I would fake it. Do the noise reduction then go back and add in the room tone/ambience. Even if this means going back to the space when it's empy or before a performance and getting a few minutes of tone. Go ahead and loop it under your track.

they won't know but you will know and if you don't tell.... :D

of course if they find out... :oops:

enn, your solution is interesting, I will try it to see if it is natural. Thanks

To administrator:

Can you please move my post to Acoustic Music section? Thanks

Can you please move my post to Acoustic Music section?

No Prob.

I live in a part of the country where air conditioning is neccesary about 7 months out of the year, and none of it is quiet! I recenlty found the silver bullet after years of trying different plugins (Soundsoap, Soundsoap Pro, etc). After trying the Algorithix Noisefree demo I decided that it was worth the interest to put it on the card and enjoy the near-silence with no musical side effects.

I also bought the "orange" linear phase EQ-- it is like buying a pair of magic microphones. Change to sound to taste but sacrifice nothing in imaging or transparency. In fact, I am tempted to suspect that this could make average mics sound like great mics for a fraction of the cash. But I won't say that!

Rich

wavelength wrote: Hi Guys:

I recorded a live recital (piano / soprano) last Friday, I used Blunmlein stereo mic. I've got amazing sound but also a lot of audience noise (about 180 people), I then tried to use various noise reduction plugins, It worked but it also took the ambience as well. I doubt if other classical live recordings use the noise reduction facilities at all, can anyone tell? Also how do you do if you have the same situation?

Having recorded over 2500 live classical concerts in my life (with audiences from 200 to 2000 people) I have never had a concert where too much audience noise was a problem. What was the audience doing that was so noisy? If you are recording live concerts the audience noise is usually coughs and the occasional wrongly timed applause but these are normally fairly easy to edit out. Was your concert different?

Live recordings are never going to sound like a studio recording and that is why they are so appealing. I would be curious to know why the audience noise was such a big factor in your recording?

I personally would never try and take out audience noise with a noise reduction plugin as it would change drastically the overall sound of the recording and trying to put the ambience back later is not the way to go especially with classical recording. I always go for the live as it happened recorded sound and not the sterile recording that sounds like it was done in a studio without an audience.

Just curious as to why you think it necessary to eliminate the audience noise and what caused it to be so noisy in the first place.

Now if you are talking HVAC noises or filament hum from the lighting instruments or rumblings from the building then I would try and eliminate as much of it as I could WITHOUT changing the overall sound of the recording. I believe very strongly in the "do no harm to the music" school of post production and would not eq or add reverb or noise reduce a classical recording unless there was some really strong need for it of if it was the only way to save a live recording due to unforeseen events that happened during the concert and I would seek the guidience of the performers if it had to be done.

MTCW and FWIW

FWIW: I always try and get about 10 seconds of ambient noise before the audience shows up....then during post, flip the phase of the noise sample, and blend a bit into the stereo mix ... not so much that it totally cancels, but enough to greatly reduce it.

Also, using Samplitude's Object editor, one can dramatically reduce the impact of background noises, spontaneous audience noise, etc...

wavelength wrote: Also how do you do if you have the same situation?

Sunday afternoon i recorded a great concert in St Sernin church in Toulouse (France) on the big Cavallé Coll organ, for the organist. The concert was free, so there was a lot of people walking around, talking, eating..., visiting the old romanesque style church... while others tried to listen Messiaen, Tournemire and Vierne. My omni mics captured all that and the very natural recording is as the reality : the pianissimo in Messiaen are very disturbed by the ambiant noise. But the "plein jeux" in Vierne don't worry about the problem !
I can't imagine trying to remove the noise without disturbing the great imaging of the recording. So the organist will have the complete witness of his concert and we'll give only the loud tracks to the radio... what a pity !

