Audio help needed..

[Brizzol]

Hello everyone
I wasn't sure which section to post this in so as a newbie i have posted it here
Its a bit of a long winded problem but i'll try and keep it as brief as i can.
I have recently got into videography after being passionate about still photography for many years and always been interested in the video side of it.
So going back last year I filmed a dance show using my DSLR and produced dvds which i got very good feedback from, and i was recommended by this dance school to a musical theatre company who were looking to have their latest production filmed.
I knew my dslr would be no good for this gig as it has a 30 minute video cap on it, so i made the jump and purchased a Canon XA11, i found it the audio capabilities of this camera to be quite overwhelming so i read up on it as much as i could and watched countless youtube videos to try and get my head around it.

So fast forward 1 month and I'm at the theatre set up ready to film, i asked their sound tech if i could get an audio feed from their desk, to which he gave me an XLR cable which i connected to my camera mic input (bearing in mind, this side of the camera i had never used and knew very little about) I asked for a sound check and played around with the settings until it sounded good to me through my headphones.
When the show started the music coming through my feed sounded great, until the singing started!
I was getting a lot of distortion when they were singing, the cast were wearing wireless mics which was all going through the sound desk, the sound from the theatre speakers sounded fine but the sound i was getting was awful.
If i lowered the volume on my camera for the sound input it didn't improve it, it was still distorted just at at a lower volume.
I asked the sound tech if he could lower the feed volume for just the cast mics which he did but still didn't solve the problem.

Does anybody have any ideas what could be causing this? Its really frustrating as the sound from the theatre speakers sounds perfect, as does the music feed i was getting.

Thanks in advance for anybody that can offer any advice.

[betwixt]

Some of the sound system is between the XLR and the level control on the camera. It seems the level from the desk was overloading your XLR input and causing the distortion before your control had chance to reduce it. There is no easy solution, you did the right thing by asking for the feed to you being reduced but it needed to be dropped even lower to prevent the overload.
Things to try: I'm not familiar with the XA11 but see if it has a setting to change the XLR input between microphone input and line input. Microphone signals are small, line level is quite high so if you were feeding line levels in and it was set to microphone they would be far too high and cause the symptoms you saw. If that fails, I would suggest getting a small external mixer of your own so the levels from a desk or any other source can be optimized before they hit the camera.

Brian.

[Cornucopia]

I agree with betwixt.

That sound mixer should have asked whether you needed mic level or line level, and he SHOULD have given you a test signal that would be equivalent to what one would receive when "all inputs running at full level" (which you should have set to somewhat below 0dB fullscale - -6dB to -24dB depending on how much headroom you expect to need) so that whatever was going to the PA should be mirrored in what was going to you.
In a perfect world, the mixer board would have a dedicated output submix (1 for you, 1 for PA, etc) with switchable mic/line, making this a breeze.

Scott

[aedipuss]

doesn't really matter at all. if they had an audio board, just ask for the recording and sync it to your video, while deleting your audio.

[Brizzol]

Some of the sound system is between the XLR and the level control on the camera. It seems the level from the desk was overloading your XLR input and causing the distortion before your control had chance to reduce it. There is no easy solution, you did the right thing by asking for the feed to you being reduced but it needed to be dropped even lower to prevent the overload.
Things to try: I'm not familiar with the XA11 but see if it has a setting to change the XLR input between microphone input and line input. Microphone signals are small, line level is quite high so if you were feeding line levels in and it was set to microphone they would be far too high and cause the symptoms you saw. If that fails, I would suggest getting a small external mixer of your own so the levels from a desk or any other source can be optimized before they hit the camera.

Brian.

Thanks for your reply.
It makes sense that the feed needed to be lowered further but the music i was getting from the feed and even the spoken dialogue was perfect which is where i dont know enough about this stuff to figure it out.
I do have an input source selector on the camera which i did have set to line in. i do like the idea have having my own mixer to prevent this happening again.
Thanks again.

[Brizzol]

I agree with betwixt.

That sound mixer should have asked whether you needed mic level or line level, and he SHOULD have given you a test signal that would be equivalent to what one would receive when "all inputs running at full level" (which you should have set to somewhat below 0dB fullscale - -6dB to -24dB depending on how much headroom you expect to need) so that whatever was going to the PA should be mirrored in what was going to you.
In a perfect world, the mixer board would have a dedicated output submix (1 for you, 1 for PA, etc) with switchable mic/line, making this a breeze.

