What format do you record at?

[Jonathan Levin]

Hi all.

I'm am wondering what you pros chose for audio format when out in the field.

My recorder does PCM 16bit 48kHz wav files, it's highest resolution. The problem I see is that six minutes of recording uses 4GB of card space for stereo, and just under 12 minutes for mono. So lots of giant cards on hand. The recorder has an option to record 44.1kHz, but IIRC, editing software likes 48.

24 bit recording has got to be a huge file producer, and I've also heard/read that it is almost preferred to do 16bit since very few people can tell the difference.

A friend who is a sound guy records MP3 on his Sound Devices, to be used only in case audio feed and capture to camera is messed up.

So what's your opinion on this? For me, the recorded audio would be my main audio, replacing scratch track from camera after sync, so quality is important.

Thanks guys.

Jonathan

[Rick Reineke]

For most audio pros, 48k/24 is SOP. 16 bit sounds good if levels are not anemic. If there's only one audio source, recording single-track mono can be more economical with memory.
One must keep in mind, some NLE's / DAW's do not support sample track events/clips of different rates.
A high res MP3 (320kbps) usually sounds good and would be fine for backup. But why.. since memory is so cheap these days.

[Andrew Smith]

I've only ever bothered to record on my Sony PMW-300 at the 50Mbit XDCAM format setting. This flavour of XDCAM records the audio uncompressed at 48k / 24bit.

Works very well for me.

Andrew

[Gordon Hoffman]

Shoiuldn't a 4GB card give you around 4 hours of stereo 48KHz 24 bit audio.

Gordon

[Greg Miller]

Your recorder uses how much memory? I assume you have never actually tried this, or you wouldn't have posted those numbers.


In reality, for stereo ::

16 bits/channel * 2 channels/sample * 48,000 samples/second * (1 Byte / 8 bits) * (1 GByte / 1,000,000,000 Bytes) * ( 60 seconds / minute) = 0.01152 GB / minute.

Therefore 6 minutes * (0.1152 GB / minute) = 0.06912 GB = 69.12 MB memory (for 6 minutes).

Inversely, 0.01152 GB / minute is equivalent to 86.806 minutes / GB.

Therefore, 4 GB * (86.806 minutes / GB) = 347.22 minutes (recording time for 4 GB).

In reality, for mono ::

half as many GB / minute
half as much memory for 6 minutes
twice as many minutes / GB
twice as much recording time for 4 GB


Not that it matters, but out of curiosity what recorder do you have?

[Rakesh Malik]

Greg beat me to it, but I was going to comment about the space requirements also. Over the summer when we were recording audio for a short, we were able to capture 2 hours of audio on a 16 GB card, and that was recording 192KHz, 24-bit audio... six channels at a time. Compared to any professional video capture format like ProRes HQ, the space requirements for capturing high resolution audio are trivial.

Premiere Pro and LightWorks and FCPX all support 192KHz audio without any problems. In Premiere Pro (latest CC version) I can load multichannel audio clips and sync them to video clips by waveform. The only NLE that I know of that's still rooted in the past as far as audio is concerned is Resolve, a constant source of frustration for my team because that's the only thing preventing us from making Resolve the centerpiece of our entire workflow outside of compositing for VFX.

Why do we record in 192KHz? Well, for the same reason that people like using 4K capture even when the target is HD: downsampling helps to eliminate noise.

We've validated this on a few occasions. In one case, I had recorded 4 channels of 192KHz, 24-bit audio to go along with a video clip. The room we were filming in has fans, but they're usually not particularly loud. This evening we were filming there on a warmer day than usual, so the fans were running hard, and a LOT louder than usual. So... there was a lot of background noise in the recording.

Audition insisted in downsampling the audio clips to 48KHz when I imported them. I suspect that there's a way to make it stop doing this, but rather than fight with it, I just moved over to Reaper. It also helped that Reaper's handling of multichannel audio is more user friendly than Audition's.

What I discovered is that when I used the noise reduction filter at 48KHz, it ruined the dialog. When I ran it on the 192KHz samples and THEN downsampled it, the result was surprisingly clean.

So, is it worth it to record at higher sample rates?

Yes, it is. It doesn't take up much space, and storage is cheap; why skimp on audio if you care about your video? The sound designers I work with prefer higher sample rates for the same reason I described above. It's cleaner. There's no significant additional cost that you wouldn't already be incurring by giving your audio files to your sound design teams anyway.

[Garrett Low]

If I'm shooting a symphony concert I record 192/24bit. For dialog, interviews, and almost everything else it's 48/24bit

As has been already addressed you will get more than enough record time on a 4gb card unless you plan on recording a marathon session. Also, there is no need for a super fast card so it would be relatively cheap to pick up a 16GB or 32GB card if you need really long record times.

[Rick Reineke]

"So, is it worth it to record at higher sample rates?"
- Not worth the storage space in my opinion. It's a pain changing cards as it is on a fast paced set.
Music and S/FX recording are another story for sure.

[Rakesh Malik]

Given that even with six channels, you get close to two hours of record time on a 16 GB card and it takes about a minute to swap cards, I don't agree. Neither do the people operating our mixer. We have to swap out camera cards a LOT more often.

[Rick Reineke]

I'm talk'in dialog recording.. Most all of my pro film production sound colleagues (major league mega-million movies) record dialog at 48/24.
Music and S/FX .. higher rates for sure.
BTW, welcome..

[Jonathan Levin]

Hi fellows.

And thank you for your input!

Sorry it took me a while to get back. I am using a Marantz PMD 660. I've attached the page of the manual where it gives recording times. I've not run into a problem of running out of disk space, but as you can see, or maybe it's just me, with the 4GB card @48kHz it LOOKS like 5 minutes and 55 seconds. I'm now thinking 5 HOURS and 55 Minutes??? The page wasn't clear enough for a novice such as I am.

Can someone clarify this? If it is actually about 6 HOURS then there would be no reason to record anything other than the highest quality.

Jonathan

[Rick Reineke]

deleted. -- double post

[Rick Reineke]

Yes, that's hours / minutes.
'With a 4GB card @ 48k/16bit: 5 hours. 55 minutes (stereo), 11hr.s 51 min.s (mono)
@ 44.1k: 6:27 (stereo); 12:54 (mono)
The mono mode would be recommended for spoken word and other single source recoding... (in the mono mode, for line level, the tip terminal of a 3.5mm TRS or mono plug.. for mic level, the left XLR input.
48kHz. is standard for audio for video, 44.1k for audio only.
As I recall, the 660 does not have a 24 bit mode.

[Jonathan Levin]

Thank you Rick!!!!

Something was off there and thanks for clearing that up. And yes, the 660 only does up to 16 bit recording.

I've used this mostly hooked up to my Shure FP 33 for two to three person interview type stuff, so I'm thinking with the stereo it would be one more level of control in post if I mess something up. I can always convert to a single mono track.

Jonathan

[Rakesh Malik]

If you can configure the recorder for dual mono, then you can either use one for each speaker or one as a backup (with higher or lower gain than the other, for example). In stereo, unless you're using two separate mics, you'll get the same result on both tracks since in stereo both channels would share gain settings.