Audio and the AF100

Not much I can add to Jeff's comments here and SteveG on Adobe - they really covered it all. Of all posts on this forum, I've learned more from Jeff and Barry than any other posters. Congrats to you for not being afraid to ask for help and to Jeff for stepping in with assistance.
 
As a cameraman...I really needed help with AUDIO. I have learnt so much in the last 10 days about it since asking around.

Difference between audio and "video" is how dynamic audio is. It never stays in the one place or stays still at all.

The hiss issue. I do not know where in the audio chain to camera that I induced it. I do not believe it is anything to do with the camera.

Something I did. ( at least I can rectify it).
 
I have been shooting interviews with Sony ECM77s going straight into the AF101 using the 48v power option and have not detected a problem. The audio monitor amp is noisy but then it is on most cameras including expensive Digi/SX/HDCAM varieties. In the cutting room all sounds fine.
 
I also have done a lot recording directly to the camera without any issue.

This time ...I used a field mixer and audio came in Line Level.
My un familiarity with that audio chain is what has caused my little issue. I can hear the hiss in my edit suite so I must get rid of it.

I will be doing more tests ( and learning) with it all later once I get this urgent job out th door.
 
Shooter said:
This time ...I used a field mixer and audio came in Line Level.
Click to expand...

Shooter, you are going to hate me for saying this, but I think it might be your choice of field mixer. I went through 4 different brands of mixers until finally being totally happy with the audio quality of the Sound Devices MixPre field mixer.

Any chance you can provide a download link to a small 30 second sample of your audio ? I would love to see what I can do to improve the finished result.
 
Hey Dingo.

You may be correct to a degree about the hardware but I did do quite extensive tests beforew I took it to the shoot yesterday and the issue was definitely not there.

The difference in a real job was the staggering amount of "volume" the Steinway put out. Kind of scared me a little.

I believe I over compensated a little in my "gain structure". Basically all my knobs were looking pretty at around 10-11 oclock ish. Master closer to 12 oclock.

I would love to post up a sample later. I am just trying to get it to client by this arvo and its in a fedex package soon after. ( Its real urgent)

Meantime by following advice I have reduced the hiss in Audition and I normalised it.

Is it best practise to Normalise then apply Noise Reduction ( or vice versa)?

BTW - these five pieces are 28 minutes long.


...and you know...I would love a sound devices mixer ( but later..). I still havent worked out a way to tell my audio mates that I have even got a field mixer.

Question: Wouldnt one hear noise (hiss) from hardware ( camera or mixer) in any situation where you have no audio signal from a source?. ie. just monitor the device for noise
 
Last edited:
Shooter said:
You may be correct to a degree about the hardware but I did do quite extensive tests before I took it to the shoot yesterday and the issue was definitely not there.
Click to expand...

I think that the hiss could have one of four sources:

1- Ambient noise from the room

2- Self-noise from your mics

3- Self-noise from the field mixer pre-amps

4- Self-noise from the camera circuitry.

...From my own tests, the weakest link is usually the field mixer pre-amps for units costing under $800, with the exception being Sound Devices MixPre.

Shooter said:
Is it best practice to Normalize then apply Noise Reduction ( or vice versa)?
Click to expand...

I don't know what the "official" order is, but the order I tend to follow is this:

1- Parametric EQ ( if necessary, which might not be kosher for a high quality music recording, but can make quite a difference if you are subtle when applying it )

2- Noise Reduction ( if necessary, to get rid of excessive hiss and anything else that might be considered noise. This is done in multiple passes by repeatedly sampling just the noise you want to remove. )

3- I apply volume and compression selectively over parts of the audio track. This is also done in multiple passes. I find that most "Normalize" functions to be far too crude compared to "hand tooling" the audio track yourself.

Once you get used to applying these steps it's a fairly fast process.


