Audio sample rates on DV tape

farss wrote on 2/28/2005, 6:01 PM
Gotta admit I'm mighty confused here. I've always thought DV25 had audio as either 16/48K or 12/32K, right?
OK, can someone PLEASE explain to me why with all my captured video Vegas reports the audio stream as 32K sampling 16 Bit uncompressed?
All the media ripped from CD is reported as 16/44.1 so that's fine.
Anything rendered out of Vegas is 16/48 so that's right too.
When I was capturing a tape last night deck shows 48K, I opened a copy of the audio track in SF and was looking in stats and noticed the audio was 32K, thinking OK, SF is setup wrong I forced it to 48K and got duck talk so the sampling rate clearly is 32K on the tape and it's 16bit?
This is from a variety of cameras so they're not to blame, mostly captured using DSR-11 but I seriously doubt it's fiddling with the sample rate.



jaegersing wrote on 2/28/2005, 6:36 PM
Hi Bob. I've come across this before, but have never seen the "official" explanation. My assumption is that the audio is being saved on the hard disk as 16 bit, even though each sample is effectively only 12 bits. It makes sense in a way, because saving and working with real 12-bit words would be quite fiddly. So the sample rate is not touched, they just add some padding bits to the data words (I think).

Richard Hunter
farss wrote on 2/28/2005, 6:59 PM
Something REALLY ODD going on here.
Just as a test I captures 10 secs of 'air' through the DSR-11 from the analogue inputs.
With deck in 32K audio mode Vegas reports audio stream as 16/32
With deck in 48K audio mode Vegas reports audio stream as 16/48

Are you confused?
I sure as heck am. So it'd almost look like Vegas reports it correctly. Except in 32K it should say 12/32. So OK, Vegas is perhaps almost reporting this correctly. If that's the case how come the camera tapes are showing up as 16/32K.
I'll run some more tests, this gets weider by the minute.
You see what has me really thrown is when I capture 90% of tapes the deck says 48K . Yet it seems when they end up on the T/L Vegas sees them as 32K. The other 10% I know someone has their camera set wrong and the deck says 32K as does Vegas.

nickle wrote on 2/28/2005, 7:23 PM
This is from my manual

VCR set
HiFi sound Stereo - to play back the stereo tape or dual sound track tape with main and sub sound.

12 bit (2 stereo sounds)
16 bit (1 stereo sound in high quality)

Audio mode

12 bit mode - the original sound can be recorded in stereo 1 and the new sound in stereo 2 in 32 kHz.

16 bit mode - A new sound cannot be recorded but the original sound can be recorded in high quality. Moreover, it can also play back sound recorded in 32kHz, 44.1 kHz or 48 kHz.

I have it set at 16 bit stereo and Vegas says it is 48kHz.

It's clear as mud to me but maybe you can find something in it.
DigVid wrote on 3/1/2005, 5:22 AM
Just a suggestion... Does any of this captured DV footage have any origination from your Sony Z1 HDV camcorder? Because, I read somewhere that there is a bug in either the FX or Z1 that causes a problem with audio sampling confusion in the DV mode. Supposedly a fix is on the way. If I can locate the link to this I will post it.
jaegersing wrote on 3/1/2005, 5:13 PM
It happens (in my case) with any DV audio. The DV standard is for 32KHz audio to be sampled with 12 bits resolution, but what I think is happening is that when this is saved on the hard disk (either as AVI or WAV file), an addiitonal 4 bits of padding is added so that the data is stored in 16 bit words. This would be much easier for most applications to work with. Of course, you don't get any additional resolution, the "extra" 4 bits are probbaly all zeros.

farss wrote on 3/1/2005, 5:18 PM
Thanks and yes there is a bug and we've had one of our cameras upgraded.
But no this wasn't FX1 footage!
As I've said I've tried footage from a variety of cameras with the same odd result.
Grazie wrote on 3/1/2005, 11:15 PM
Bob - I've emailed you. - G
jaegersing wrote on 3/2/2005, 2:25 AM
Found this discussion at the link below. It is not data padding after all. Seems there is a complication due to the 12bit audio in 32KHz DVbeing non-linear.


"audio capture problem - Adam Wilt

> ...It is of course two stereo tracks at 12bit or
> one at 16bits. But the VX1000 only uses 12 bit and doesn't record onto the
> second stereo track.
> The fact that MS6.0 only refers to 16 bit suggests that Sony's incarnation
> of DV is no standard but this still doesn't explain why some shots are OK
> and others are not- they all look the same on the DHR1000 ie 12 bit
> stereo.
Sony's implementation is absolutely standard. You can look it up in the
Blue book, or IEC 61834 (!
However, most computer-based editors don't know what to do with 12-bit
sound, so they'll convert it to 16-bit linear on input and reconvert it to
12-bit nonlinear on output. The good ones at least don't perform
rate-conversion unless you ask them too; they'll report the audio as 32kHz,
If MS6.0 refers to VX1000 audio as 48kHz or 44.1kHz, then it's performing
a rate conversion. You might be able to change this in MS's setup menus (I'
m not familiar with the program). Doing so will give you somewhat better
sound -- how much better depends on the quality of MS's rate-conversion
code, of course!
Adam Wilt"

Bob Greaves wrote on 3/2/2005, 4:03 AM
Im not in front of my camera but I seem to remember that my samsung camera permits me to choose bit depth independant of sample rate.

Bit depths are 12 and 16, sample rates are 32 and 48. So depending on how I choose I can get all the following audio configurations:

12 bit 32 KHz
12 bit 48 KHz (not standard but valid)
16 bit 32 KHz (not standard but valid)
16 bit 48 KHz
farss wrote on 3/2/2005, 4:22 AM
Got it, thanks.
farss wrote on 3/2/2005, 4:35 AM
now there's a range of choices! Now what's really odd is Sony cameras like the PD150, when you select 48K say "48K NS" in the viewfinder and the default setting IS 32K!
Now I've got no way of really testing just what the bit depth is that I know of (Red feel free to tell me how). However whichever way I try both Vegas and SF insist that the sample rate IS 32K, despite the camera saying 48K!

Now this really worries me because there's a big difference between 32K and 48K in terms of quality, well there's an even bigger difference between 12 and 16 bit. What's even more rattling is that when we play the camera tapes back, the damn DSR-11 deck says 48K! Yish time to really get to the bottom of this and run some more tests, once I can my hands on a camera!

From what Adam Wilt is saying this could be a MS6.0 bug but surely Vegas works things out for itself. If it started treating 48K audio as 32K there's be very noticeable problems I think, treating 12 bit as 16 bit shouldn't matter at all though.