Take the 2-minute tour ×
Sound Design Stack Exchange is a question and answer site for sound engineers, producers, editors, and enthusiasts. It's 100% free, no registration required.

Hi there,

I'm currently doing a dialogue edit for a film, and there's a substantial amount of voice over. Knowing the studio where the recording was to take place well, I asked the recording engineer to be very careful with mic choice, placement and general position of talent within the booth to eliminate the horrible room reflections the studio is known for. Unfortunately... the recordings arrived today, and all I hear is that damn room.

The narrator is male, fairly deep voice, with nice character. But he sounds like he's speaking inside a box. I've done some eq to highlight his tone, but can't seem to get rid of the room altogether.

There's no possibility of re-recording at this time.

Anybody have any success with this problem before? I usually do my own voice over recordings, so I haven't run into this problem a lot.

Any help would be greatly appreciated.

Thanks,

NB

share|improve this question
1  
Do you have a decent quality impulse response of the room? Deconvolution is a possibility –  kibibu Apr 11 '10 at 11:22

12 Answers 12

Nick,

I've been working on a show that was shot on greenscreen in a soundstage. Fine for the indoor scenes but some take place outside.

From a little searching I found a plugin called De-Verb made by SPL. I use it as a plugin and automate it during the mix as sometimes I can get away with less of it when there's lots of sound and music going on.

Have to say I've been really impressed. It works really well, unless you push it too far then it starts to destroy the dialogue. If used subtly then the results are quiet excellent. Best of all it only cost €60/$80.

Ian

share|improve this answer
    
Have not heard of De-Verb before. I'll be sure to grab the demo when I get home tonight..could be a lifesaver! –  VCProd Apr 12 '10 at 14:17
    
De-Verb sounds like it's exactly what I need. Thanks for the tip! –  Nick B Apr 13 '10 at 6:03
    
I've always heard that once you have natural reverb in a recording, you can't take it ou. I'm gonna check out this De-verb thing. Thanks for posting! –  Hubert Campbell Aug 13 '10 at 19:28
    
It worked really well, firstly it really took the edge off the reverb during the dialogue edit stage but once we got to mix stage it sounded great, you'd never know it was once there after you added the music and fx. –  ianjpalmer Aug 13 '10 at 21:21

Depending on your budget there are several noise reduction tools you can try:

  1. Z-Noise from Waves. Does a great job at a reasonable price but the algorithm doesn't hold up very well against too much reduction (ie. your dialog will start to sound "underwater")

  2. WNS from Waves. More advanced than Z-Noise, approaching the quality of a CEDAR. Is very impressive, but beware dialog pumping from too much processing.

  3. CEDAR. It's what the pros use. Very expensive, but if you can borrow one for a while or if you know someone who owns once and can do a test-pass on some of your material it would be worth your while.

UPDATE: A friend of mine recently turned me on to SPL De-Verb. It works really well at reducing…you guessed it -- reverb! It's a fantastic one knob solution that surely does the trick.

Best of luck!

share|improve this answer
    
Thanks, I am a loyal user of z-noise and soundsoap 2 pro, Have used a cedar box before but no access to it now. Unfortunately it's reflections I'm trying to eliminate, and I've never been successful removing them with any NR plug I've come in contact with. That said, I haven't ever used WNS. Thanks for the help. –  Nick B Apr 13 '10 at 6:06
    
Hi Nick - OK, I get what you're talking about now. Sorry my initial advice wasn't really geared to your exact question. Since you're familiar with how a Cedar works I wouldn't bother checking out WNS since it serves basically the same function but for less money. I'm thinking EQ is pretty much your only option on this one. –  Jay Jennings Apr 13 '10 at 6:44

Izotope RX has saved many a project around the office.

share|improve this answer
    
Love RX, and it's certainly saved me many times too. Especially that de-clipper I mentioned above. But don't have the time to go in on a "spectral analysis" level to remove tiny reflections. The funny thing is, director thinks it sounds fine, and it's me who's obsessing. I just can't stand poorly recorded voice overs! Thanks for your thoughts. –  Nick B Apr 13 '10 at 6:18

try EQing out the resonances of the room...Grab an EQ 3 , and with a notch selected with very low Q and high gain you start moving the frequency until you get a strong resonant band, then cut it in the 6dB range...and so on with all troublesome frequencies.

share|improve this answer

This is probably a bit too late, but I absolutely LOVE Reaper's ReaFir for noise reduction.

share|improve this answer

It sounds like you're trying to get rid of echo, not noise specifically.

Echo Other than clipped audio, it's one of the hardest things to fix. The reflections are coming in at the same frequency as the vocals, so you can't EQ them out. Try setting a noise gate pretty agressively so you can get just the higher level vocals, not the reflections.

Noise For noise reduction, as birdhousesound has pointed out, there are lots of very nice tools. Most of what I come across I can fix with Adobe Audition 3.0's built in adaptive noise reduction. It's a great tool bundled with a reasonably priced software. Adobe's help files have good instructions on how to avoid the warbly/underwater sounds.

