Noise Removal Advice (with samples)

[The_T]

Hey there. I recorded some vocal audio commentary with multiple people in a room that was not great for recording it, and so consequently there is a ton of noise, some people are quieter than others, etc. The latter I can work with and mess with later, the main issue I am having is…

No matter how I attempt the noise removal, the voice still carry a hint of “boomy-ness” from the room. I hoped that I could post some samples here and get some advice on what the best sort of settings I should be using.

Thanks for anything you can do!

The first sample has the noise sample I used as the noise profile. It has some controller clacking in it that… ultimately I can just noise gate out if it’s not completely removed so that’s not a big deal; should I avoid including controller clacking in the noise profile?
https://www.dropbox.com/s/pquzgghpcxtghv0/test1.wav

The second sample has a lower voice level that I want to make sure, in the final product, the words are at least still understandable (“Are we going to sync what”), even if the quality on them has to be lower because of their low volume.
https://www.dropbox.com/s/exj2kcte1ff8zg5/test2.wav

The third sample has some booming that… I’m not sure, either someone hit the table the mic was on, or something, but. I figured I’d include it (as well as the audio that follows as an additional sample) to see what the experts can suggest on that as well!
https://www.dropbox.com/s/rrrt0lld0mw1sz8/test3.wav

[kozikowski]

I got #3. That’s a straight noise removal job.

Use the segment of room tone at the beginning of clip wav3 for a profile and then apply it according to the settings in the illustration. Also: clip before and after. A common problem people have with Noise Removal is trying to make a “Studio Recording” out of a noisy room. That’s pretty much doomed. The best you can do is gently reduce the noise which should not damage the show at all.

Koz

[kozikowski]

Getting rid of the clunk is much more interesting. You’ll have to edit that out either by selection and muting which will sound odd, or edit it out and put a segment of room tone back in. Since people are talking during those sounds, you’re dead.

Koz

[kozikowski]

Same thing with wav2. Select some room tone by itself and use that for the profile step in Noise Removal. Juggle that first setting, Noise Reduction for tradeoff between quality of voices and background noise.

http://manual.audacityteam.org/o/man/noise_removal.html

You can do sync the same way Hollywood does. Stand where the camera can see you and yell “Camera Sync” and clap your hands. That gives you a visual and aural, easily discernible, sync point.

Koz

[kozikowski]

“…I wanna wander about.”

There’s always a solution. You can do that with lavalier microphones and radio packs. You can wander wherever the radio receiver can receive your signal.

Of course that’s not cheap…

Perhaps the lady can be convinced to stay in one chair with something tasty from the buffet and a large glass of red wine.

Koz

[kozikowski]

multiple people in a room that was not great for recording

Nobody thinks about this before the fact, but if you continue doing that, it means you need to go through the Noise Removal and patching exercise every show. Editing may already add significant time to the production of the show and you’ll have Removal on top of that.

And that’s assuming Removal works to the quality you’re after.

Nobody in Sound liked the room style idea of bare floors and walls and simple, severe furniture. Those rooms are a nightmare for recording.

http://kozco.com/tech/audacity/clips/EchoSample.mp3

We can’t get rid of echoes. Sorry.

Koz

[kozikowski]

People have been known to shut down air conditioners and refrigerators during recording. You’ve seen those big Hollywood soundstages that look like you could land a small plane in there? When the director yells Roll!, the building air conditioning shuts down to zero. It’s dead quiet when he yells Action!

Koz

[kozikowski]

Wav1 has several problems. I posted a picture of all the overload points in the show. All those red bars are places where the the voice signal got too loud. It overloaded the sound channel and you’re stuck with the crunchy, harsh sound. Sorry.

The sound clip is Effect > High Pass Filter: 100Hz, 24dB, and noise reduction as per the above posted settings. I think that’s good to go for the presentation. It’s easy to ignore whatever is left. You may want to juggle the settings.

Listen to the whole clip before you decide what to do. The first part of the clip is not at show volume.

