48 KHZ or 96 KHZ? to record for film

HerzogisGod

Well-known member
I own the Fostex feild recorder and Im using it to capture sound to be edited into my films. Sould I record 24bit 96khz, or 24bit 48khz? Is there a huge difference.

I intend on offloading all my data from the unit straight into my computer for further modification. Is there a specific program that's affordable that will allow multi track layering and enhancements?
 
I own the Fostex feild recorder and Im using it to capture sound to be edited into my films. Sould I record 24bit 96khz, or 24bit 48khz? Is there a huge difference.

I intend on offloading all my data from the unit straight into my computer for further modification. Is there a specific program that's affordable that will allow multi track layering and enhancements?

If you were doing a lot of mixinmg in the digital domain then there might be a difference between 48k and 96k but in most cases probably not.

Multi-track editing, etc ... Vegas, Audition, Nuendo and bunches more. Define "affordable" as related to your wallet.
 
I could record 16 Bit, but isnt 24B better?


I guess around $300 for some sound editing software. There must be something for that price :)
 
48kHz, 24bit has become the standard for non-linear recording on set. 96 would be overkill for just dialog. Though could be useful for capturing sound effects that may later be manipulated in a DAW
 
Regular DVDs as well as HD-DVD and Blu-Ray all have natively 24bit 96Khz audio (as do the all but market-irrelevant SACDs and DVD-A music discs).

In a perfect world, everything would be recorded as such from the git-go, but alas, ours is not a perfect world.

With high quality AD/DA converters (like Digidesign 192s), the sonic difference between 48k and 96k is very subtle, but the difference between 24 bit and 16 bit is quite meaningful. So I'd say, as HansosnD suggests, that whenever possible you should record 24 bit. 96k probably isn't necessary, and using it eats twice as much of everything - hard drive space, DSP plugin resources, backup time, etc. And anything can be upconverted to 96k later for the DVD formats. You won't gain any quality by doing so, but for the most part, that shouldn't really matter.

With marginal quality converters, 96k is much more significant, but all the dowsides still apply of course.

If you want to be a little bit more of a purist, record any music scoring at 24/96. For myself, knowing the pros and cons, and having good converters in my rig, I'm sticking to 24/48 at the capture stage for everything (when possible, which for example, isn't in the case of recording directly into my HVX-200) unless otherwise compelled.

My $.02
 
Regular DVDs as well as HD-DVD and Blu-Ray all have natively 24bit 96Khz audio (as do the all but market-irrelevant SACDs and DVD-A music discs).

Excuse me? This is simply NOT true. Standard audio on a commercial DVD is AC3 encoded 44.1/16. So not only is it not 96 it is basically an MP3 file. Theatrical Dolby Digital is also 44.1/16 and encoded. You can put 96k audio on a DVD but that is the "market-irrelevant" DVD-A.

In a perfect world, everything would be recorded as such from the git-go, but alas, ours is not a perfect world.

Somewhat arguable. I think probably the 1bid stream is a more "ideal world" format but...

With high quality AD/DA converters (like Digidesign 192s), the sonic difference between 48k and 96k is very subtle, but the difference between 24 bit and 16 bit is quite meaningful. So I'd say, as HansosnD suggests, that whenever possible you should record 24 bit.

True, even with better and a bit worse converters.

96k probably isn't necessary, and using it eats twice as much of everything - hard drive space, DSP plugin resources, backup time, etc.

Very true.

And anything can be upconverted to 96k later for the DVD formats. You won't gain any quality by doing so, but for the most part, that shouldn't really matter.

True, strange and pointless. You can up convert anything, you wouldn't do that for DVD's and it's pointless because you don't gain anything up converting. You can't get back what you didn't record in the first place.

With marginal quality converters, 96k is much more significant, but all the dowsides still apply of course.

What? That is a very strange comment and I can't think of why you would think that this would be so. In fact everything I have read would point to the oposite being true. A dodgy converter is going to be most dodgy when it's at the edge of what it's supposed to be able to do. So a dodgy 96k converter is probably a pretty decent 48k converter, making the difference between the settings closer not farther away.

If you want to be a little bit more of a purist, record any music scoring at 24/96. For myself, knowing the pros and cons, and having good converters in my rig, I'm sticking to 24/48 at the capture stage for everything (when possible, which for example, isn't in the case of recording directly into my HVX-200) unless otherwise compelled.

My $.02

In general I agree with this. However the converters in your editing rig don't really come into it much in this case. Your not doing any A/D conversions in the edit so those high quality converters are being used strictly for playback. As to the original post the question would be whether the converters in his recorder are good 96k converters or not. My guess is that they are in the middle somewhere.

We did a bunch of tests with a Nagra V (which has some pretty good converters and an excellent preamp). 48 VS 96 was just noticeable on BRIGHT sounds (slamming metal plates together etc) and pretty undetectable on everything else. Live acoustic music would be one place I think it would probably be noticeable with. 16 bit VS 24 bit was instantly noticeable and we could tell the difference blind. For the heck of it we also did the 44.1 VS 48 test (this was the raging debate in DAT days) and just as when we did the test with DAT nobody could tell the difference.
 
I would suggest going 24bit, no matter what sample rate you use.

That way, you can run your levels relatively low and have lots of
headroom, in case there's an unexpected loud sound. Or if you
unintentionally record something at an extremely low level,
you can boost it without getting a lot of yucky quantization noise.

Keep it 24bit through post, if possible. At the end, you can maximize
the gain, then convert it to 16bit, if you'd like.

Les
 
If the camera have 48kHz/16bit, edit and record sound at the same :thumbsup:
 
To maybe put the two previous posts in perspective, since both are valid to a certain extent.

If your doing the sound in your NLE and it can't deal with mixed sample rates/bit depths and or it can't convert your on camera sound to 24 bit then phd is right. If your using any of the current crop of pro to prosumer NLE's or your doing a separate audio post then go with 24 bit.

It used to be a big pain with mixed bit depts or sample rates but this isn't the case any more and it was also common to go back out to tape so you would be stuck at the DV tape 48/16. Now most people are going to DVD and files and conversion is fast and done all the time so it is usually not an issue.
 
Last edited:
If at all possible, record everything at 24 bit instead of 16 bit. Each bit represents 6dB of dynamic range. Going 24 bit allows you to not worry as much about running your levels right up as close as possible to 0dB.

48kHz vs 96kHz is much more subtle.
 
You don't want to go "as close to zero as possible" because of converter overshoot. So you should try to make sure not to peak above -3dB. You also don't want to record really low and 24 bit's isn't going to really help you there. The logic is that you can trow away some bits and record low and still get the same resolution that you would have had before. Logically that is fine but recording low drops you farther into the noise floor and so you have higher resolution, noisier files.
What you want to do is basically record where you should (traditionally) and the 24 bits will give you better sounding files that are smoother to message in post.
 
However, many mic preamps, especially the kind that aren't super expensive, can sometimes bring up hiss at a greater rate than they add gain. In other words, as you turn up the gain knob, the rate of hiss climbs faster than the rate of signal. By using 24 bit audio, you can back off on the gain just a little bit, give yourself a little more headroom to protect against peaks, and bring levels up a little more in post, getting a cleaner overall result.

Of course, if you overdo it too far in either direction, you're asking for trouble.
 
Back
Top