Choosing between 16 and 24 bit depth

[redge]

A question for people who deal with this question regularly...

If you have a choice between recording at 16 bit depth and 24 bit depth, what considerations do you take into account in choosing between the two?

I've read the ramps debates on this issue, and my sense is that they tend to be highly theoretical, which may explain why they are so inconclusive. I've tried to frame my question in a way that invites discussion based on real world practice; that is, what are people actually doing?

Thanks. 

[Jeff Wexler]

I have followed all the discussions as well and I don't think it is all that theoretical particularly if you go at the questions from all ends: the first thing being the original master recording (what we do in production) and the last thing being the completion of the project in post. Many of the discussions have their origins in one or both of these two poles. Firstly, not all production machines can record at 24-bits, so many of those who had machnies that could only do 16-bit were often the ones most concerned with this issue. On the other end, there were many systems in place in post sound editorial that were also not 24-bit capable. The other factor, concept or consideration, goes at looking at the final product and the age old discussion of "how good does it have to be to play the 5" speaker in the car or the mp3 player from Wal-Mart (and all the variations between depending on at what point in tech history you're talking about).

It is unquestioned at this point (as far as I know) that 24-bit audio is superior in all respects to 16-bit audio. Even if the project ultimately will only exist in what I call the "mp3 space" that does not mean everyone should wimp out and head towards that resolution. The Deva has always recorded natively at 24-bit and we were able to deliver at 24-bit or at 16-bit. In the very early days when the majority of the ProTools systems were sadly still 16-bit, I did deliver 16-bit recordings. Later, as I discovered what these systems would do to 24-bit recordings, I decided to deliver 24-bit anyway and hope everyone could deal with it.

I'll write some more later...  got to go now.

-  JW

[Philip Perkins]

As I see it the dicussion as far as movie sound goes is entirely post driven these days.  If you are recording for a project in which there is a lot of audio from video cameras, the project will be in 16 bit in the edit system and it's easier for everyone if you record 16 bit as well.  If you are doing a full double -system video or film project and the audio post people are good with 24 bit, then there are many advantages, such as higher headroom and greater resolution of detail.  The storage requirements are greater, but so far all that means is that DVD burns take a bit longer.

Philip Perkins

[Guest repete86]

I personally use 24-bit because it gives me a little more room before overloading.  In addition, even though all of the projects I've worked on so far have been shot on video, we've been recording the audio completely independently of the cam and have used the mic on top of the camera simply to aid the syncing process since we're just using  a standard slate. 

[Mike Borlace]

I would recommend recording in 24 bit as a way to "future proof" the production recordings.  At some point all picture and sound systems will be able to work at 24 bit and then 16 bit recordings will be a liability.  It's like the producers who shoot their productions in Hi-Def even though they may only be releasing them today in standard def.  These productions will have a longer "shelf life" as in the future they will have a value when standard def productions become the equalivent of black and white or silent productions.

[Philip Perkins]

I would recommend recording in 24 bit as a way to "future proof" the production recordings.  At some point all picture and sound systems will be able to work at 24 bit and then 16 bit recordings will be a liability.  It's like the producers who shoot their productions in Hi-Def even though they may only be releasing them today in standard def.  These productions will have a longer "shelf life" as in the future they will have a value when standard def productions become the equalivent of black and white or silent productions.

This is true but most of the projects that a lot of us work on have no shelf life at all (commercials, corporate) so the audio bit depth doesn't matter so much.  The major issue is if the production has decided to digitise @ 16 bit from video cameras, and intends for the audio post people to work from OMFs exported from the AVID or FCP without reconforming from original sources.  A big slice of video and TV productions work this way today @ 16 bit.  I don't think this will change until the cameras start recording 24 bit audio.  Of course, in some areas we are going on the OPPOSITE direction--re: HDV audio etc..

Philip Perkins

[David Terry]

I've tried to frame my question in a way that invites discussion based on real world practice; that is, what are people actually doing?

Thanks. 

R,

In my world of commercials and industrials, workflow, when I can wring that info out of the producers typically guides me to 16 bits.  On commercials shot on film the digibeta from telecine is the 16 bit audio that continues forward and the DVD-RAM from Deva is no longer used in most of my clients' workflows.

For a Metacorder setup next Wednesday, I asked the editors (Avid and Pro Tools) which bit rate among other things, do they prefer.  Even though the Metacorder section of this shoot is a small concert, singer-guitarist, and RF lavs on our heroes in the audience etcetera...and you think that they would want the music to sound as good as possible, my bet is that post will ask for 16bit files since the other scenes will be ENG style digibeta.

Of course if I were making my own documentary, I would go 24bits.

DT

[takev]

Hi,

So what is the feeling for 32 bit float, which seem to be the native internal processing format of a couple of audio programs (some are doing 64 bit float), including Boom Recorder. If these programs are internally using floats isn't it better to keep the audio files in the same format?

As floats are an exponential/logarithmic format is has more dynamic range than integers and are more accurate in filter processing (it certainly is easier to program these filters). An other advantage is that floats don't clip when processing samples through effect filters that amplifies the signal, most filters will handle sample values beyond -1.0 and 1.0.

I've seen the same thing happening with high resolution image processing for film frames, that now process and store pixels in floating point.

Cheers,

    Take