Bit-depth info / question

A

aughii

New member
Hi, I've searched around trying to figure out what info is stored or is MEASURED by the bits when recording audio. So far I've only found out that bit represent volume. The higher bit resolution, the higher accuracy in VOLUME, and that 16 bit have abot 65535 possible volume-levels possible.

My question is: is it that simple, or is bit-depth something more than "volume-measurement" while recording digital audio?

And the higher bit-rate, the higher or better dynamic range. Is amplitude the equivalent word by meaning, as volume?

Thanks :-)
 
As I think you've discovered, the bit depth is a measure of the ACCURACY of the measure of volume/level on each sample. However, a higher bit depth doesn't equate to a louder recording. 0dBFS is the same whatever bit depth you're using and the extra bits extend down to the quiet signals rather than making things louder. There's a point of diminishing returns where you get a much more accurate representation of the noise floor!

Volume is a function of the alignment between the digital domain and the analogue for playback--most bits of gear are set so that 0dBFS in digital will be the equivalent of +18dBu, though sometimes this can be +22 or even +24.

The above applies to integer-based systems but it gets more complicated with "Floating Point" systems. Used only for editing/mixing, not recording, a floating point system slides the values around so they can be used as needed. This gives an artificially large dynamic range but means that in effect you can mix without fear of clipping.

However, even there, 0dBFS is the same thing when you convert your tracks to analogue to listen to them so you don't get any louder.
 
don't post wiki links... it's just noise. when a question hits the forum, it's because one need to get another angle or additional information
 
Bits represent volume steps. The more steps, the better the low volume resolution.

Every cycle of a wave crosses the zero point. with 16 bit, as the wave hits -96dbfs on the positive side, it will jump to -96dbfs on the negative side. With 24 bit, that jump doesn't happen until -144dbfs, which is way below the noise floor of any signal being recorded.

So all the resolution gained happens at the zero crossing of the waveform. More bits, less distortion at that point
 
aughii said:
don't post wiki links... it's just noise. when a question hits the forum, it's because one need to get another angle or additional information
Click to expand...
Dude.............
 
Last edited:
More bits doesn't mean higher resolution, it means increased dynamic range (difference between loudest peak and noise floor). The amount of volume change represented by one bit is exactly the same regardless of the word length ("bit depth").
 
bouldersoundguy said:
More bits doesn't mean higher resolution, it means increased dynamic range (difference between loudest peak and noise floor). The amount of volume change represented by one bit is exactly the same regardless of the word length ("bit depth").
Click to expand...

If I could ask a question for clarification. I've been using 16bit recording and paying quite a bit of attention to recording fairly hot to make sure the noise doesn't creep up too much if I have to boost a level. I've heard this isn't as much of an issue at all with 24 bit resolutions - that you can record comfortably at say maybe -18db and still not have noise be an issue. W/ 16bit I would try to record hotter than -18db if I could because it seems noise can become an issue. Is this a correct appraisal of a significant benefit of 24bit? Thanks

edit: I'm probably not being precise when I say -18db ---> I'm thinking more of pushing up into allowing more yellow and occassional red on the meters w/ 16bit - but have heard this isn't much of a consideration w/ 24bit and can back off on recording levels without having to worry about noise. thanks,
 
Last edited:
The relaxed levels requirement and greater headroom is probably the most tangible benefit of recording 24 bit audio. Another (possibly theoretical) benefit is that 24 bit digital audio is subject to fewer rounding errors when converting from floating point math (eg in the DAW) to 24 bit integer (audio file) as compared with 16 bit integer.

Conventional wisdom suggests that tracking and all digital processing should be done at 24 bit. Then, if down-sampling/dithering is required, it should be performed as the very final step before export.
 
PRHunt said:
The relaxed levels requirement and greater headroom is probably the most tangible benefit of recording 24 bit audio.
Click to expand...

Yep. I don't want to spend the session getting stressed about levels, I want to spend it getting good tones and good takes.
 
