Audio levels too low

Overview
Audio levels are too low in both Resolve & Vegas Pro. My audio recordings barley wiggle on the audio timeline.

Test
Not having an oscillator, I hooked two wireless mics to my Blackmagic URSA camera and placed them in front of a speaker. Played a 1khz sine-wave tone, and adjusted the volume so that all green LED on the camera were on. I recorded a few seconds of this tone, moved it into Resolve.

Problem
In Resolve the tone only comes in using about 10% of the vertical audio scale, instead of what I should think would be about 80% of the vertical audio timeline scale. It plays back at -16DB on the VU meters

I rendered it out as a Quicktime MPEG4 file, before importing it to Vegas. In Vegas, it also only uses up a tinny bit of the audio vertical scale, and plays back at -16DB on the meters.

Questions
1. What could be wrong that this audio isn't showing up as a more robust signal on the audio timelines?
It seems that with all green LEDs on, the audio should come in at more than -16DB.

2. Is there a way to increase the audio waveform vertically, for easier editing, without changing the actual audio level?

Not sure how the URSA's meters are calibrated or what's the bit depth of their converters, but -16dBFS is an absolutely reasonable level at 24 bits or above. Audio apps usually allow changing their meter calibration (e.g. Apple's Logic can display as either exponential or linear-segmented, the latter making the meters "appear" fuller). Actually, I never go above -10dBFS for any one track, and aim to have my master at -6 (a kind of standard for mastering studios). Working with analogue emulation plugins means even lower levels.

As long as you don't hear any noise, you should be fine. Being able to scale the waveform display would be really nice though (all audio apps can do that), really helps with editing.

Dani:

Since posting, I raised the tone to light all the orange LED's on the camera. That brings the waveform up to about the 80% level, and plays back at -4dBFS. So, I may have been looking at this all wrong. I'm old school and thought I should be in the green with only an occasional orange LED lighting. Now, I see I'll have to drive the audio at the camera so that all the orange LED's at lit with only an occasional red. Because even with all the orange on I'll have to make up 4db in the program with the resultant loss of a little S/N.

Your conservative levels (-10db on channels, and -6db for final output) is interesting. Is it to ensure enough headroom.

Thanks for your response.

Dave

Hi Dave,

The long answer, if you're interested. I hope I don't get you confused. I'm an audio engineer by trade.

Yes, my level suggestions pertain to headroom. What would be important to know is whether the URSA has 24-bit converters (I record with a dual system, so I know mine do). At 24 bits, you have enough sampling detail to record (and work) as low as needed with zero quality loss.

The slightly more technical reasons for recording lower has to do with analogue levels. The analogue level reference in, say, mixing desks, audio outboard, etc, is +4dBu (note the "u" at the end), which is another way to say that the analogue meter on it will display 0 VU (volume units, i.e. target reference volume) at 1.23 Volts, measured from a 1kHz sinewave. This is the point where the analogue electronics behave well. Of course, most equipment, like the mic preamps in the URSA, can go much higher without distorting. Other audio gear exhibits a kind of pleasant distortion at this super-hot range. Bear in mind, this is used for music recording purposes, and would probably sound bad for lav mics or shotguns on booms.

The point is that converters are universally designed to accept much higher levels than +4dBu, both for the possibility of overdriving, as well as to provide some good headroom. They meter on a different scale, called dBFS (dB full-scale), which means simply "how many dB before clipping". So, a typical consumer level AD converter would be calibrated at -14dBFS, meaning that when its meter is -14, the reference level at its input is the nominal standard of 1.23 volts (otherwise known as +4dBu, displayed on analogue equipment as 0VU). A high end converter can be calibrated even lower, -18, or -22 dBFS, in which case the nominal level is achieved there.

The converter does not care if the audio is at nominal level. Most decent ones are perfectly linear above the usually ridiculously low noise floor (waaaay better than what your mic and preamp does in terms of SNR). However, the URSA also has mic preamps plugged into it, and there is no way to monitor the analogue level between the preamp and the converter (both internal components). In this case, most people would assume the -14dBFS standard, to get the preamp to 0VU, and thus using it at its best.

On the playback side, the level of an audio file depends on the DA converter (i.e. your audio interface, or the DA inside your, say, Ultrastudio, or whatever you're connecting your audio desk, amp, or powered speakers to). If that has the same calibration, then a -14dBFS file will land right on the Unity mark on your mixer, with no gain adjustments. Amps and powered speakers can handle quite a bit more with no problems. My setup, for example, uses -18dBFS converters, that feed an analogue desk. If I then want to manipulate the volume I'm listening to, I'm using the monitor controller on the desk, but that is now getting a clean signal. If I output something from iTunes at full blast, it immediately hits the reds on my desk.

The -6dBFS for final mixes is entirely arbitrary. It turns out that for most material, peaking at -6 means that most of the normal levels are at about -14, always depending on compression. Remember, -14 is not meant as a maximum, but as an average. It's perfectly ok to go higher, that's what headroom is for, to be used. My mastering guy usually turns those down some more, to fit the calibration of his system.

So, you see, digital levels don't mean much by themselves. It depends on your system's calibration, much of it you can't change. For the URSA, the best thing would be to connect a line level signal at +4dBu, and see where the meter lands. Then, use that as your reference level for the mic preamps, i.e. the average point around which your meter should move; it's ok for transients to go above that, as long as you don't see that red.