Can you possibly post a short clip so we can hear what you're hearing?

share|improve this answer
    
Hey thanks for the reply. Tried gating but there's just too much and things would sound too choppy. Can't post a sample just yet, unfortunately, but you're right, what I meant by room is source reflections, essentially, echoes. And yeah, they're tough...that's why I tried to have the engineer not capture them to begin with. For clipped audio you should try Isotope RX declipper... pretty incredible. –  Nick B Apr 13 '10 at 6:09

I second VCProd. My experience is little however I worked on a 15 minutes movie with roomy sound, and I went about it as follows:

  • I first EQed out critical reverberation frequency. If you have a not-too-long-to-come reverberation (which is the case in normal size rooms as opposed to cathedrals), the frequencies the room accentuates will be heard as the talent is emitting sound. What happens? For this particular freq, well there is twice more! The amount of freq from the voice plus the reverb. You can tame that down and you'll lose some echo without impairing the body of the voice too much.

  • you then expand. Set your expander so that it glides down at the end of the words of the character. If I was to paint what I'm trying to explain, I'll paint the side of a mountain with a person and a kite gliding down, 3 feet off the ground, accompanying the shape of the terrain all the way down. Well, the mountain is the waveform of your voice, and the kite is your expander accompanying the waveform in its rises and its falls from very close.

The production tracks I got were hyper roomy and I tamed them! The funny thing is that I was just suggested to add a little bit more reverb. Well, I'll just have my kite flying a little higher :)

share|improve this answer
    
Yeah, did what I could with eq already and it helped a touch, but not enough. I think gating is what you're getting at, rather than expanding (essentially the same tool though, of course). Unfortunately, gating just doesn't do justice to this particular narration. Thanks though! –  Nick B Apr 13 '10 at 6:16

Seeing as you can't re-record, I think a noise expander / gate is your best option. Someone mentioned earlier to set it up so that only the loudest parts get through, which is absolutely correct. It takes a bit of trial and error though.

For starters, some parts are inherrently going to be louder than others so one gate setting will not work. What to do here is adjust the amplitude of each word so that the average is approx the same. Do NOT use compression for this because it will squash your peaks down closer to the level of the unwanted signal and you'll find it harder to make the distinction with your gate later. Once your peaks are uniform, you should be able to find a general gate setting which cuts the reverb/echo from the tails of each word. An expander with quick transients is best for this otherwise you'll find the words "chop in". Also, be sure to set the release so that words don't end too abruptly. This won't suddenly make your recording sound fantastic, but if you do a before and after comparison you will notice at least a 70% improvement. It's a shame that the engineer didn't do a better job to begin with - using a decent expander you really can improve and near eliminate room noise. I use a Behringer INTELLIGATE XR2000 and get good results, but there are newer ones for around £200 which are well worth the money.

share|improve this answer
    
When you say adjust the amplitude for each word, you literally mean raising the volume on the mixing board for the track for each word? Man he's got a lot of work ahead of him. –  Hubert Campbell Aug 14 '10 at 1:04

Agreed with GMatijas. I like to do this (filter out the resontant tones), followed by the C4 in the premix chain with sharp Qs and release times - helps round the edge off of the room. Also, dipping at about 450-500Hz, Q3.5-4.0 with an EQ - that's where the roominess in dialogue tends to sit, traffic drone sits more around 400Hz. It's different all the time, so it way require some sweeping to find it.

I've just started playing with this De-Verb in my chain after the C4, and I must say I am quite impressed with it when used sparingly. So that said, my chain of preference for bringing out dialogue clarity and receding the roomy muddiness is:

  • EQ (notch resonant tones and roominess)
  • wave C4
  • SPL De-Verb
share|improve this answer

I would go with either Izotope Rx, SPL Trancient Designer or SPL De-Verb

share|improve this answer

Another problem you'll be hearing is the actual tone of the room, or the room mode. This is where a frequency being played is the same wavelength (or multiple of) a dimension in the room (W, D or H).

You should find what frequency (or frequencies) excite the room and notch it out with an EQ. This will help a lot with getting rid of the room sound.

The bigger the room is, the lower the room mode will be. I suggest starting low and working your way up until you find the correct frequency.

Other than this, I strongly agree with Ian Palmer's suggestion of SPL De-Verb. However, whilst this plug is great at removing a reverb tail, it struggles with the early reflections, which means you're still stuck with a characteristic of the room, albeit a less intrusive one.

Hope this helps!

Cheers,

Fred

share|improve this answer

After reading this i started editing and mixing a new tv show today and ran into the same problem. There where interviews in gymnasiums and showers with wrong mic with bad positioning. Downloaded the De Verb and played with it for a few minutes. I must say i'm impressed. Always great with two buttons plugs when they work :) Maybe i was lucky with what i tried it on, but the dialogue got a lot closer and easier to understand. I'll probably buy it.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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