducks for safety..

If so, and if it turns out that my workstation really doesn't have a built-in dither option, could I perhaps solve the problem by applying a noise gate to all my tracks during mixdown?

No gate. (See the RME thing
a) Whatever noise (this one of the base quality' or lack of you have in your recordings) is also happening while the music is playing.
b)This is way down, likely below the noise in and from your recorded signal.

Also, isn't it likely that a lack of dithering might not even be noticeable as long as I stick to recording pop and rock tracks and avoid quiet string ensembles?

That's may take on it.
BTW, which Yamaha ya got?
ps. A nice experiment-Try recording in 16 bit (let alone 24) something nice and open' like accoustic guitar low enough to hear the converter noise above your analog noise. Now consider the dynamic range you have above that.

 

[HapiCmpur]

Hey Hapi', I went back and dug up this info from RME on dither, relative noise and the effects of bit reduction. It's pretty old (from 99') but still relevant. I found it a good primer for laying this stuff out in a clear way.
http://www.rme-audio.de/english/techinfo/dither.htm

Thanks for the link, Mixsit. It's a bit over my head (okay, it's a lot over my head), but I think I get the gist of it.

Yes. There was a nasty thread a while back here at HomeRecr's where fur was flying as to whether +/- dither could be proved to be heard at all. (ducks for safety..

I think I saw that thread when I ran a search before posting this one. It did get a bit ugly. More to the point, it got so esoteric and jargon-laden that I couldn't follow much of it.

BTW, which Yamaha ya got?

I have an AW2816, circa 2001. Like you said, if the machine will dither an outgoing signal, it makes sense that it should also dither a stereo mixdown, but there's nothing in the manual to suggest it.

ps. A nice experiment-Try recording in 16 bit (let alone 24) something nice and open' like accoustic guitar low enough to hear the converter noise above your analog noise. Now consider the dynamic range you have above that.

I have a strong suspicion that I won't be able to hear any difference between dithering and truncating. I don't think I can even hear converter noise in the first place. My ears just aren't that good. What worries me is that other people are going to hear noise in my recordings that I can't hear myself. I don't know how to get around that except to follow the procedures that other HomeRecrs swear by, and I guess that includes dithering.

 

[mixsit]

They say 'Don't sweat the small stuff'. If making and capturing good music is, say '100 on the scale?, this is .1.