When not using an audio card to its full potential...

pchorman

pchorman

New member
For those who only intend to digitize their analog recordings for transfer to CD (i.e. with no extra processing), it seems pointless to record at anything beyond 16 bits. My question is, what happens when a 24 bit card is commanded to record at 16 bits - are the 16 most significant bits or least significant bits used? Are all 24 bits used at the hardware level, only to be discarded by the controlling software, or are only 16 bit samples taken by the audio card?

Thanks for any insight you can offer.
 
The converters always operate at 24bit. Technically with a 24bit converter you will get a better quality 16bit file than using a 16bit converter. This is because you'll have usabe dynamic range well out past the 96db 16bit limit, so all 16bits *should* be good information. In an average 16bit system, you might have a bit or two of junk.

Of course your mileage will vary. There's about a zillion variables that come into play here, and we could wind up in another stupid 16/24bit debate.

Slackmaster 2000
 
Slack: I don't completely follow, but I'd like to, so please help me out...

Does this imply that the 16 MSBs are used by the recording software if I ask for 16 bit recording through a 24 bit card? If I had a couple bits worth of junk as you mentioned (at the noise floor) with 16 bit sampling, what does that imply about the number of bits worth of junk using 24 bit sampling - 8 more bits below the noise floor? That's the part which throws me. If the same number of bits are toggled at the noise floor using 24 bit sampling (assume 2 here), then the 8 extra bits are really at the upper end of the dynamic range. Only the 16 LSBs would ultimately get used in my application. See what's puzzling me?

The advantage of using all 24 bits seems to be more headroom before clipping. What do you say?

If I do what you suggest and take the recordings at 24 bits and then reduce to 16 for disc burning, does it matter which way I dispose of the 8 bits - dithering versus truncation?

thanks again
 
The advantage of using all 24 bits seems to be more headroom before clipping. What do you say?
Click to expand...
This is a myth. Recording 24 bit does not give you more headroom. There's no way to go above 0 without clipping in the digital realm. When people say 24 bit gives you more dynamic range, they're talking about the possible volume levels between -infinity and 0. In other words, going up in bit rate means that you have the freedom to be really quiet, not really loud! There are other people on this messageboard that can explain this better than I can.
 
wrong! 24 bit DOES give you more headroom - It was explained to me in detail by a soundcard developer (teratec) - ever noticed how the 24 bit systems have a max db of 106? and 16 bit has 96
 
Well, being able to be really quiet is the same thing as having more dynamic range. It allows you to record at a lower base level without noise and have the ability to peak.
 
At 24bit the noise floor is lower and you have more resolution per decibel over -96db. Therefore you certainly can record at lower levels which means that there's more distance between your peaks and 0db which means that you have more headroom. No myth, just relativity.

Whether you need to work at 24bit resolution is certainly up to your own abilities and the capabilities of your system.

Slackmaster 2000
 
I believe Full Scale Deflection (a left-over term from the days of D'Arsonval meter movements) corresponds to the following digital word for a 24 bit ADC:

1111 1111 1111 1111 1111 1111
And the corresponding minimum is of course all zeroes.

For a 16 bit ADC full scale is:
1111 1111 1111 1111
And the minimum is 16 zeroes.

Since the maximum signal level in both cases must be the same (i.e. one sound card can't pass "louder" signals to the hard drive or to DSP software than the other), then the extra 8 bits allow more "footroom" rather than headroom, as we use the latter term.
 
It was explained to me in detail by a soundcard developer (teratec) - ever noticed how the 24 bit systems have a max db of 106? and 16 bit has 96
Click to expand...
It's not the "max db" that's 96, it's the dynamic range, the difference between the quietest possible sound and the loudest possible sound the sound card can record. A 24 bit card can record a quieter sound than a 16bit card can, not a louder one.
Since the maximum signal level in both cases must be the same (i.e. one sound card can't pass "louder" signals to the hard drive or to DSP software than the other), then the extra 8 bits allow more "footroom" rather than headroom, as we use the latter term.
Click to expand...
Exactly!
Therefore you certainly can record at lower levels which means that there's more distance between your peaks and 0db which means that you have more headroom. No myth, just relativity.
Click to expand...
But the original post implied that you could somehow feed a louder signal into a 24 card and not have it clip. Sure, the distance between the quietest sound and the loudest sound you can record is greater, but it doesn't mean that you can push the signal above zero and not having clipping. Am I missing the point here?
 
In my mind (it's scarey in here), having more headroom doesn't necessarily mean that your system can take a *louder* signal. In a digital system 0db is always the "loudest" anything can be. Hear me out.

In *theory* you can record 48db lower in a 24bit system and still have the same bit resolution (e.g. 8bits*6db = 48db) as a 16bit system. But of course in the real world you don't get a full 144db of dynamic range out of a 24bit converter. While you'd be maintaining bit resolution (e.g. the number of values used to represent an audio signal), you'd run into a lot of mechanical/electrical/*real life* noise. I think think that the noise floor specs I've most often seen for 24bit systems have been in the -100 to -110db range.