[quote=Midlandmorgan]FWIW: I always try and get about 10 seconds of ambient noise before the audience shows up....then during post, flip the phase of the noise sample, and blend a bit into the stereo mix ... not so much that it totally cancels, but enough to greatly reduce it. quote]

You lost me here... ambient noise recording at one point in time should be basically uncorrelated with the ambient noise recorded any any other random point in time. So how would flipping the phase and blending it reduce the noise at all? Seems like adding two uncorrelated noise sources should increase the noise in the final result.

Or are you using the ambient signature for some kind of frequency-based noise reduction algorithm?

What am I missing here?

Michael

Like Tom, I follow the audio hippocratic oath: "First do no harm." This holds true for noise reduction as well, I would think.

Random noise (audience shuffling, wind, rain, far-away door slams, footsteps, occasional coughs, sneezes, etc.) doesn't usually require too much removal, unless it's something really bad in a critical moment. Static noises like HVAC/air handler sounds, fixed pitches, and even cell phone jing-alings are often fair game for removal, if it can be done without harming the music. (Rant: As "ambient" as baby-noises can be, that's one incursion that annoys me greatly. I"m a parent and family-oriented, but I think uncontrollable children & hungry babies have no place in a serious music concert, and for the life of me, I cannot understand how people let their little darlings ruing such beautiful moments for other people.)

Samplitude/Sequoia has some pretty extensive noise reduction tools, and I've been lucky enough to get some pretty impressive results, often a layer at a time, carefully removing/extracting things as I go, like hum, buzz, dimmer noises, etc. The trick, however, is knowing when to stop, or how much/how little to remove without damaging the "Good" audio. (And how do we decide what is "Good", as well?) As soon as I hear any kind of change in the audio for the worse, I back off the NR and re-assess.

For really pesky stuff, I have also begun using Algorithmix - in my case it's the reNOVAtor plug in tool, and it's nothing short of astounding when it works. It doesn't fix everything, but wow....when it works, it's just fantastic. A short explanation is kind've like using Photoshop's "clone" tool....highlight an area and crop/wipe it out, using information around it. (In this case audio data info immediately before and after the sound you want to remove.) It's a spectral analysis tool, and you can zero in on a very tight area, if nec., as well as larger tones, based on time (left to right) freq (top to bottom) and amplitude (colors).

As for flipping phase on a noise sample and some other things like that, I'm not sure I'd go that route, but if it works.....it's worth a shot.

Midlandmorgan wrote: FWIW: I always try and get about 10 seconds of ambient noise before the audience shows up....then during post, flip the phase of the noise sample, and blend a bit into the stereo mix ... not so much that it totally cancels, but enough to greatly reduce it.

Also, using Samplitude's Object editor, one can dramatically reduce the impact of background noises, spontaneous audience noise, etc...

Not sure how this would work. Could you be more specific? 10 seconds of noise with no audience present mixed back in with the ?whole program? equals ? Is this your way of getting rid of amblient noise such as AC and building noises? Not sure what you are accomplishing but maybe I don't understand correctly all that you are doing. Thanks!

JoeH

I too wonder why people bring small babies that are prone to crying to public events. We did a recording of a Children's concert recently and the person with the baby sat right under out main microphone set up. The baby cried for most of the concert. Finally about two minutes before the intermission the mother got up and moved to the foyer. When the second half of the concert started she was back in her seat under the microphone with the crying baby AGAIN. We have a 90 minute concert with a baby crying for maybe 45 of those minutes and right under the microphones. Since our microphones were located in the first row of the concert hall and since we had our yellow ropes up the mother had to disassemble the security in order to sit where she did. I talked to the director afterwards and she was NOT PLEASED that someone would sit in the first row with a crying baby for most of the concert. I think their will be a posting in the next program suggesting that people with very young children sit in the back of the hall or STAY HOME! This was the rudest person I have ever seen at a concert (well maybe the person that was carrying on a 10 minute cell phone conversation in the middle of a live opera recording might be just as rude) In any case crying babies are hard to remove from the overall mix so my NUMBER ONE suggestion is to HIRE A BABY SITTER and leave the little ones AT HOME!

MTCW

Sorry for not being specific....