Scott

Thanks for your reply.
It makes sense what you have said but unfortunately the only sound check i had to use was just music without and singing.
I had my input volume controls on manual so i could monitor the levels on screen and adjust them if i needed to, do you think if i had them set to auto, the camera would have been able to adjust to the correct levels?
Thanks again,

[Brizzol]

doesn't really matter at all. if they had an audio board, just ask for the recording and sync it to your video, while deleting your audio.

Thanks for your reply.
As far as I'm aware they weren't recording anything, would that be a normal practice in this situation to record the audio from the performance?

[Cornucopia]

Normal practice? It varies. Smaller shows or venues it is only like 20% but it grows with larger shows or larger venues. It is never 100% because of inevitable licensing issues.

Auto levels is a cheat, and is almost never right unless the levels vary WILDLY and you cannot afford the dynamic range.
Plus, it would do nothing to help fight distortion if the distortion is pre-existing due to overloading the input stage (prior to digitization).


I'm curious, what level did you set your cam/recorder to, given the test feed?

For example, if I were getting a line level feed from the board, and the board op set their master output with a 100% level (aka +4dBu on pro boards) of a 1kHz test tone, that should be set to something like -18dB digital fullscale on the cam/recorder, to allow for 18dB of headroom.

My guess is that:
1. The board op didn't give you anything close to 100% with his/her test level
2. You probably didn't give that much headroom, if any (-10, -6, -3, 0dB)

So as soon as they started adding channels/voices, it tipped over beyond your headroom.

These happenings are unfortunately not uncommon if one or more parties is not a seasoned vet w pro audio practices.


Scott

[Brizzol]

Normal practice? It varies. Smaller shows or venues it is only like 20% but it grows with larger shows or larger venues. It is never 100% because of inevitable licensing issues.

Auto levels is a cheat, and is almost never right unless the levels vary WILDLY and you cannot afford the dynamic range.
Plus, it would do nothing to help fight distortion if the distortion is pre-existing due to overloading the input stage (prior to digitization).


I'm curious, what level did you set your cam/recorder to, given the test feed?

For example, if I were getting a line level feed from the board, and the board op set their master output with a 100% level (aka +4dBu on pro boards) of a 1kHz test tone, that should be set to something like -18dB digital fullscale on the cam/recorder, to allow for 18dB of headroom.

My guess is that:
1. The board op didn't give you anything close to 100% with his/her test level
2. You probably didn't give that much headroom, if any (-10, -6, -3, 0dB)

So as soon as they started adding channels/voices, it tipped over beyond your headroom.

These happenings are unfortunately not uncommon if one or more parties is not a seasoned vet w pro audio practices.


Scott

I only asked if it is normal practice as someone had suggested to ask for the audio recording, and as I'm new to the video/audio side of these performances i would never have thought that the audio team would record the sound from the performance.

I kind of guessed that the auto audio levels wouldn't get it right which is why i stuck to manual, even though i didn't get it right, i have certainly learned a lot from the experience, a lot more practice needed.

"For example, if I were getting a line level feed from the board, and the board op set their master output with a 100% level (aka +4dBu on pro boards) of a 1kHz test tone, that should be set to something like -18dB digital fullscale on the cam/recorder, to allow for 18dB of headroom."

That's all a bit over my head, but when i was given a test feed i set the on screen audio levels so they were around 10,
But even after the voices kicked in and i knew it was distorting, adjusted the levels back down again but i was still getting the distortion, it was almost as though the actual feed i was getting i was getting was too much for my camera to process. ( kind of what betwixt was describing )

[Cornucopia]

As per pg. ~75 of the manual, you should be setting the input level of the audio to ~ -18dB (it says the scale goes from -Infinity to +18, and says it will show red if it hits or goes over 0dB). If you are setting it to 10 (?10dB?) I would expect that to be too much, as that is 28dB higher than the recommended.
Also the manual mentions using a combination of Input Sensitivity switch and Attenuator on/off switch. THOSE should assist in preventing input overload.

