Yes, there is more headroom available in a 24 bit recording, letting you get a better signal to noise ratio. you also have more volume to play with. Some s/w programs do sound a tiny bit better when using 24 bit, mostly because of the s/n ratio.
Plus, the advantage is that recording at 24 bit 48KHZ can provide DVD-Audio and DVD-Video compatibility, and you can easily downsample to CD-Audio (16 Bit 44.1KHZ).
16 Bit is good enough for most applications, but if you are doing Video Production at all or 5.1 surround sound, I recommend going to 24 bit, at least when you are due for your next upgrade. I recommend a MOTU 2408 MKII, Tascam TM-D4000 or DM-24 and Mackie SDR 24/96. The only problem right now is the fact that Vegas Video can not import BWF files, so you can save time coming from the Hard Disk Recorder, dumping via USB, and openning your tracks up in Vegas Video...

Thanks for the explanations. My cams are SONY...PD-150 & DSR-250. Don't they record audio in 16 bit?
Andyd: Are you suggesting that I should get an audio set up that records in 24 bit to supplement the Cam's audio? Or are you suggesting I convert the 16 bit Cam audio into 24 bit audio in post?

I'd say dear Andy's shooting a trifle over your head, there... Try two or three LIGHT YEARS above your head. With the type of sound you're typically recording with your camcorder, I'd say you would barely heard the difference between 8 and 16 bits!

For your specific application there is no need for 24 bit. All the advantages Andy speaks about are specific to studio environments with extremely highqual recording equipment and heavy destructive processing. Besides, if the material is originally in 16-bit (which would be a fair educated guess) there's even less sense in 24-bit.

"Yes, there is more headroom available in a 24 bit recording, letting you get a better signal to noise ratio. you also have more volume to play with. "

This is not true. This is a misunderstanding I've seen many times in these forums. There is a better signal to noise ratio in 24 bit, but it is not due to an increase of dynamics range. The increase signal to noise ratio is due to less distortion created by quantization errors. 0dB is 0dB in the digital world, no matter if you're in 16bit, or 24bit. No extra headroom, no extra volume to play with.

Here's a quote from a post, which I've described this very thing before.

"Here's how it works: Think of sampling audio by putting an analog waveform drawn on a piece of graph paper, and assigning it a point using an X/Y axis. Think of your X-axis divisions as the "Sampling Rate", and think of the Y-axis as the bit resolution. Let's say my actual audio is 1 second long and has a digital peak level of 0dB. If my sampling Rate is 44.1Khz, then my grids on my graph paper on the X-axis will have 44,100 points divided into that 1 second of time. If I increase my sampling rate to 48Khz, then I will have 48,000 divisions within that 1 second of time. So on my x-axis the resolution of accuracy increases, as I increase the sampling rate when plotting points that correspond to my original analog signal. Now for my Y-axis, as I mentioned this corresponds to the bit resolution, which actually corresponds to the amplitude of the original analog waveform. So if I have 16 bits, I have 2^16 (65,536) number of divisions representing the amplitude from -Inf. to 0dB. If I have 24 bits I now have 2^24 (16,777,216) number of divisions representing the amplitude from -Inf to 0dB. So basically the grid lines on my Y-axis has increased (ie become closer together)....thus the term "increased resolution" in assigning a numerical value. So the digital number that get's assigned to the sample is more accurate to the actual waveform.

The term signal to noise ratio, is actually mis-used in this scenario and misleading. A better term to use is "signal to error" ratio. Thus your signal to error ratio increases when you increase bit resolution. If your sound card has noise at the -50dB level, the noise will still be at -50dB, no matter if it's 24bit or 16 bit that you record at. The only difference is that the noise will be more accurately recorded on a 24bit recording than a 16bit recording, thus will have less distortion added to that noise, due to the larger space between "Quantization levels" (that's a fancy word for the spacing between the numbers on the Y-axis)."

Here's the actual link for the post on this discussion:

http://www.sonicfoundry.com/forums/ShowMessage.asp?ForumID=19&MessageID=101299

A simple answer for the general 16/24-bit discussion would be; you can easily be as much as 30 or even 40dB's too low with your levels on a 24-bit system and still have useable stuff to work with. Think high definition film, better to zoom in if needed, more fine details recorded. But then you need to start with 24-bits, and that's not happening on a DV camera of course. However, it's great for music recording, like a live concert when anything can happen and you want to stay safe without any surprise overloads.

Martin

I agree with you, but I think saying you can be 30-40 dB low is a bit of an exaggeration.

A run-off-the-mill 16 bit converter will probably give you useable resolution to approx. -90 dB. A 24 bit converter of the quality that most users here have will probably do about -105 dB of useable resolution (maybe a little more). My point is that if your 24 bit levels are 40 dB below 0 dBfs, your recording is going to sound worse than a 16 bit one with peaks at -5 dB. It is best to come as close to 0 dBfs as possible without *overs* but the 24 bit converter will be much more forgiving with lower levels as you stated.

English is not really my main language, sorry if I didn't make myself clear enough. I'm not saying that you should aim at peaks of -40dBfs, just that you can probably get away with it on a 24-bit system, never on a 16-bit. Tried to give a simple anwer to why 24-bits is there in the first place.

Your way of calculating useable resolution is correct (there are indeed very few true 24-bit converters in the real world...), so using that extra resolution and peaking at say -15dBfs would be pretty optimal (in case you need the extra headroom).

In the studio, I usually go as high as I dare, but then again it all depends on the type of source you're recording, and sometimes even how the customer feels about it. I've had producers getting really nervous ;-)

Martin

Another way to think of it is that each bit of resolution gives you 6db of dynamic range. 16bits (6db X 16) gives 96db of dynamic range. As mentioned most good converters max out at 105-108db dynamic range which divided by 6 gives 18bits. 18bits is the most you can use. The other 6 bits of the 24 get filled with noise. This is why some "24bit" converts actually do only 20 bits.

There are other arguments for more bits like quantization error which I'm not expert enough to address.

As a side note if at all possible use balanced connections. Noise floors around 100db are very difficult (if not impossible) get with real world (not test bench) unbalanced connections.

Larry