All of the venues I do these live orchestral type recordings are relatively old...in some of them, air conditioning was retrofitted (please keep in mind I'm in West Texas, so it do get HOT!) with poorly placed vents, loud blowers, etc...And, these things run nearly 24 hours a day, so that ambient noise is ALWAYS there. More times than not, there are just as many vents/blowers on the performance stage as there are in the audience... This is what I try and eliminate with the phase reversal technique of which I spoke earlier. This works well for any ambient noise that is constant...but as you all pointed out already, doesn't do squat for random sporadic things...

Again, sorry for any confusion. Most places I record remotely are not what anyone would consider acoustically wonderful...

Audience noise - I hear you! Especially here, where the average classical audience is 60-years-old+, coughing, hawking and "loud breathing" is an overriding problem during the concerts but also extremely audible on recordings, from the most delicate pianissimo to the loudest fortissimo. They just don't seem to care.

I once suggested our conductor to distribute free Vicks or cough lozenge before the concerts. Don't know if it would work, but it's worth trying!

Where I grew up in Wales, there's a tradition of giving people nicknames according to their job or personal characteristcs; Dai Bread would be a baker called Dai, or Thomas the Milk a milkman called Thomas (or Mr Thomas), etc.. Following this tradition, I, and many other engineers I know with a Welsh background, commonly refer to our old concert going friends, Mrs Jones Bronchitis (the old lady who goes to every concert and sits under the mics coughing and spluttering), her friend, Thomas the mint (an old gentleman who sits next to her unwrapping boiled sweets or mints wrapped in noisy celophane and kept in a rustling paper bag and Davies the Dust (an ex miner suffering from respiratory problems which make him wheeze and puff interspersed with apologies delivered in a force 10 whisper to people around him - "Sorry boy, I got a bit o' the the dust, see"). Each of these is prone to deafness, so they "whisper" in a voice loud enough to be heard at the back of the balcony, and inclined to fall asleep, snoring, in the quieter/slower musical passages!

On Ellegaard's point, I regularly record a festival where, at every door to the main hall, there's a little stand supporting a bowl of revolting tasting herbal cough lozenges for people to take as they enter the hall. The hall does seem to be a little quieter than most I work in. I've no idea whether that's because the festival is in late summer when there aren't so many colds and coughs around (apart from hay-fever sufferers) or because the sweets are working. They might be acting as a deterrent, people being terrified of coughing in the hall in case someone forces them actually to eat one of the foul tasting things!

Being slightly more serious, the acceptable noise threshold obviously depends upon the nature of the recording. For classical concerts, a silent background broken only by appropriately timed applause seems to be the preferred norm; for jazz, especially live recordings in jazz clubs where people are eating and/or talking right next to the stage/performers it's a different matter. A certain amount of background noise is unavoidable and as long as it doesn't get in the way of the perfomance I don't have a problem with it - a bit of atmosphere and audience reaction is something I like. I find that in most classical halls/church concerts/etc., the worst excesses of audience noise can be tamed by careful micing and perhaps a polite announcement or two before the performance. Most people seem to have no idea that they can be heard; almost every time I've mentioned it to an audience member they seem genuinely shocked that it's a problem. There seems to be a common perception that microphones only pick up the music and that a mic that isn't obviously pointed at someone can't "hear" them. It's a little frustrating having to compromise even a little the mic setup and overall balance to help produce a reasonably quiet recording but I find that slightly less than my preferred balance can be preferable to a great sound that's beautifully recorded with every sneeze or cough - or at least that's what my clients seem to prefer.

Having said that, I've occasionally resorted to "cleaning up" recordings. One of my major clients is a Classical FM radio station which uses a "broadcast processor" (they hate me calling it a compressor or Finaliser!) on it's station output. I've taken to post processing (a bit of careful EQ and level adjustment) their broadcast masters in the quiet sections of some concerts to reduce the worst effects of the compressor binging up levels (by up to 25dB or so) and with them, traffic noise, aircon rumble, etc.. I'd rather I messed about with levels in a humanly controlled way so that the machine doesn't have to do as much in it's unthinking and uncaring way. Either way, the listening copies that are sent out to artists/sponsors/etc. are left alone - unless there's some particularly bad noise problem to fix.