For the editing, the best thing would be to mix your sound in an audio app (export an OMF or AAF from Vegas, with a ref video proxy render and all the audio files separate). You'll have a much easier time, more tracks, better editing capabilities for sync etc, sample-accuracy, proper meters, demonising plugins, and the ability to scale the waveforms at will. Then you mix your audio to a single stereo file (-6dB if you go to a mastering house with it, -1 for web delivery, dolby levels if you're doing it for cinema, whatever is relevant) and combine with your master image sequence for final output.

Dani Iosafat is right
some additions in "laymen terms"

Since posting, I raised the tone to light all the orange LED's on the camera. That brings the waveform up to about the 80% level, and plays back at -4dBFS. So, I may have been looking at this all wrong. I'm old school and thought I should be in the green with only an occasional orange LED lighting.

occasional orange that is most of the time correct!
-4dBFS means 4 db before clipping and thus to loud.
People think that 0dbfs is the sweet spot, like on "old" analogue gear. as Dani stated (0 /+4 dbu is not 0 Dbfs)
But everything above 0dbfs sounds horrible.
-4Dbfs means roughly +14Dbu.Not exactly because it also depends on the gear and country, but close enough

If i'm not mistaken
EBU in most European countries, specifying +18 dBu at 0 Dbfs
SMPTE (America) standard defines +20 dBu at 0 Dbfs

Now, I see I'll have to drive the audio at the camera so that all the orange LED's at lit with only an occasional red.

that's definitely to loud

Because even with all the orange on I'll have to make up 4db in the program with the resultant loss of a little S/N.

when you do that you got probably peak clipping, not recommended!
Even after mastering (which is a finishing proccess after the mix) levels should NEVER exceed 0Dbfs.
Recommencement max level is -0,03 on the loudest peaks

Your conservative levels (-10dbb on channels, and -6db for final output) is interesting. Is it to ensure enough headroom.

Yes for headroom and is not conservative at all, especially in multi track/channel formats.
When the channels increasing the "mixbus" output get louder.
The effect is the most in overlapping frequencies

I hope that did make (some) sense

Regards,

Remco Johannes

Dani:

First let me say how much I appreciate your taking the time to provide this stimulating background on audio levels. I would also like to thank Remco and Simon for their input as well.

I'm embarrassed because my head was adrift in the world of old analog audio. I was narrowly thinking of the audio/video files brought into Resolve, and hardly appearing on the audio timeline, with out zooming back to think about all the other issues. My head is still spinning, but the fog is clearing.

The URSA camera specs are not very complete, so I'm not sure if the converters are 24-bit. Somehow I think they are, but could not find information on that.

I think I can work this through now. Again, Thank you all for your help.

Dave

Some great advice in this thread. I would be very wary of hitting levels too hard in any camcorder (BMD included), because the preamps in those cameras are cheap, crappy, and prone to overload -- in my opinion. I generally put a mixer at the head of everything, have a limiter in place (to prohibit peaks over about -5dBfs), and I aim for average dialog levels to peak at about -10dBfs. If I have the ability to record isos for separate mics, those are at least -5dB lower.

I would do extensive tests to make sure the level out of the wireless mics is compatible with what's going into the camera. If the camera can accept line-level signals, use that. Be aware that there are consumer line-level (-10) vs. professional (+4) signals, and the latter can overdrive the former. A surprising number of allegedly pro camcorders use consumer levels. All the ones I've used that employ XLR connectors are using pro levels, but there's always a bonehead exception out there.

If possible, always record backup audio for anything going to camera. Things can and will go horribly wrong when only recorded to camera. Something as minor as the camera running off speed, losing a sync reference, or the camera operator bumping a knob can derail the entire process.

Marc:

Good points all. I will say that the audio the URSA provides is, in my opinion, pristine, so it's a good starting point, and worth the calibration effort.

Thanks,
Dave

Good points all. I will say that the audio the URSA provides is, in my opinion, pristine, so it's a good starting point, and worth the calibration effort.

The metering and connectors on the Ursa and the Alexa are actually pretty good, and I've heard good audio on both on occasion. In situations where I've worn my sound hat, I've delivered sound files for Alexa projects and the producer told me, "eh, we listened to the Alexa audio you gave us and it was great, so that went in the finished commercial." Luckily, the scratch track was fairly pristine, though not quite as perfect as the sound recorder audio. On the plus side, I got paid.

http://www.independentrecording.net/irn ... metergain/

Simon:

You're a man of few words, but you hit a home-run with this one. The interactive chart on the link you sent is absolutely wonderful. Full of information, clearly written, and easy to find and read. It has many things I had forgotten, and lots that I didn't know. It is simply a wonderful explanation on the whole topic.

What a great help – thank you, thank you!

Dave

thanks dave, this is a very impotent topic and hard to understand this chart !

http://www.independentrecording.net/irn/resources/metergain/

Interesting but how do you use this source. I am a dummy in sound engineering. The audio levels of my imported audio files are too low while outside davinci they are normal and optimized for use. Since I am an abobe audition user I've fixed the audios there. In davinci the project settings is set at its default:
Target Loudness level: -23 LUFS
Loudness scale: EBU + 9 scale

How do I change the settings so that the audio files are played and rendered as they were before being imported.

The audio levels of my imported audio files are too low while outside davinci they are normal and optimized for use. Since I am an abobe audition user I've fixed the audios there. In davinci the project settings is set at its default:
Target Loudness level: -23 LUFS
Loudness scale: EBU + 9 scale

These settings only apply to metering, and do not affect the source audio.

viewtopic.php?f=21&t=69747&hilit=+loudness#p389317