You should always record at line level and make sure that the peaks don't go over 0dbfs. Everything else will take care of itself.

The reality is, 16 bits is enough not to have to worry too much about recording levels. Most of the time, the digital noise floor is going to be below the analog noise floor of the signal you are trying to record. As long as that is the case, you are not losing any dynamic range or 'resolution'.
 
smili said:
I've heard this isn't as much of an issue at all with 24 bit resolutions - that you can record comfortably at say maybe -18db and still not have noise be an issue. W/ 16bit I would try to record hotter than -18db if I could because it seems noise can become an issue.
Click to expand...

Jay gave you the right advice. I record stuff that averages -20 dBFS using "only" 16 bits all the time. My home studio is very quiet, but the acoustic room noise still dominates. Recently I accidentally recorded a track through a guitar amp that peaked at -40 dBFS and that too was fine. The background noise was about equal to what you get from a good analog tape deck recording near 0 VU.

Of all the things to fret about when recording, not capturing a loud enough signal with 16 bits is near the bottom of the list.

--Ethan
 
I guess the other practical consideration is that 24 bit is now the native and default word length for many devices.

With my Echo Audiofire, I don't have the choice to use anything other than 24 bit word length. (This is under Linux).

The take-home message is understand your gear and work within its capabilities.
 
The bottom line is that the wordlength is not going to be the thing that limits your recordings potential.

Metallica's Black album was mixed to 16 bit dat. It's one of the better sounding albums and it sold over 20 million copies.
 
PRHunt said:
24 bit is now the native and default word length for many devices. With my Echo Audiofire, I don't have the choice to use anything other than 24 bit word length.
Click to expand...

Are you sure about that? I don't have an Audiofire to test, but I've never seen a modern converter that can't record at 16 bits. The bit depth is set in the recording software.

--Ethan
 
Are you sure about that? I don't have an Audiofire to test, but I've never seen a modern converter that can't record at 16 bits. The bit depth is set in the recording software.
Click to expand...

Hi Ethan,

It is actually the case that I can not select 16 bit sampling. However, this is not a limitation of the Echo hardware, but a software design implementation decision.

Here is why:

I use the Linux operating system. Firewire audio device support in Linux is provided by a unified driver called FFADO, and the developers took the decision to lock in at 24 bit.

Evidently the packet size for firewire data is fixed, so if a smaller word length was chosen, the data packet just would be padded out to fill the unused portion of the buffer. There would therefore be no firewire bandwidth benefit to using a smaller word length. Although it would technically possible to use 16 bit, the devs have not made this a priority.

Hope that clarifies! Paul
 
bouldersoundguy said:
More bits doesn't mean higher resolution, it means increased dynamic range (difference between loudest peak and noise floor). The amount of volume change represented by one bit is exactly the same regardless of the word length ("bit depth").
Click to expand...

Not quite, but close. More bits means greater precision and also more dynamic range. In fact, precision and dynamic range are really the same thing.

More bits means that you can reproduce a value more accurately. When you take a measurement with higher precision, you can detect a difference between two small values; by contrast, if you measured those same values with lower precision, you would get the same value. Therefore, by sampling at a higher precision (bit depth), you are able to reproduce tiny differences in value, and thus quiet sounds. Thus, the increased precision results in a bigger difference between the maximum value you can reproduce and the digital noise floor (the minimum value you can reproduce).

Put another way, increasing the bit depth lowers the digital noise floor. Because noise is additive, less is better. Therefore up to the point at which the analog noise floor dominates, increasing bit depth results in better recordings. Of course, if you're going to truncate to 16 bits in the end anyway, this may or may not matter, depending. :)
 
dgatwood said:
Not quite, but close. More bits means greater precision and also more dynamic range. In fact, precision and dynamic range are really the same thing.
Click to expand...

My comment is meant to counter the idea that more bits means more bits per given amplitude difference. One bit represents exactly the same amplitude difference whether there are 8 or 16 or 24 per word. The resolution is always the same, 6dB per bit.
 
You must log in or register to reply here.
Back
Top