Koz

[The_T]

@kozikowski

I got #3. That’s a straight noise removal job.

Use the segment of room tone at the beginning of clip wav3 for a profile and then apply it according to the settings in the illustration. Also: clip before and after. A common problem people have with Noise Removal is trying to make a “Studio Recording” out of a noisy room. That’s pretty much doomed. The best you can do is gently reduce the noise which should not damage the show at all.

At first I want to say that, to me the results on that are… definitely not enough. The commentary audio is going to be overlayed over the gameplay audio and having so much dead noise would be rough. On the flip side, I guess I could try noise gating it and seeing if the voice parts still sound okay. But ultimately I still think I MIGHT prefer it being a little stronger?

@kozikowski

You can do sync the same way Hollywood does. Stand where the camera can see you and yell “Camera Sync” and clap your hands. That gives you a visual and aural, easily discernible, sync point.

Not that I really needed any advice on this (everything’s long since been recorded, audio and video) but there’s no camera involved. It’s a video game. The easiest way to sync is just to say “1, 2, 3” while moving the cursor from one menu option to the next as you count.

@kozikowski

“…I wanna wander about.”

There’s always a solution. You can do that with lavalier microphones and radio packs. You can wander wherever the radio receiver can receive your signal.

Of course that’s not cheap…

Perhaps the lady can be convinced to stay in one chair with something tasty from the buffet and a large glass of red wine.

Once again, video game. She was sitting perfectly still, she was wandering around in the video game. Comments like this come off pretty creepy, really.

@kozikowski

multiple people in a room that was not great for recording

Nobody thinks about this before the fact, but if you continue doing that, it means you need to go through the Noise Removal and patching exercise every show. Editing may already add significant time to the production of the show and you’ll have Removal on top of that.

Nobody in Sound liked the room style idea of bare floors and walls and simple, severe furniture. Those rooms are a nightmare for recording.

We can’t get rid of echoes. Sorry.

Well there are only 3 (hour+) clips (to note: all 3 of the above test wavs are from the first hour+ clip) to do and being able to use one noise removal setting across all 3 (or in the very least one setting across each one clip) would be preferred, so I kind of assumed doing it “every time” just came with the job. That said, I don’t exactly know how to make a living room into a better recording space. Electronics have to be around; at least one PC is stored in another room at least. But air conditioning has to be on or else we’d be dead from the heat.

@kozikowski

Wav1 has several problems. I posted a picture of all the overload points in the show. All those red bars are places where the the voice signal got too loud. It overloaded the sound channel and you’re stuck with the crunchy, harsh sound. Sorry.

The sound clip is Effect > High Pass Filter: 100Hz, 24dB, and noise reduction as per the above posted settings. I think that’s good to go for the presentation. It’s easy to ignore whatever is left. You may want to juggle the settings.

Listen to the whole clip before you decide what to do. The first part of the clip is not at show volume.

To me, this one still sounds pretty bad. You’re probably going to think I’m insane for it, but for past recordings (with a single-user microphone rather than a group one) I would always Normalize, Noise Removal, and then Compressor the audio to get a consistent volume level for everything and deal with any clipping that occurs. I basically planned on that for this, but I uploaded the unaltered clips in case I could get any better advice.

I’m probably going to try those noise removal settings and then use a noise gate and see how it sounds, but I’m already thinking that having as little air-yness as possible would be the most preferred thing, from past experience…

[kozikowski]

There’s only so much you can do with a noisy room or a recording made in a noisy room.

In Noise Reduction, increase the first number (Noise Reduction (dB)) by two or three every jump until you notice the voices turning honky or bubbly. That’s the limit. Remember the Profile Capture step is all important. If you get a tiny snippet of voice or other manual activity (shuffling paper) in the profile by accident, Noise Removal will try to remove that from the show in addition to the air conditioner. Also try increasing the sensitivity control.

Noise Gate may not be much help. Have you tried it? It’s very sticky to get a balance between suppressing the background and passing the show undamaged. The first time somebody comments with a quiet voice, Noise Gate will remove them or mute some of the words.