Since the development of CEDAR's ReTouch (which so impressed me that I actually bought a Sadie in order to use it!) I've been able seamlessly to remove things like coughs, chair scrapes, bumped mic stand noise and even police sirens(!). I don't do too much work on a normal concert - I'm not trying to make it sound like a studio recording - but having the option to tweak it later does give me more leeway on micing and can save a really badly timed sneeze/siren from destroying an otherwise atmospheric recording I generally find that conventional de-noising processes are of limited use on live recordings - unless there's a steady state noise like aircon rumble to remove - most of the problems I get are with random noises like coughs or external noises like sirens, aircraft noise or taxi brake squeals leaking into the building and the average noise fingerprint based de-noising process does little to help with these.

The main problem I find with most de-noising processes on live concerts is that, by the time they're doing anything really useful to the background noise, they're also screwing up the sound of the music. I'd rather hear some background noise, much of which, after a few minutes, the brain naturally filters out, than have a near silent background and a recorded sound full of processing artifacts and digital wierdness. I guess it's just another case of listening carefully and less is more.

Ellegaard wrote: Audience noise - I hear you! Especially here, where the average classical audience is 60-years-old+, coughing, hawking and "loud breathing" is an overriding problem during the concerts but also extremely audible on recordings, from the most delicate pianissimo to the loudest fortissimo. They just don't seem to care.

I once suggested our conductor to distribute free Vicks or cough lozenge before the concerts. Don't know if it would work, but it's worth trying!

When I worked at the Conservatory we had an older gentlemen who use to come to all the concerts of a certain type and he would cough in only the quiet places. I got so use to him being there that the concerts did not seem "real" if I could not hear his coughing. Suddenly the coughing stopped and I wondered if he was still with us or if he had gone to be with his maker. I asked around and found that he indeed had died of Emphysema and that he had been a heavy smoker for most of his life. I also found out that he was a voice teacher in the Conservatory but had retired about the time I came on board. I felt sorry for his passing and somewhat missed the coughing that made the concerts seem more real to me. It was like an old friend (who I never met) had passed away.

FWIW: I always try and get about 10 seconds of ambient noise before the audience shows up....then during post, flip the phase of the noise sample, and blend a bit into the stereo mix ... not so much that it totally cancels, but enough to greatly reduce it.

Sorry, but this technique just can't work for most ambiant noises that are talked about here. When you add two uncorrelated signals their energies sum up, even if you shift the phase 180°. In fact you just add noise to your mix, that's all.

However, your point about taking a sample of this background noise is excellent and this approach can make the noise reduction plugins work much more efficiently than with the standard settings. FYI, most of these plugins work in the frequency domain (as opposed to the time domain).

Best,
Thomas

Well, it can work to a degree. As you state, it's a great way to get noise reduction plug-ins to work FAR better. However, say an AC unit gives of a 42 Hz rumble. This is not random, uncorrelated sound, rather a very specific sound pattern, one which can be easily cancelled. However, the Brownian noise (the sound of air molecules colliding in space) are uncorrelated and relatively random (which if you study random, you know there truly is no such thing). Therefore, you'll make a slight difference in using this technique.

However, I've found both this technique as well as every noise reduction program I've ever used to impart sonic problems that I just can't live with.

Here's my solution to simple or even mildly aggressive noise issues:

1st - understand that on a "live" disc, noise is to be expected, even by your most picky client.

2nd - be aware of the fact that, during passages containing regular program material with a good selection of frequencies that registers say -25 dB or higher, most noise will be masked and is often best left alone.