Shooter said:
Question: Wouldn't one hear noise (hiss) from hardware ( camera or mixer) in any situation where you have no audio signal from a source?. ie. just monitor the device for noise
Click to expand...

I test by recording the mixer's 1 kHz test-tone with the levels showing -12 dB, then shut the test-tone off and keep recording your set-up's noise-floor for a few seconds. All pre-amps should be set the same way that you are going to record at. Once done, bring this recording over to your audio editor and see where the 1 kHz test-tone is metering ( it should be at or close to -12 dB ), and then see where the noise-floor meters.

With my Panasonic HMC-150 recording LINE level from the MixPre I normally see a noise-floor around -63 to -65 dB. My Panasonic consumer HS700 camera recording via a passive BeachTek adapter shows a noise-floor of -82 dB, which surprised the heck out of me the first time I tested it.
 
Shooter said:
The hiss issue. I do not know where in the audio chain to camera that I induced it. I do not believe it is anything to do with the camera.
Click to expand...
Shooter said:
I also have done a lot recording directly to the camera without any issue.

This time ...I used a field mixer and audio came in Line Level.
My un familiarity with that audio chain is what has caused my little issue. I can hear the hiss in my edit suite so I must get rid of it.
Click to expand...

Based on the above, it sounds like the issue is likely your line level output/input. How many dB is the hiss? At roughly what frequency? How much gain are you applying to the audio and are you hearing the hiss before the gain is applied?

I hope you don't mind me muddying the waters, but I would recommend not normalizing audio. Ever. It's like using auto-focus: it works okay most of the time, but has the potential to do a bad job because it's not analytical. If you're having noise floor issues, it's just going to make them worse. Normalizing also often involves a brickwall limiter on the top end to bring the peaks down on the transients, which you really want to avoid with solo piano (or any classical music recording). It is a really bad choice best reserved for automated output.

Of course, you're trying to fix a bad recording so you don't have good choices. [Insert token lecture about doing it right from the start]

My thoughts on fixing this:

DEALING WITH THE LEVELS:
You said you're looking for about 6dB gain, so I'm assuming your recording is peaking at about -6 or -7dB. That's actually pretty good, and I wouldn't worry about adding too much gain. However: you want to try to make sure that whatever gain you add doesn't get added uniformly because that will affect your noise floor. If you normalize the audio and add 6dB across the board, your noise just got 6dB louder. The solution is to add gain when the volume is above a certain level - higher than the noise floor, but lower than the bulk of the piano sounds - but taper the gain off when the sound falls below a certain level.

A good expander should be able to do this for you, or a gate followed by a gentle compressor. Cytomic's The Glue compressor may be able to do this using a really low compression ratio (even 1:1) and adjusting the range. Not sure off the top of my head and can't test it at the moment.


DEALING WITH THE HISS:
This is a lot harder, and mostly I'm speculating because I haven't heard it.

EQ is sort of your friend in that you can shelf where the noise is, but pianos are very full range, so I'd use this sparingly. Depending on where the noise is, shelfing a couple cents off the top is probably a good starting point. Possibly shaving a couple of cents off with a very narrow band if you can isolate the specific frequency range where the hiss is centered.

A gate will help if the hiss is only audible in the silences, and may be part of what you're doing to deal with your levels. Learning how to set up a gate well is pretty easy with a bit of practice.

Noise reduction might work, but I've always always heard serious artifacting with NR plug-ins. It's impossible to affect a specific frequency in one sound source but not in another. Make sure to test an appropriate section of the recording - that is, if the hiss is around 10khz, listen to how the noise reduction affects a quiet part in the upper register of the piano.


One other reason to use an outboard recorder: fixing this stuff is less of an issue. If your audio were recorded 24/96, you would be able to make more drastic adjustments without it having as much of an impact. As it is, your audio is the same quality as what will be output, so edits and modifications will be more damaging. Like shooting uncompressed, if you have more information than you need, you have more of a cushion to work with.

Best of luck.
 