At any rate, if you push the meters in a 16bit system you are using 65,536 discrete voltage levels to represent your audio stream, and your noise floor might be at -90db. But if you're on a 24bit system with 110db of dynamic range, which is 20db lower than the noise floor of the 16bit system, and you record such that your levels are down at around -20db, you are still sampling with a resolution of over 2 *million* discrete voltage levels, and with the same amount of real dynamic range as the 16bit sytem.

When you turn up the 24bit file from the above example by 20db (assuming the gain increase is constant), you end up with what is essentially the equivalent of the 16bit file that was tracked in the yellow. Therefore, in this example, you can record in the -20db range on the 24bit system without sacrificing fidelity, and have a whole lot of extra room for that guitar player who suddenly decides to go crazy.

Of course those are just a bunch of stupid numbers. In real life I don't know squat...but technically, it does demonstrate that you can track lower which essentially means that you have more headroom. Of course we could get into an argument over the meaning of headroom or something...but to me it's all relative.

Slackmaster 2000
 
I see what you're saying. Since you can record at a much lower level, than you have a lot of room to go up from there. That makes sense. I'm pretty sure the strict technical definition of headroom is this: "The level, over 0, that you can push a signal without clipping." In this sense, no sound card has any headroom. You can't send a signal louder than 0 into a sound card without it clipping.

Could it be that what 0 is (in voltage) changes from sound card to sound card? Maybe that's the key to what we're talking about. Decibels are relative.

In any case, I'm really confused. :confused: I'm going to bed.

Can someone much more knowledgeable than myself try and explain this to us? Harvey? Blue Bear? Maybe even Sonusman? (Just try keep the condescension to a bare minimum, Sonusman.)
 
Waaaaaaaaaaaaaaay too much is being thought into something that is very unscientific in nature.

Headroom is more or less a slang term used in audio. It's technical meaning is only relevent to any given audio chain, which includes the first conversion from sound vibration to voltage (usually microphone....) and end with your speakers. Wait!!! Your ROOM has something to do with headroom too, and so does the source sound!!!


Headroom is:

The distance between average volume and peak volume. PERIOD!!! If the average volume of something is only -6 lower than the equipments maximum operating level, then you have 6dB of headroom. PERIOD!!! You can start talking about many other things that may make more PERCIEVED volume, but quite assuredly, once you have reached the gears maximum operating level, the peak volume does not get any loud. PERIOD!!!

I am too tired to explain how this is relevent to the difference between 16 and 24 bit. But, the low level distortion in 16 bit systems means the audios average volume must be higher to retain a large sound to noise ratio and dynamic range when compared to 24 bit. But it won't take too long to research the difference on your own. http://www.digido.com is a good place to start.

Eddie
 
Doesnt everyone know that condescension is a given in any Sonusman post ;)

I really dont understand why there is any debate as to the merits of 24 bit vs. 16 bit. It's not like the analog/digital debate. Although I'm sure there are people who will argue that 1/2" tape is just as good as 2".

If you like digital and have the capability to use 24bit then just use it. Why waste all this time worrrying about it? 24bit has a better SNR than 16bit. The more tracks you are using and the more processing you are doing the more that will help you.
 
When recording in 24 bit resoltion the dynamic range does increase. This will be really useful when using many instruments. Since adding more instruments adds more noise, the noise level will increase overall.
Anyway one thing to remember is to make sure that when recording in 24bit, that the max volume of any single instrument must be close as possible to the max dynamic range ( just below clipping, 3dB?). This will ensure that better resolution is used. If you record too close to the noise floor then you will never attain optimum sound quality.
Correct me if I am wrong ;)
 
Tex, it's just fun to talk about.

amt,

I think it's always true that you'd want as much S/N as possible...but if the noise floor is down at -100db, it might not be as crucial that you track in the yellow, which in a round-a-bout sort of way means that you can create "headroom" (I said sort of). That's what we're talking about....or at least that's what I'm talking about...maybe I'm on the wrong page :)

Slackmaster 2000
 
I really dont understand why there is any debate as to the merits of 24 bit vs. 16 bit. It's not like the analog/digital debate. Although I'm sure there are
Click to expand...

Tex: I don't think anyone contested that, hopefully not on this thread (like the dozens of others). The original post asks what happens when a 24 bit card is commanded to record with only 16, for direct transfer to audio CD without the types of processing that lead to cumulative round-off errors, and errors due to finite word length. So we took a minor detour through headroom and footroom and whatnot, no big deviation.

Slack pointed out that it's still advantageous to use all 24 bits in this case, and now I can see the advantage after reading about dithering. I will do exactly that from now on. Who'd ever think that adding pseudorandom counts of noise just below the threshold of the LSB could actually improve audio intelligibility? Great revelation for me.
 
You must log in or register to reply here.
Back
Top