Comments like this come off pretty creepy, really.

We can’t see what you’re doing.

I recorded some vocal audio commentary

Doesn’t say anything about a game. My mental image is someone actually wandering around. Food and wine is usually a good way to nail someone down.

You have an aggressive, hostile recording environment. If I was going to shoot this, I would put a wired lavalier on each person to get the microphone as close as possible to each voice, and even then getting rid of machine noise and air conditioning would be a challenge. If it was important, I’d use headset microphones like the ones used in the later TED talks.

I have used an AKG C555L headset microphone which picks up good quality voice with little or no room noise.
http://www.sweetwater.com/store/detail/C555L?utm_source=Google&utm_medium=PPC&utm_campaign=livesound&device=c&network=g&matchtype=&gclid=CODK7I3qp70CFUqUfgodE2gAGA

It’s just a microphone, but it does require a mixer. Three (or more) of those would be perfect for this kind of show.

with a single-user microphone rather than a group one

I’m still foggy how you were doing this. Someone snuggling up to a single microphone can do well in a noisy room. Distance between the mic and voice is a big deal, so trying to use one mic between three people would be very noisy. The other possibility is you lucked out the first time and the solution doesn’t scale up. There’s a sticker shock between one and three microphones. Suddenly you need a mixer, etc.

There’s the speaker thing, too. You didn’t comment on the powerful “earthquake” rumble on the clips. That’s what I was getting rid of with the High Pass filter. Mixing on poor speakers can be a challenge. If you had a mixer with multiple microphones, you could apply the 100Hz filter in the mixer which would also help with overload.

Koz

[The_T]

@kozikowski

In Noise Reduction, increase the first number (Noise Reduction (dB)) by two or three every jump until you notice the voices turning honky or bubbly. That’s the limit. Remember the Profile Capture step is all important. If you get a tiny snippet of voice or other manual activity (shuffling paper) in the profile by accident, Noise Removal will try to remove that from the show in addition to the air conditioner. Also try increasing the sensitivity control.

Right. What I am trying to do is find the right levels of noise removal and sensitivity to make it sound as clean as possible. I would rather it be closer to the “wonky” side than the “constant annoying air noise”, but not too far on that side. Just enough. But I’ve tried plenty of values and I never feel like it sounds… right.

@kozikowski

If I was going to shoot this, I would put a wired lavalier on each person to get the microphone as close as possible to each voice, and even then getting rid of machine noise and air conditioning would be a challenge. If it was important, I’d use headset microphones like the ones used in the later TED talks.

I have used an AKG C555L headset microphone which picks up good quality voice with little or no room noise.
http://www.sweetwater.com/store/detail/C555L?utm_source=Google&utm_medium=PPC&utm_campaign=livesound&device=c&network=g&matchtype=&gclid=CODK7I3qp70CFUqUfgodE2gAGA

It’s just a microphone, but it does require a mixer. Three (or more) of those would be perfect for this kind of show.

$149 is an insane price for a microphone. Not even mentioning that it would have to be 4x that and the price of a mixer.

@kozikowski

I’m still foggy how you were doing this. Someone snuggling up to a single microphone can do well in a noisy room. Distance between the mic and voice is a big deal, so trying to use one mic between three people would be very noisy. The other possibility is you lucked out the first time and the solution doesn’t scale up. There’s a sticker shock between one and three microphones. Suddenly you need a mixer, etc.

If I’m using a headset, I’m recording one person. If I’m using a group microphone, I’m recording a group. Sorry if that wasn’t clear. I’m not going to try recording a group with one headset.

@kozikowski

There’s the speaker thing, too. You didn’t comment on the powerful “earthquake” rumble on the clips. That’s what I was getting rid of with the High Pass filter. Mixing on poor speakers can be a challenge. If you had a mixer with multiple microphones, you could apply the 100Hz filter in the mixer which would also help with overload.