Thank you for all that Dingo.

I will follow up on all that testing as you say.

Just had the job approved by client ( musician , teacher and her team) so I am burning DVDs right now.

They think the recording I did of the Steinway is fabulous! They really know their pianos so I take that as high praise.

BTW: I still had a hint of hiss after doing a NR process in Audition. I dropped the Premiere De- noiser on it with a minute amount of attenuation and the hiss dissapeared completely without any adverse affect on the piano sound. Amazed me. Clean as a whistle by my humble standards.
 
Shooter said:
I will follow up on all that testing as you say.
Click to expand...

Let us know what you find. I am curious to see if it gives any clue about your hiss issue.

Shooter said:
Just had the job approved by client ( musician , teacher and her team) so I am burning DVDs right now.

They think the recording I did of the Steinway is fabulous! They really know their pianos so I take that as high praise.
Click to expand...

A happy client is always good news. ( often we are our own harshest critics, where everyone else sees a masterful work, we will focus in on the flaws that we could not fix )

Shooter said:
BTW: I still had a hint of hiss after doing a NR process in Audition. I dropped the Premiere De- noiser on it with a minute amount of attenuation and the hiss disappeared completely without any adverse affect on the piano sound. Amazed me. Clean as a whistle by my humble standards.
Click to expand...

Good to hear. ( it's kind of like applying "NeatVideo" to image noise, a subtle touch always works best )
 
When you do this..

I test by recording the mixer's 1 kHz test-tone with the levels showing -12 dB, then shut the test-tone off and keep recording your set-up's noise-floor for a few seconds. All pre-amps should be set the same way that you are going to record at. Once done, bring this recording over to your audio editor and see where the 1 kHz test-tone is metering ( it should be at or close to -12 dB ), and then see where the noise-floor meters.
Click to expand...


...does that mean you set your tone in the mixer to -12dB ( Master) then you set the camera input (VUs) to show -12dB?

This is the area that I am well confused on regarding mixer generated tone and what is should "read" on the camera.

There is so much advice that says Mixer tone out at 0dB and match camera in at 0dB. ie matched meters.

then ...there is the advice that says -18 or -20dBs atthe camera. (At that level I dont get a reading at all from my mixer once I go to mics.)
 
nadir3000, just to say thankyou for your detailed & visual description of sounds & the shapes they make, and hence the equipment needed to describe it while recording.

Building the soundscape so to speak.

mike
 
Last edited:
Shooter said:
...does that mean you set your tone in the mixer to -12dB ( Master) then you set the camera input (VUs) to show -12dB?
Click to expand...

There is no volume / level control for the 1 kHz test tone on my SD MixPre, you simply turn on the tone which is calibrated by the manufacturer to be the loudest signal that the mixer can generate without distortion. Technically I guess this would be 0 dB output.

The reason you set your recorder/camera to -12 dB ( or -20 dB on the higher end units ) is that there is no "headroom" in the digital world, so by setting our maximum signal to record as -12 dB we are creating our own 12 dB of headroom, which will guarantee that nothing will ever clip at the recorder end of things. Super loud sounds might overload your microphone and distort, but the recorder will always have that 12 dB of headroom to act as a safety-net.

With higher end recorders, some people will use -20 dB for their "headroom", because these recorders have a noise-floor of -100 dB or even lower, so throwing away 20 dB to use as "headroom", will still leave them with 80+ dB of dynamic audio range.

You might find that setting your recording level to -6 dB will give you that 6 dB of extra dynamic range ( compared to recording at -12 dB ), but you should test to be absolutely sure that nothing will ever clip when recording this way. ( thus a -6 dB recording level, with a noise floor of -63 dB for my HMC-150, the dynamic audio range would be 57 dB )

I hope that some of this helps. It can be very confusing when you first try to wrap your head around this stuff. :)
 
Can you post an AIF or WAV file now? .. I'd love to hear it .. and might be able to analyze the
noise as well.

jeff
 
Shooter said:
When you do this..