Spend extra time to test all this ahead of time, as once you've got distortion, it is near impossible to get it out.


Scott

[Brizzol]

As per pg. ~75 of the manual, you should be setting the input level of the audio to ~ -18dB (it says the scale goes from -Infinity to +18, and says it will show red if it hits or goes over 0dB). If you are setting it to 10 (?10dB?) I would expect that to be too much, as that is 28dB higher than the recommended.
Also the manual mentions using a combination of Input Sensitivity switch and Attenuator on/off switch. THOSE should assist in preventing input overload.

Spend extra time to test all this ahead of time, as once you've got distortion, it is near impossible to get it out.


Scott

Ok, apologies, I wasn't very clear in how i explained that.
So what you are referring to is the manual audio level controls on the camera, i had these set to 4 ( with 5 being 0db ) the on screen audio meter goes from -40db to 0db, which i was keeping at around the -10db mark.
But as you say it does recommend -18db, although that does seem very low when looking at the audio meter.
I have watched a lot of videos and read as much as i can find on learning to use this camera, although most people seem to recommend not using the attenuator as it will make the rest of your audio too quiet, I haven't tried it so cant comment on that.

I'm very much a beginner but i appreciate all your advice.
Thanks

[Cornucopia]

When I was doing audio production full time (which I did for 17+ years), one would set the level for an RMS (aka near steady-state) test tone, and it would be lower than a layman would think should be put. But dynamic range comes into play. The difference between even that simple sine wave test tone from steady-state RMS level vs. instantaneous Peak level is 3dB. Everything goes up from there. Some percussive sounds, for example, have a difference of 32dB (between RMS and Peak).

If it was just spoken word, depending on how variable the speaker/source is, you might set it at -6, -8, down to -15, though I usually chose -12dB.
If it was music, or mixed, it would be more like -16, -18, down to -24 or even -28dB. This varied depending on the kind of music - contemporary pop variations would all be on the less dynamic side (especially as mixed post-90's) needing less headroom. Classical OTOH or certain kinds of Jazz would use the most dynamic range, and so would need more headroom. For general purpose, I often chose -18dB.

The beauty of digital is that you normally have either a LARGE amount of dynamic range to work with (96dB at 16bit) or a HHUUGGEE amount of dynamic range to work with (128dB at 24bit)), sometimes more. So it doesn't hurt it much to drop it down a little bit more than you would think, or more even than necessary, in order to keep a clean signal. 96dB - 18dB = 78dB, which is still better than any FM radio station (~60-65dB). You might hear a bit more hiss, but that is much better than distortion.
The weakness of digital (at least normal integer digital) is that if it hit 0dB full scale for more than 1 sample, it is instantaneous square wave distortion, a different, usually more obnoxious character.

As to distortion, it depends on whether it is analog or digital in your signal chain. Both are bad but in different ways. Setting the levels lower (as just above) adjusts the Analog-to-Digital conversion, and thus the digital level. Setting the sensitivity and the attenuator (at least AFAICT from the manual) adjusts the analog input gain stage, pre-digital. Listen to the character of the distortion to see which one is at fault. If it is more "oomph-y" or "crunch-y" it is likely analog stage. If it is more "hash-y" or "screech-y" it is likely the digital stage. Adjust accordingly.

**There is a trick you can do, especially if you are recording dual mono, which is to "bank" the levels (think of the word in reference to taking a curve at high speed). Set one track to a standard dynamic range input level, set the 2nd (most people have 2 channels) to a lower level - say -6dB or maybe -10dB or maybe even -20dB - and then use the track that works best with the given section of recording (you would have to do cuts with ramped crossfades, etc to make in not too noticeable that you are switching tracks). Or it is even possible, though not easy, to mathematically combine the 2 tracks and create a single track which both has the headroom needed and greater dynamic range than is possible with just 1 track. You DO need to use a higher bitdepth when you do this (e.g. go from 16bit to 24bit, 32bit, or perhaps just use floating point). The process is quite similar to how one stitches together bracketed camera shots in order to achieve a High Dynamic Range image (HDR) in post.

Hope that helps,

Scott

[Brizzol]

Thanks Scott, that is a wealth of information, that's given me a lot to think about and research.
I shall read that a few times and try and absorb it.
Thanks again.