So, how many bits are we being asked to lose ... for a good recording ?

[fifty8th1]

so.... after reading extensively ... and proving to my own satisfaction ... that gain staging is the way to go....... how many bits are we shaving off. This is from a guy who felt that a bit was worth equal to and arm or a leg.

OK ... so maybe it doesn't matter how many bits are going to the wayside..but does anyone have a guess ? Would you say losing a dozen bits sound right ?

And now here is the biggest question .... by recording CD's at top volume .. are not we shaving off a lot of head room ... and losing out on a great sounding CD ? Isn't this a large part of the reason why CD's suck ? Why not record a CD at a peak of -6 ...

Thanks a lot for any comments...

[Greg Malcangi]

When you say "Gain staging is the way to go", I'm not really sure I understand what you mean or what your subsequent question is. Whenever you record or playback anything you are gain staging. You may or may not be gain staging correctly but you are gain staging. Am I right in assuming that when you say "gain staging" you are referring to referencing the calibration level of your PT system to 0dBV (+4dBu)? If so, and if your system is calibrated to the factory default of -18dB and if you do not exceed -18dB when recording then theoretically you are "losing" 3 bits of dynamic range. However, the massive dynamic range provided by 24bit more than easily accommodates this loss, in fact that's pretty much why 24bit was invented in the first place!!

"by recording CD's at top volume .. are not we shaving off a lot of head room ... and losing out on a great sounding CD ?" - Again, not quite sure what you are getting at here. A CD doesn't really need any headroom, although it's wise to give it a few tenths of a dB. The problem with the sound quality of many CDs is the overuse of compression and therefore the reduction of the dynamic range to only a few dB, the so called "loudness war". If the death penalty was introduced for anyone exceeding -6dB on a CD, it still wouldn't make any difference to the loudness war. Most CDs would still be massively over compressed. The only difference a consumer would notice is that their amp would be 6dB higher than it is currently.

I'm going to take a guess at this point and ask if you believe that more bits = higher quality? If so, this might explain your questions and your fears. Without going into any detail, you do know that more bits just means more dynamic range and not higher sound quality?

[albee1952]

Well said above. My take on it(much less technical) is that using 24 bit to record will give you lots of headroom(I tend to record just into the yellow on most things). Dithering when going to a 16 bit final file loses 8 bits, but the low level noise(that is dither) makes for a cleaner and more linear response at the bottom of the dynamic range(as opposed to truncating by converting to 16 bit without dither). Another personal choice I make is to always record, mix and bounce at 24 bit, and leave dither as part of a totally separate process, if I need it(like if I need to provide a client with an audio CD, I run the 2mix thru Wavelab, slap on a basic mastering chain of hi-pass, tube warmer and light limiting, with dither as the final process, render the file, run the DC offset process and save as a 16 bit.44.1K file). If the project is going to mastering, then the raw bounced file is the end of the line for me.

[formfunction]

Originally Posted by fifty8th1

by recording CD's at top volume ..

I record music.. not CD's. I think you are trying to reinvent the argument that has been going on since 1984.... And evidently doing it in ways that no one understands. Good Luck.. FF

[fifty8th1]

I'm going to take a guess at this point and ask if you believe that more bits = higher quality? If so, this might explain your questions and your fears. Without going into any detail, you do know that more bits just means more dynamic range and not higher sound quality?[/QUOTE]
.....................................

To Greg and Albee thanks for the really terrific answers...

YES I do .. or did ... think 24 bit was higher quality than 16 ...or..more is better.

This is why ...the gurus that were telling us how to record back-when explained that 16 bit ..means that digital recording is able to detect only a certain number of level changes ... it is not analog and cannot record 100 percent of volume changes .... So, it goes that 24 bit can record more level changes .... 32 would be even better. I have seen picture diagrams.

That is the way I understand it.... so less bits mean less accurate copy of the original ... SO ... you are saying Greg that this is not true ?

Thanks for chiming in on my question.... Your answers deserves a metal.
And albee I am going to reread your post a few more times to...

Oh . and Formfunction .... don't ever post to my questions again...

[fifty8th1]

From Greg
referencing the calibration level of your PT system to 0dBV (+4dBu)? If so, and if your system is calibrated to the factory default of -18dB and if you do not exceed -18dB when recording then theoretically you are "losing" 3 bits of
----------------------

That was the question precisely .. thanks...

[fifty8th1]

Quote from Greg..

A CD doesn't really need any headroom, ....
....................................

Thanks again... that was just what I wanted to know.

SO.....I am guessing here that a CD is not ... bare with me here ... A CD is not a digital thing ... it is just material that has the ability to copy. Whether or not audio has headroom or not ... the CD doesn't know ... or care. So ... if this is true .. and I know I am sticking my neck out here..... If an audio track is recorded very very low on the disc.... it will not suffer any lose of sound quality..

So, also.... whether I believe that less bits is less quality ... it would make no difference to a CD .... does any of this sound right ?