What I like to do is simple - say there's a violin solo, such as in Scheherazade, where nothing else is going on. On top of this, you add air handler noise. Most of the noise that you are wishing to remove falls well below the spectrum of said solo violin. I take a gentle crossfade into the solo and put as aggressive of a high-pass filter as humanly possible without altering the sound. Depending upon the instrument, I'm occassionally able to reach HPFs of as high as 220 Hz. Most HPFs though factor in at around 100 or so Hz and impart nothing on the sound except for getting rid of the pesky air handler. The noise which falls within the same frequency spectrum as the solo has two issues.

1 - it will be severly masked by the program material (this is the bonus)

2 - it will be noticeable if you attempt to remove it b/c you WILL get artifacts.

In my humble opinion, it is, in almost every case, unacceptable to remove the sound of the hall using a noise reductin plugin and replace it with reverb. This simply breaks every rule I follow for natural recording.

Sometimes, you just have to live with a little noise. Anyone in attendance at the concert will be well aware of the noise that was there and will be reminded of it on the disc that they purchase.

Of course, I also make it perfectly well known in advance to those in charge if I am hearing excessive noise from an air handler, etc. Of course, old deaf dudes in the audience can't be helped.

Okay, 2 quick stories here - one funny, one tragic.

1 (funny) - I recently recorded a rather major orchestra at a live event. There was apparently a nearly deaf and blind fellow sitting in the second row, center. He had apparently come with someone dear to him, as he kept LOUDLY asking the same question -- "What's happening now?" To which she would reply - "The musicians are playing beautiful music" equally as loudly.

2 (tragic) - I do a lot of recordings in High School auditoriums here in the area. Unfortunately, many of these auditoriums are made as multi-purpose halls and very little is done in the way of acoustics for these halls. There are a few which are exceptional and really do work wonders for just about any group.

There was recently a school constructed which was touting the fact that their auditorium was to be of superiour acoustic design. IT IS! It's a wonderful hall.

However, here's the problem. The A/C contractor apparently didn't give a damn about the acoustics. During the winter, when the heat is on, the blowers at peak volume reach 82 dB at the center mic array. Yes, you read that absolutely right!!!!!!!! 82 dB!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

During the summer, with the AC on, it's far more tame and only registers at maximum 71 dB!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Here's the great part - it's all handled by a centralized system over 50 miles away and cannot be turned off under any circumstances.

UGGGGGGGGHHHHHHHHHHH!!!!!!!!! :evil:

J..

Cucco wrote: However, I've found both this technique as well as every noise reduction program I've ever used to impart sonic problems that I just can't live with.

Then you should try the Algorithmx Noisefree demo and be pleasantly surprised!

Rich

Great post Cucco: I agree with most of your analysis and approach, and had a good laugh with your stories. But sorry, the aforementionned technique just can't work.

2 posibilities:

- if the noise sample and the noise in the recording are uncorrelated, the noise will just sum up. There is nothing to do about it.

- if the noise sample and the noise in the recording are strongly correlated (like a rumble, or a high pitch due to the lights), additional noise and unwanted artifacts are still introduced by the 'technique'. Take the example of your 42Hz AC rumble. What you will get with the time substraction technique is the very same rumble modulated at a very low frequency (depending on the stability of the 42Hz pitch). In the worst case you multiply the rumble by 2.

My apologizes for the technical rambling, but I though important to mention this is no noise reduction technique at all.

Best,
Thomas

Hey Thomas -

No need for apologies. You are absolutely right. I wouldn't suggest that technique for most anything. However, if you do have a strong fundamental noise and nothing seems to get rid of it, it might be worth introducing additional noise in other bands if it helps to rid the problem. Again, me personally, I'm usually willing to live with the noise so long as the program material can be heard.

The clients who are usually the fussiest about noise issues, already play in halls where this isn't too much of a problem. It's one of the situations where a paradox works to our advantage...

Rich,

You're right, I haven't tried the Algorithmix stuff. I'm afraid of the price tag. Strangely though, while I was looking at Sadie stuff, I did use the Cedar No-noise and still found it a little lacking. Perhaps I'll get in touch with Algo and see if I can demo some of their neato stuff. I hear their EQ is to die for too - beats me saving my dough for the Weiss I've had my eye on.

J.