quick audio question

Hi Guys

I have a rush job in two days time to to record a live stage play. Vision wise I'm sorted but I have a question about the best audio format to use. I will be recording some audio on a Tascam DR22 placed as near the stage as possible, and as level as possible with the performers, and using a Rode NTG - 3 mounted on the camera, in this case a Canon XF 305. I'll select/balance the best audio when I edit in FCPX.

My question is: should I record on the Tascam at maximum quality, in this case, 96kbs at 24bits, to give me the best chance of handling volume variations/quality issues or will there be problems syncing dialogue with action? Common sense tells me it's a no brainer, use the best quality available, but the Tascam and dual sound approach is new to me so any suggestions will be more than welcome.

Sample rate will have to match for sync. 96kHz won't sync with a camera's 48kHz audio. You'll get audio drifting more and more out of sync.

Thanks for the reply, Ben. I'll stick to the standard which is 24bit 48kbs. I've found out that I can take a feed from their board into the Tascam so I should be fine. Thanks again.

That's sensible advice, 96k would confer no better quality - it would have to be converted to 48k at some point.

For info, the 'k' is not kbps, it's kHz, refers to the sample rate of the recording.

Thanks, ffg. Agreed, as it was to be boiled down to 48kHz (earlier typo but thanks for pointing it out) there would not be much point.

The gig went well. The feed into the Tascam was from two high quality boundary mikes either side of the stage. I left the T on auto gain and apart from some brief pumping during the quiet moments, easily sorted in editing, the audio quality was better than I expected.

I did an auto sync (with the audio from the NTG3 on the camera) in FCPX using audio as the guide track and it worked perfectly. I was expecting it to be quite slow in syncing but it was a matter of seconds for some takes as long as 30 minutes. Sync was spot on. Really impressive.

Thanks again guys.

Useful post, for I used to wonder the audio and video unsync problem, but for checking this post , it do shed some light on me. Thx guys !

It's not entirely true. 96k makes for better maths and processing done with that signal.

Like the difference between 16-bit and 32-bit image processing. Yes you can get away with working in 16-bit, but 32 does have its merits and will make your work more future proof.

Another example is if you wanted to manipulate your visual footage, than capturing it at a faster frame rate using a better codec like 422HQ or 444, give you more information to work with, so you can manipulate the footage, even slow it down a lot more before it begins to fall apart.

Yes you can work at a lower quality, but if you have plans on doing a bit of work with the audio, than a higher quality is best. The downside is you need a system that can keep up with it, but if it's just audio it shouldn't be too much of an issue.

I work with 96 as much as I can and there is a difference working at 96 then mastering at 48 just as there is a difference working in 32bit and mastering at 8 or 16 bit with video.

FCPX won't fall out of sync, I often end up with mixed sample rates, especially when i start to add effects to a project. it's stays in sync with the video just fine. Exporting as Quicktime can also preserve the quality if you wish.

FCPX is smart enough to handle multiple formats on the timeline, not just with video, but with audio as well.

Hi Simon. Really interesting post, thanks.

I must admit, I couldn't see how a '96" file would lose sync but I was too rushed to experiment. And I like your analogy with 422/444. The bit about future-proofing and the possibility of wanting to manipulate later, maybe with new software that might be developed, etc., makes a lot of sense. Another analogy would be re-wroking raw still files with the latest raw converters. There is often a significant improvement to be had.

In my case, with this particular video, it didn't matter. The quality I got from 48/24 was better than I had hoped in the short time I had. But I will certainly do some tests using 96, with the aim of keeping that as the master audio file and then working with it at whatever suits the project. Compared with 422HQ and the 444 files I sometimes (rarely, thank goodness) work with, the audio files are very modest in size so storing and working with them is no problem.

It's good to know FCPX will handle a mixture of audio rates in the same way that it does with visuals. I was massively impressed with the speed and accuracy of its auto syncing. Thanks again.

96kHz is used on studios to do heavy processing, but that is to give overhead. Problem with comparing video to audio is that 96kHz isn't audible to humans. Anything over 41kHz really isn't. If I play a raw recoding done in 48kHz, and the same recording done in 96kHz, humans can't hear any difference. But, it gives wiggle room when processing through a lot of digital filters, or so the theory goes. Consider most recording studios record the raw material at 44.1kHz.

Nothing wrong with 96kHz recording, but if you're not doing a whole lot of digital filtering, you won't hear any difference.

An yes, FCPX will end up with audio drift using different sample rates, I see it weekly. Audio is not handled the same way as video.

Fact of the matter is if anything over 48k wasn't audible, then recording systems would have stopped there.

The difference is subtle but it's there. We don't directly hear it's effect but we can tell it's interaction with the lower bands. Frequencies over 48 interject "air" into the recording, or atmosphere. Audio sounds less suffocated or stifled. Those extra harmonics will interact with other signals recorded over 48, which will interact with other processing. All small quantities, but accumatively adding up to something bigger.

True you lose some of that down converting, but some of it is still preserved because the interactions produce side bands well into the audible range.

Furthermore, when you approach 48k there is a steep lowpass filter that is applied, this is because, according to the Nyquist Theorom, audio frequencies will interact with the sample frequency ceiling and produce side bands that introduce intermodulation in the lower frequencies. This is nasty stuff. Unwanted side effects. So to avoid that the audio undergoes steep filtering at 48k, which seems to have the effect of suffocating the sound.

Recording at 96 will push that ceiling right up, well beyond the range of what is heard, and so the steep filter has no side effects in the audible range. No stifling. Plus that sense of space and air harmonics are actually preserved.

It's a case of a little bit of a lot adds up to something more significant.

Years down the track, 96 will become commonplace, and just as we can tell the difference with the quality of earlier style recording, more and more will be able to tell the difference with 48 and 96. Our eyes are already becoming more sensitive to color apparently, our ears will soon follow.

So while it's not necessary, it's definitely advantageous.

Oh and I have mixed audio on the timeline and it's all locked in.

Oh yeah. And when you come to conforming timing, stretching time will produce better quality results at 96k than 48. Just as it produces better results if you have 60 frames as opposed to 30.

It just has more information to work with and is able to produce better sounding results.