[Greg Malcangi]

OK, I see the problem. I have to say, your misunderstanding of how digital audio works is not uncommon. This subject has been covered numerous times in the past here and on other forums like gearslutz but I'll try to give you as simple an explanation as I can, without going too much into the math and scientific theory which details how it all works.

The misunderstanding of digital audio exists mainly because of two incorrect assumptions: The first is the assumption that more data means higher quality and the second is diagrams or waveform views in programs like ProTools. If you look at a waveform view in ProTools at say 24bit / 96kHz and compare it with a waveform view of say 16bit / 44.1kHz. The 24bit version will look less blocky and more like the smooth, continuously varying waveform we would see in an analog recording, so it's a natural assumption that when this 24bit waveform is sent to the speakers it's going to sound smoother and be more accurate. Unfortunately, this assumption misses some vital information: When you look at a waveform view in PT (or any other digital audio software or diagram) you are not actually looking at a waveform, you are just looking at a graphical representation of the digital data stored on your disk. This graphical representation is NOT what will be sent to your speakers! First of all, this digital data will have to pass through a Digital to Analog Converter (DAC) and one of the processes in the DAC will be to apply what is called a dithering quantiser. The Dithering Quantiser converts all those blocks of digital data into a smooth, virtually perfect, linear recreation of the original recorded signal. Without going into the math, it's difficult to explain how this happens but I'll give it a go :)

If we record a signal with say 2 bits of data, we have 4 available values, the points at which we measure the signal's amplitude (the sample point) is unlikely to match any of our 4 available values so our digital system will be forced into assigning the nearest available value. The difference between the assigned digital value and actual value is called the quantisation error. What a dithering quantiser does is use some clever math to randomise these quantisation errors and this randomisation is perceived as white noise. So with 2bits (4 values) available we can in fact achieve a virtually perfect reconstruction of our original signal, unfortunately there would be so much quantisation error that this perfect original signal would be almost completely masked by the resultant white noise. This is the basis of the Nyquist theorum. With me so far? OK, if we add another bit (3bit) we double the number of values available (8 values), which proportionately reduces the amount of quantisation error (noise), by half. Half the noise is -6dB, so each time we add a another bit of data, we double the accuracy and therefore reduce the noise by 6dB. So with 16bits, 16 x 6 = 96, 96dB is approximately the dynamic range of CD, 24 x 6 = 144, 144dB is roughly the dynamic range of 24bit.

In other words, 24bit is of course more accurate than say 1bit (or 16bit) but that added accuracy results ONLY in less noise (and therefore a larger dynamic range), the waveform coming out of the DAC is just as virtually perfect at 1bit as it is at 24bit. The obvious proof of this is the existence of SACD, which is a 1bit format. Obviously with 1bit SACD there is a huge amount of noise from all the quantisation errors but this is dealt with by moving ("noise shaping") the noise into ultra sonic freqs so it can't be heard. Using the theory that more bits = higher quality, a CD should sound roughly 32,000 times better than an SACD! The fact that if anything SACDs often sound higher quality than CDs proves that sound quality cannot be directly related to bit depth.

With all this in mind, you can see that even a 16bit CD is a bit of an overkill, as it has about 10 times more dynamic range than a vinyl disk and therefore 24bit is unbelievable overkill (about 1,000 times more dynamic range than a vinyl disk), so what's the point of 24bit? Basically, when recording and mixing, 24bit allows us to leave 18dB (3bits if you want to think of it like that) of headroom and not even have to think about it. Remember, headroom is only required to give us some leeway for processing and summing signals together (when mixing) or to accommodate unexpected peaks when recording. None of these considerations exist for the consumer and therefore headroom is not required on a CD.

Your last post, about CDs: Again, the CD is just a storage device for digital data, that digital data has to pass through a DAC before being sent to the amp or speakers. However, the lower you record to the CD the closer you are getting to the -96dB noise floor, not a problem until you get so low that the consumer can hear the digital noise floor as they crank up their amp to hear the music. We still have to consider signal to noise ratio (SNR) when going to CD, even though we have a fairly wide margin of error.

Hope this helps, G

PS. Although Formfunction was a little intolerant, I think you were maybe too harsh in your response. ProTools, as the name implies is a Professional Tool, for recording engineers, mix engineers, mastering engineers and other audio professionals. With the word "engineer" in the job description you really need to have at least a basic understanding of the underlying engineering. While myself and many others here will be happy to help with specific questions, it's just not possible to provide an audio engineering course on this forum or to write in a few posts what has taken us many years to learn. Maybe your "extensive reading" could do with being a little more extensive?

[WernerF]

Incredibly well said. Thanks for underscoring the fact that there is no substitute for real information that is stated in a clear way.

[fifty8th1]

Greg ..

All of us own studios here ... So, when we come here with questions we deserve some respect. You will always be welcome on my threads and posts.
Thanks for all the time you spent here on this.... Clearly I have read some poor info.... over and over again !