Sound Design Stack Exchange is a question and answer site for sound engineers, producers, editors, and enthusiasts. It only takes a minute to sign up.
Sign up to join this community
Hey, does anybody have any suggestions for removing the low level hum of a microphone?
It's noticeable enough to be heard across cuts.
I've tried removing it with Izotope but I'm not confident that it is worth it because it somewhat effects the audio quality of dialogue, etc. How much is too much when you affect the quality of stuff through noise reduction? I ask this because I realize that other people don't hear things the same as the audio community.
When you reduce noise you often have to put something else in there instead, in order to mask the artifacts of noise reduction. You can also mask the noise without removing it with a nice background atmosphere.
Noise will always be a part of the sound you capture. You do not necessarily need to remove it, if it isn't annoying at normal listening volume. I would rather listen to a bit of noise than I would listen to an extremely noise-reduced sound.
If you get a lot of broadband self noise from the microphone it can be because the gain is too high or there is an auto-gain in the camera, constantly amplifying unwanted noise. This can be pretty hard to remove, so try a combination of noise reduction, gain automation and masking with background tracks.
One trick I like using is one I learned when someone taught me how pickpockets steal from you.
Now, hear me out:
Pickpockets bump into you at another spot of your body, like your arm, and while they bump into you they also reach in your pocket and steal your wallet. The trick is that you're being hit in the shoulder or arm and this is where your attention goes to and you don't notice that you also have someone reaching into your pocket to steal your wallet.
I have sometimes (not all the time) used this to distract the audience's attention to something blatantly obvious like an out-of-sync ADR line that can't be helped or a noisy recording.
The last time I used this successfully was in a music video where the director wanted to hear the person the camera was focused on but I didn't have the original dialogue because it was a b-roll camera which wasn't taking good audio so I didn't have the line at all. I edited it together from other lines the same person said in a different scene and the sync was OK but not perfect. The scene was inside a diner, so I put in an effect of someone dropping plates right when the camera focused on her so you hear this a bit more predominantly than the dialogue, but you still hear the dialogue. This helped the fact that it wasn't what the person originally said but it worked because you don't really notice it was a bit out of sync because I used this "pickpocket" type of technique, distracting the audience for a moment with another sound and it worked pretty good I think. At least the executives on the approval lines didn't seem to notice. Beware over-doing it because in most cases you'll need that noisy line of dialogue to be clearly understood, in which case this technique is pretty much useless. But I thought I'd mention it anyway! :-) Also, always check your pockets when someone "accidentally" bumps into you in the airport or subway or other crowded places.
Excellent points, Morten.
The other thing about "hum," is that it has a lot of additional content at octaves. For example, when we refer to the sound of 60-cycle hum, the most prevalent frequency we're actually hearing is aactually at 120hz. There is additional content at 240, 480, etc. A comb filter can help reduce that, but it will need to be used with care because of the effect it will have on other content within the signal (also notice that I only said "help").
Are you using the Spectral Repair part of your Izotope Rx ? Pulling out specific frequencies as opposed to the broadband 'wash' often means you get less artifacting for this sort of noise.
Or, if you have the waves Q10 it's amazing what you can do with a tight, tight bandwith to remove specific frequencies. If they're constant it's really easy... if they move.. it's a bit trickier. As Shaun says they'll be based around whatever your power runs at (50hz, 60hz) but could be any of the harmonics (multiples) above that number, 60, 120, 180, 240 etc.
... in response to your comment:
@Chris, ah ok, I think the 'hum' in the subject line might be at odds with the kind of noise you're dealing with. Have you tried some fast downward expansion in the upper ranges? If you have the C4 or ML4000 try setting the top two bands of the expander to a range of eg.-6db and set your thresholds so that when there is no dialogue there is -6db of expansion, but when there is dialogue it is opening up to 0db. Have the attack as fast as you can, and with the release find the line between following the decay of the words naturally while avoiding 'blooming' where you can hear the noise tailing out.
It's amazing how much you can trick the ear simply by turning things down. This will avoid the artifacting you're talking about but maybe introduce other problems... (take a listen to exterior scenes on tv shows that have obviously not been ADR'd and you can often hear the ext noise moving with the dialogue.)
Lesser of two evils?
Wow, I have now discovered the purity of downward expansion.