Unsure what you mean by “speaker thing”. The only rumble I pointed out was in the third clip, and I’m going to assume it was someone hitting the table (after I repeatedly told him not to…). The high pass filter you did was on the first clip, so I’m not… understanding.

[kozikowski]

But I’ve tried plenty of values and I never feel like it sounds… right.

It never sounds like the studio you should have been recording in. Velvety quiet backgrounds are terrifically difficult to achieve. Past the obvious pounding air conditioning and gamer computer fans (now that we got that straight), you have the noise that some microphones and electronics are making internally. That’s straight “rain in the trees” hiss and that is impossible to remove.

Did you try Steve’s Noise Gate? I think he wrote it.

$149 is an insane price for a microphone. Not even mentioning that it would have to be 4x that and the price of a mixer.

Actually, that’s a medium price and the mixer can be $120 USD.
This is a broadcast radio show I shot. See that gold thing in the middle?
http://www.kozco.com/tech/audacity/pix/JMASoundShoot.jpg
That’s one of these:
http://www.sweetwater.com/store/detail/C414XLII/

They didn’t have to do any effects, noise removal or cleaning to my side of the show. It was cut and went straight to air. The conference room is soundproofed and please note the double furniture moving pad on the desk. Every little bit helps.

That mixer is one of these:

http://www.sweetwater.com/store/detail/PV6/

I’m using the analog digitizer in the Mac. The USB version of the mixer would be a bit more.

If I’m using a headset, I’m recording one person.

And that’s the one that sounded pretty good, right? You have good, short, stable separation between your lips and the microphone pickup (or tube depending on the headset). Even I could make that work. It’s when you tried to do a group that it fell apart. Open air microphones with multiple people in a noisy room takes a different skillset. I can’t do that. I’d be making calls.

You did miss one configuration. A gamer headset on each person. I built the analog adapters to do that with a mixer for a friend once. I can’t do it for a USB headset.

There’s also the NPR ‘This American Life’ technique. He pretty much pioneered using a long-distance shotgun microphone up close for interviews. We all gasped in horror, but those people turn out very high quality interviews like clockwork.
Attached is a similar interview. The microphone is in the lower right. I think it’s one of these.

http://www.sweetwater.com/store/detail/NTG1/
… in its windsock (right-hand illustration).

The only rumble I pointed out was in the third clip

That wasn’t so much rumble as somebody giving the microphone a good smack. Rumble is the metrobus or dump truck going by — or an earthquake. Or your air conditioner. Can you pick up your speakers with one hand? That’s usually the sign they’re not up to critical mixing. And you are critical mixing.

Well, it’s early Sunday morning and my web site has gone down, so it’s time to go to bed.

Laterz,

Koz

[The_T]

I don’t know what “game computer fans” you’re talking about. While the game console is in the room, it’s quiet compared to the air conditioning. The laptop is also quiet. The main recording PC is stored in another room to avoid issues with it’s noise.

I don’t see how trying to insult my financial stability is going to affect my ability to process these recorded videos that are done already and still need to processed.

I don’t hear any “rumble” at all. I hear the air conditioner and other dead air noise, and it’s the noise I’m trying to remove. I would still appreciate help on this, but it seems like you want to move the conversation completely outside of that…?

[kozikowski]

I don’t hear any “rumble” at all. I hear the air conditioner and other dead air noise

It’s there. There’s a very low pitch growl component to the air conditioner that is apparent on larger sound systems. If you can’t hear it, it just means some of your listeners are going to hear it and some not. In any event, you can get rid of it with the 100Hz high pass filter without damaging anything else and nobody will ever hear it.

ability to process these recorded videos that are done already and still need to processed.

When you changed your recording process/hardware, you may have moved beyond your ability to rescue the recording to your satisfaction.

Our attention then shifts to not making any more of these with similar problems. If the software rescue idea fails, then the only other production tool left is a reshoot.

Koz

[The_T]

Please stop making unfunny jokes. If you don’t want to offer any further advice, just say so.

[kozikowski]

It’s not a joke. When the pros decide the current shoot is never going to work, they shoot it again and fix the problems. Koz