...does that mean you set your tone in the mixer to -12dB ( Master) then you set the camera input (VUs) to show -12dB?

This is the area that I am well confused on regarding mixer generated tone and what is should "read" on the camera.

There is so much advice that says Mixer tone out at 0dB and match camera in at 0dB. ie matched meters.

then ...there is the advice that says -18 or -20dBs atthe camera. (At that level I dont get a reading at all from my mixer once I go to mics.)
Click to expand...

If your camera / meter is calibrated to a +4db reference, it will be close to -12db.

I'm a little less conservative than Mr. Dingo. The test tone is the loudest signal that is going to be
emitted from your pre-amplifier. If a signal is presented to your pre that pushes it louder than this, it
will start to distort the pre-amp. As such, I back off between -3db and -6db.

The reason is is that camera audio is 16bit .. and that's 96db of dynamic range. If you back off
-12db, you're now down to 84db of dynamic range. A grand piano can have a dynamic range of up to 90db ..
It's not going to clip .. but .. the quieter passages of the music may not be as detailed/distinct.

Once you've set your "max" on the camera, you then set the pre-amp gain such that audio signal
peaks at around -6db. Sometimes peaking is hard to gauge .. so if you're not sure, you can be
a little more conservative and dial it back a bit more.

If you have a noisy pre-amp, you may have to play with the output level of the pre vs. the input
gain that you apply to minimize the noise/hiss through the pre. If your pre has a good limiter,
I recommend using them for live performances ... but, you try your best not hit the level that
engages the limiter.

As with video .. take a "test shot" and analyze what your getting and make adjustments as needed.
Sometimes you don't get lucky with a plug-in.

love to hear what you got...

jeff
ps: as a reference 6db is "twice as loud"
 
If your camera / meter is calibrated to a +4db reference, it will be close to -12db.
Click to expand...

AF102 - I have no idea what the meter is callibrated to. The AF102 User manual ( ha ha) also confuses me with the paragraph that states this camera is set up 8dB higher than other panasonic broadcast cameras. THe reason for that is unfathomable to my simple brain!

Will post up some audio parts today now the rush is over.


Thanx for taking interest you guys. I found it a very absorbing process - playing with the audio side of what I do.
Craig
 
Last edited:
TheDingo said:
The reason you set your recorder/camera to -12 dB ( or -20 dB on the higher end units ) is that there is no "headroom" in the digital world, so by setting our maximum signal to record as -12 dB we are creating our own 12 dB of headroom, which will guarantee that nothing will ever clip at the recorder end of things. Super loud sounds might overload your microphone and distort, but the recorder will always have that 12 dB of headroom to act as a safety-net.
Click to expand...

This strikes me as similar to saying the WFM should never go above 75% for fear of a bright light suddenly appearing.

You're recording 16/48. If you're setting your maximum dB at -12, then when you bring the volume up 10 or 12 dB while finishing it you're guaranteed to end up with noise.

Trying to solve a potential problem this way guarantees another.
 
The Dingo and Jeff

Sent a link to the files in a PM

Here are pix of the set up. Forgot to get a picture of the AF102 because I was so focussed ( freaked out) on the audio set up and also...I was setting up lights, camera and the mics by myself! All had to be set up in 2 hours before the shoot.

Shot 720p PAL. ( pixel peepers note ..lack of banding in the ungraded frame grab) I have never yet seen any banding on anything I have shot BTW.


AuditionEditStill001.png


smP1000812.png
 
Last edited:
Nadir3000

Certainly want to thank you for your input as well. Very valuable and interesting info.

There is certainly more than one way to skin this cat and I have now read / heard a few of them.

BTW - Normalise. My audio was peaking around -12dB hence why I thought 6dB would be helpful. This is for DVD presentation.



Craig
 
You must log in or register to reply here.
Back
Top