2

I am working with actor video where he unfortunately recorded on-screen in a room and off-screen in an almost anechoic space. While I have been able to get rid of a lot of the frequency differences between the recording microphones (one's a lavalier and the other is not) and can almost match the reverberation, the recording in the room is still quite boomy.

As I was listening, I remembered that in architectural acoustics class, we learned that the reason people enjoy singing in the shower isn't the reverberation, but the room modes, where wave radiation patterns amplify certain frequencies as they reflect off surfaces; think resonance. I believe that is the problem with matching my two sources.

Does anyone have any experience or hints as to how to add the modal sound to my studio version? I am only trying to match the two sources, and it's easier to add than to take away. I tried to add some reverb confined to lower frequencies, but it just didn't sound right. Does anyone have a tried and true way to add these? Delays? Some sort of resonance plugin?

4 Answers 4

0

Try the bark of Dog plugin. It's a resonant EQ, most use it to when a VO/Narrator needs a little more "boom" to the voice but you should be able to use it to match recordings also. An issue that usually comes up in in a situation like yours is that you can get very close on the EQ curve itself but the talent wont have the same projection in the quiet space as in the louder one.

2
  • Is this the Boz Digital Labs Bark of the Dog plugin? I will try this out first probably because this seems like the simplest but if it doesn't work, I bet Joel's response will work, albeit will require more work. Aug 10, 2015 at 2:37
  • I ended up using this and a little EQ but I couldn't get it exactly right. I think the space was different and so was the microphone, so it wasn't too easy to get it right. In the future, the actor should just record the same setting for all the audio if it is meant to match. Jun 29, 2016 at 14:23
1

You need to use EQ/Equalization on the dry voice and boost resonant frequency with a narrow notch filter. It might match closer if you then apply a basic reverb afterward, but it will take some trial and error.

The important thing about modal frequencies is that practically they are just a higher amplitude of that frequency. If you can identify one or two of the boomy frequencies in the original (by reducing them with an EQ) you can try boosting them in the other audio. Practically stick with finding the lowest boomy frequency and try boosting an additional one or two multiples of that (e.g. 100hz; 200hz, 400hz)

I'm not sure why you need to be so precise, but it takes time to match the correct frequencies and additionally the correct reverb pattern. There aren't any other special plug-ins to rely on other than Convolution reverb, which works from an Impulse-response of the original room, but you still have to take into account the tone of each of the microphones.

5
  • “important thing about modal frequencies is that practically they are just a higher amplitude”... nah, it's really not that simple. To adequately model room response with IIR EQs you need an extremely complex mix of comb filters, shelving, and indeed narrow boosts; it's so many parameters that I really wouldn't recommend going this route. A convolution reverb can do all this in a single step, and much more accurate & systematically. Aug 9, 2015 at 18:56
  • EQ is indeed simple. The main problem is that the room recording is boomy(er), not the the recording needs to be objectively identical. Even in acoustic room treatment it's essential to look at the most offending low frequencies. Of course an impulse response would be nice but it's also idealistic and not something even audio professionals may have the luxury to utilize. Aug 9, 2015 at 21:43
  • I might not need to be super precise, but if my client tells me that I didn't match it well enough, I'd like to know how to do this, especially because he will be coming back to me with more work. Also, curiosity got the better of me. Aug 10, 2015 at 2:32
  • David, it's unusual for you to be looking for an alternative workflow like this for future reference. It is fairly arbitrary to be asked by an actor to match the reverb of a room; why can't you simply use a microphone that captures more of its sound, like a boom mic? Otherwise, you could typically get away with choosing a reverb to your sound designers taste and not based on some independent "objective" reason. Aug 10, 2015 at 3:02
  • @JoelPinteric: I don't see what's “fairly arbitrary” about this. It's not unusual that you'd like to record a lot of footage in some prestigious location, but can't afford it. Well, means you need to fake it, and as for the audio side, convolution with a single IR is very much the right way to do that. I also don't see why this would be a “luxury”: IR capture requires neither much equipment you would't have anyway, nor is it an issue processor-wise. Aug 10, 2015 at 9:54
0

If you can, go back to the room that was used for the recording, and get an impulse response, ideally from where the actor's head was to the same lavallier mic. An IR is a full “fingerprint” of a room's acoustic character: if you combine the anechoic-recording with a convolution reverb using that impulse response, you can get virtually the same result as if it was recorded in that room, with all resonances.

In practice, this doesn't always work as perfectly (some correction may be required for the speaker/clap used for the test impulse), but it'll generally be vastly better than anything you can achieve with manually-tweaked EQ and generic reverb.

1
  • I will try to get the actor to do some claps for me before the next round of recordings, which may be the best he can offer. Besides the issue of using a clap rather than a full IR source through a speaker, I think that because he uses a lavalier microphone for the on-screen voice over, I think the IR recording might fall short because of the relatively huge difference in source location. Aug 10, 2015 at 2:35
0

As said in an other answer, resonance is merely a peak/peaks in the spectrum. Therefore it's replicable by a simple EQ peak or multiple peaks (some EQs allow you to gain/reduce the harmonics/inharmonics of the main peak as well).

There are however more sophisticated tools as well such as if you can find a "harmonic synthesizer" that works in the vocal range (I've only seen "bass extender" types).

Also there's a design called "resonator bank" like this:
https://www.kvraudio.com/product/resonator_bank_by_starplugs

Also, a convolution reverb might be able to add resonances if the used impulse is "resonant".

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.