32 bit, 24 bit, 16 bit, Who is lying?

[jmiller]

Hey folks.

So I recently received some files from a tracking session for me to do some overdubs and mixes on. The tracking engineer apparently used Adobe Audition to record the tracks. I am using Pro Tools.

According to the workspace in PT, the files I received are 16bit 44k. Naturally, this is not ideal (made even less ideal by the very low recording levels). I downloaded an evaluation demo of AA to open the original "sessions" to see what was going on. When viewing the files in AA, it tells me they are "0.24 type 3", or "32 bit mono".

I do understand the whole 24 bit/32bit float BS. But what I'm trying to figure out is if the files were in fact recorded at 24 bit (and stored as 32bit for internal processing in AA) OR at 16 bit.

I tried exporting the audio files from AA as regular "24 bit type 1". Then I imported those files AND the "16 bit" files into PT. They null. But what IS it?

Anyone dealt with this?

On a side note, I'm bringing this question here first, rather than the Audition forums, because I don't want to get into a debate with any AA evangelists over why I should ditch PT and switch to AA.

[vvv]

I use CEP2.1.

If ya want, put a track on yousendit or somewhere and link it up; I'll download and letcha know.

[The Scum]

You should be able to tell simply by the file length.

16 bit 44.1k files will be 88200 bytes per second (~ 5 MB a minute).
24 bit 44.1k files will be 132300 bytes per second(~7.5).
32 bit 44.1k files will be 176400 bytes per second(~10).

Each +/- a bit of overhead data. Stereo tracks will be double that.

(There may be a bit of confusion, though, because I seem to recall that some apps stored 24 bits in 32 bit words, simply because it's easier to waste the storage space than to have to do a lot of modulo 3 arithmetic to try to get a stream of 24-bit values aligned)

Or try this handy little app to peek inside the files:
http://www.railjonrogut.com/HeaderInvestigator.htm

[Corey Y]

Are you sure about that? I'm looking at a 16 bit/44k.1 kHz wav file right now, 7.316666 minutes, 73.8 MB file size. So roughly 10 MB/min.

It was recorded at 24 bit and bounced down to a 16 bit stereo wav, in Logic 9 in OS X Snow Leopard.

Not being smarmy/sarcastic, I know very little about sample rates and all that. Just enough to not screw stuff up.

[rushofblood]

The Scum probably meant per mono track?

[rushofblood]

Stereo tracks will be double that.

[The Scum]

Let's do the math:

16-bits means 2 bytes per sample.

Times 44,100 samples per second = 88,200 bytes per second.

Times 60 seconds per minute = 5,292,000

In reference to computer memory, a 5 megabytes is actually 5,242,880 bytes (1024*1024*5).

Those of us who lived through the early days of sampling knew these numbers well...it dictated how many floppies we'd have to shuffle to load up a sample set.

[Corey Y]

Ah...my eyes glazed over a few figures in, but I'll take your word for it

[jmiller]

Thanks, guys.

That util from Rail is pretty handy. In my case, however, I couldn't use it to "change" anything in the files (it just rendered them silent).

I'm pretty sure that the files are "32 bit" whatever. If I open them in Audition or Samplitude it says so. Oddly, the Pro Tools workspace and the "import audio" dialog all declare them as 16 bit files. Rail's header utility says "undefined" (though to be fair, the utility was made for Pro Tools files and Pro Tools doesn't do all that 'save a 24 bit audio recording as 32 bit' rigamarole).

I have to wonder what's being done, then, when PT imports and thus copies/converts the audio to proper 24 bit. But I'm almost too exhausted from thinking about it to care anymore. They don't seem to actually be 16 bit but i dunno. In any case I have other issues to deal with anyway.

Scum, I remember the days of 10-floppy sample sets on the Emax II. I remember getting a SCSI ZIP drive and feeling like I was living in fucking Total Recall or Robocop or something. And I've still got my Mirage, though it's getting hard to find 800k floppies

[vvv]

There is a adjustable parameter in some DAW's, like Cool Edit and presumably Audition, what lets you make conversion to 32 bit on file-opening a default.

So, ya might could wanna check on that.

[leigh]

I do understand the whole 24 bit/32bit float BS. But what I'm trying to figure out is if the files were in fact recorded at 24 bit (and stored as 32bit for internal processing in AA) OR at 16 bit.

I tried exporting the audio files from AA as regular "24 bit type 1". Then I imported those files AND the "16 bit" files into PT. They null. But what IS it?

I have to wonder what's being done, then, when PT imports and thus copies/converts the audio to proper 24 bit. But I'm almost too exhausted from thinking about it to care anymore. They don't seem to actually be 16 bit but i dunno. In any case I have other issues to deal with anyway.

See if you can track down a bit scope or bit range meter plug-in. These are included as part of Spectrafoo Complete ($$$), but there may be some freeware versions out there somewhere of these.

A bit scope or meter will tell you more about the information in those files than header info, or calculating expected file sizes.

When Pro Tools copies in 16 bit files to a 24 bit session, it pads the last 8 bits with zeros. So, your files sizes go up by 50%, but you obviously don't gain any new audio information from that process. This is why your null test worked (comparing 16 bit and "24 bit" files). And, if I understand everything you're saying, then those files in AA were recorded through 16-bit AD conversion, whatever file format they subsequently wound up in.

That make sense?

cheers,
Leigh

[Bill @ Irie Lab]

Different programs have their own way of dealing audio internally but should spit out a standard .wav file at the quality you specify.

Maybe you've tried this -

In Windows right click on the file, choose Properties then Summary.

It will display # of channels, bit depth, and sample rate.

(be sure Advanced is selected or it will be all grayed out)

[Bill @ Irie Lab]

PS


Thanks for the tip about Header Investigator.

Sound Devices's Wave Agent is a utility along those same lines, with playback and mix functions.

[leigh]

In Windows right click on the file, choose Properties then Summary.

It will display # of channels, bit depth, and sample rate.

Again, though, this is just reading header info.

What the OP was looking for was:

...But what I'm trying to figure out is if the files were in fact recorded at 24 bit (and stored as 32bit for internal processing in AA) OR at 16 bit.

So he needs to examine the content of those files to figure this out. As you say, programs will spit out WAV files at whatever you tell them to. But if you've recorded at 16 bit, and tell your program to spit out 24 bit files, you don't get 24 bits of information. You get 16 bits of information with 8 bits of padding added.

Cool?

Leigh

[Bill @ Irie Lab]

Thanks Leigh,

Headlines aren't truth - just see Fox News if anyone needs proof.

Now we face the truncation problem!

Cheers,

Bill