Sound Level on EX1r

[Denny Kyser]

I asked this in the audio section with no response, since it is a camera specific hoping for some help here.

I am really working on getting great audio, and am unsure about the level that is ideal in the EX1r.
I like the sound at -10 but not sure if that is giving me enough headroom.

I turned on the 1k test tone, and that puts the level to -20, seems that auto level also shoots for -20 so I am thinking this is where the ideal audio level should be.

I am setting up my SD 302 mixer and trying to determine where the audio level in the camera should be when the 1k tone on the meter is on.

[Vincent Oliver]

Generally I tend to keep my sound level to -12 or even -18, that should give you sufficient headroom.

When producing DVDs I set my maximum sound level to -3.

Hope this helps

[Bob Grant]

-18 is way too low. Keep in mind with digital audio the lower the recorded level the higher the distortion.

The EXs have a limiter and it seems impossible to get them to clip. Not that you want to have every peak being whacked by the limiter but the odd impulse sound from a cough or something being dropped should be OK.

[Vincent Oliver]

using -18 would depend on how close you are to your subject, I know of one professional sound guy who has almost everything set to -20. Personally I tend to use -12 for most of my shooting. This is on my portable Sony PMW50 recording device, I only ever use the cameras sound as a guide track or with radio mikes.

[Denny Kyser]

Thanks for the information, will do some more testing but looks like between -20 and -10 seems to be the sweet spot.

[Dave Morrison]

Originally Posted by Bob Grant

<snip> Keep in mind with digital audio the lower the recorded level the higher the distortion. <snip>

Could you explain this one Bob? That left me scratching my head.

[Bob Grant]

Originally Posted by Dave Morrison

Could you explain this one Bob? That left me scratching my head.

Sure. The lower the signal the less bits are used to define it.
Take a sine wave at 0dBFS. There's 16 bits used to represent it. Reduce the signal by 6dB and then there's only 15bits to define it. Reduce the signal to the lowest level and you end up with only 1 bit to define the sinewave and the binary values will represent a square wave. Of course real world recordings are not sine waves and not all the components of the sound are at the same level, the quieter parts are always more distorted than the loudest, unless you clip.

Digital audio has zero overhead unlike analog audio tape. You could push the signal past 0VU and the system would saturate quite pleasantly. With digital audio the signal just clips above 0dBFS and sounds horrid. With analog systems distortion tends to decrease with level. With digital systems the opposite is true. The best quality digital audio recording is achieved by just not clipping. I've been working with some audio recordings where things really got out of hand and I'm having to apply 40dB gain, it does not sound nice at all.

[Dave Morrison]

I guess it was your use of the term "distorted" that threw me off. I'd agree with your description that fewer bits are used at these low levels but I'd take issue with the use of distorted. If that were the case, then every digital recording of a quiet orchestral passage would be an unlistenable sea of distortion, no? These low-level audio signals may be lacking in resolution, but inherently distorted?

[Olof Ekbergh]

For audio levels in an interview, where conditions are controlled. I use -12 or so, -18 if I don't know the person well.

If it is outdoors or the interview may get lively or someone may laugh loud, I use -20.

For BG sound I set it to -30 or so, just in case of something loud.

[Bob Grant]

Originally Posted by Dave Morrison

I guess it was your use of the term "distorted" that threw me off. I'd agree with your description that fewer bits are used at these low levels but I'd take issue with the use of distorted. If that were the case, then every digital recording of a quiet orchestral passage would be an unlistenable sea of distortion, no? These low-level audio signals may be lacking in resolution, but inherently distorted?

Total Harmonic Distortion is a measurable quantity and yes the quieter passages will be more distorted than the louder in an orchestral recording. They'll not be an "unlistenable sea of distortion" because although they are more distorted the amount of distortion is well within acceptable limits for the average listener. Those with "golden ears" would beg to differ of course and proceed to write 10,000 words :)

Keep in mind though that any serious recordings of orchestral works will be made at at least 24bits and then mastered to 16 bits if going onto audio CD.

The original question related to "headroom". The EX cameras use peak level meters. The levels they display show how much headroom you have. Recording at -12dB or -18dB peak level means you have wasted that many dB of headroom. On the other hand if you eyeball an average of what the meters are telling you -12dB is probably just safe for speech, the peak reading of the meters will be getting very close to 0dB and possibly you are just hitting the inbuilt limiter. Recording at an average reading of -18dB is certainly not too bad for speech.

The issue I find is one persons average becomes anothers peak and then someone thinks a bit more headroom cannot hurt. You still are unlikely to end up with unlistenable distortion however if you apply enough gain in post to get it sitting right in a mix and then intercut to some ADR recorded in a good studio the difference is quite distracting.
There's probably millions of words written about how to get the best images out of the EX cameras. A few about how to get the best audio would probably not go astray. No one would say "oh, just set the iris to F8 and you'll be close enough". I do understand the difference between video's 8bpc and audio's 16bpc but still.

[Chad Johnson]

If you are calibrating the EX camera using the 1k tone from an SD mixer, then set that tone level to -20db on the camera. Then no matter how loud the signal is through the mixer you will not clip. Also make sure you do not turn the EX level dials above 6, or you will be adding noise. You can leave the external audio dials to 5 or 6, then set the trim level through the menu to hit the level you need.

[Denny Kyser]

Originally Posted by Chad Johnson

If you are calibrating the EX camera using the 1k tone from an SD mixer, then set that tone level to -20db on the camera. Then no matter how loud the signal is through the mixer you will not clip. Also make sure you do not turn the EX level dials above 6, or you will be adding noise. You can leave the external audio dials to 5 or 6, then set the trim level through the menu to hit the level you need.

Thanks, that is VERY helpful, I had been turning them up above that to get to -20.

[Dave Morrison]

Originally Posted by Chad Johnson

If you are calibrating the EX camera using the 1k tone from an SD mixer, then set that tone level to -20db on the camera. Then no matter how loud the signal is through the mixer you will not clip. Also make sure you do not turn the EX level dials above 6, or you will be adding noise. You can leave the external audio dials to 5 or 6, then set the trim level through the menu to hit the level you need.

Chad, should that audio signal coming out of the SD mixer be taken in on the EX1 using Line Level or Mic Level setting?

[Denny Kyser]

Originally Posted by Dave Morrison

Chad, should that audio signal coming out of the SD mixer be taken in on the EX1 using Line Level or Mic Level setting?

I have been using Line level, but will see if that is what Chad suggests.

[Dave Morrison]

That was what I assumed, Denny, but I wasn't sure. I used to have an MP2 from Sound Devices but would love to get a 302.