Remove childs crying - edit audio track

sheraj wrote on 5/11/2006, 12:40 AM
I am editing a video. While the main person is talking, there is a child crying loudly in the room. Is there any way to cut out/delete/cover up the sound of the child crying from the audiotrack. without affecting the main speakers voice? Can we see separate tracks for each voice on the video ? Is there a way to accomplish this or will the entire video be a waste because I cannot remove the crying ? - I have Vegas Movie Studio Platinum Edition. Appreciate any creative ideas.

Comments

PeterWright wrote on 5/11/2006, 1:20 AM
Because the child crying is probably at the same time as the main person talking, this will be very difficult - you may be able to isolate the frequency of the child and reduce it in Track EQ, or if you have Noise Reduction this may help, but I doublt you'll be able to totally remove it, as it's all on the same audio file.
B.Verlik wrote on 5/11/2006, 1:31 AM
There is no way to delete the childs crying, without affecting the main speakers voice.
There may be a way to help the final project.
Don't worry other suggestions will follow.
This is the Vegas Video forum, so I don't know Movie Studio Platinum at all.
One thing you can do, is place a volume envelope on the audio track (if this is possible in Movie Studio). Then you will have to 'zoom' in on the track and drop the volume to zero or near zero when the speaker is not speaking, and bring it back up to normal when he is speaking. (very time consuming)
You can use an another audio editor, if you have to. This is very tedious work.
I'm sure others will offer advice about noise gates and maybe a different approach to the situation, but there is no way to just erase the child's crying.
The way I suggested, while time consuming, should give about as good a result as you can expect. You may be able to mask the editing with some other background noise (since there will be "holes" in the sound), or maybe try some reverb on the speakers voice.
The Parametric EQ may help cut back on the childs crying frequencies too. You'll have to experiment to find the right frequencies, but probably between 1K and 5K.
Good luck, it won't be easy. All the sound, even if Stereo, is embedded with the room noise, so there are no separate tracks for each person.
You will probably be able to hear some signs of the childs voice, no matter what, but you can improve it.
craftech wrote on 5/11/2006, 2:19 AM
I used to run into this all the time when I shot theatre video because I used to have my mikes back by the camera. Between the little kids crying and the parents trying to coo coo them in a loud voice it was pretty bad. There was no way to remove it in post.

My solution was to move all the mikes to the stage and run a feed. You can still hear the crying, but now it is very faint and perfectly acceptable. I am not sure where you had your mic/mics, but the bottom line is don't mess with it unless it is as loud as the talent.

John
farss wrote on 5/11/2006, 3:12 AM
Couldn't agree more. Basically mission impossible. You MIGHT be able to use a gate to get the crying out between the words but then it's going to stick out really badly that you've fudged something, you'll have the words with room and then dead silence, sound really, really bad.

vicmilt wrote on 5/11/2006, 3:27 AM
Tell us what the movie is - and what the intended audience is, please.
PeterWright wrote on 5/11/2006, 3:36 AM
Whenever you need to reduce volume between words, as Bob says, sudden silence sounds awful, so find a piece of room ambience (with no talking) from elsewhere and put this in the gaps - best way is to render a new wav file and put this on a separet audio track, then drag it so it loops and helps to disguise the drops in main track volume.
epirb wrote on 5/11/2006, 7:11 AM
Ahh another reason/use to have gaffers tape on the shoot.....
seriuosly though, definately use the nat sot if you have it, but agin it might still sound strange if the crying is both during the person talking and the silent periods. Sometimes trying to make it go away is more distracting to the listener than the actual sound.
Does it make the talents words indistinguishable?
sheraj wrote on 5/11/2006, 8:58 AM
The Video is from my childs day care and the intended audience are the parents(for mothers day). While all the children are playing and enjoying there is one child who was crying very loudly throughout. I cannot pass the video as is to the mothers since the mother of the crying child will be upset and concerned.

From the responses I have seen so far, it seems like an impossible task.

I am not very familiar with using Audio Envelopes/Parametric EQ's. I tried last night but found it confusing. Manual not helpful either. Can anyone explain in simple terms(if you have tiime). Thanks
Chienworks wrote on 5/11/2006, 9:24 AM
Can you get in touch with the main person talking? Perhaps you could invite them to recite what they said again, in time with their original speach while watching the video. You could record them and sync the new version to the original video. As has been mentioned before, the difference in sound quality and ambience will be very noticeable so adding in some of the original background noise would probably be helpful.
JJKizak wrote on 5/11/2006, 9:24 AM
There is another way. Get the guy that was speaking and have him repeat the identical words as the video plays and record it. Do it many times until his mouthing matches the words. Then using the equalizer blow his voice and the child voice out of the original clip to retain the baclground noise, then just sync up his new voice track with the old clip.

JJK
johnmeyer wrote on 5/11/2006, 9:27 AM
I restore things, including audio, and have to deal with this sort of thing all the time. You definitely cannot remove the screaming but, depending on the nature of the main audio, you may be able to salvage certain things.

First try to find some audio -- perhaps from B-roll -- of the background noise that matches what you would hear without either the kid screaming or the talent talking.

Second, take this audio and put it on a track below your main audio. Next, find the places between the points where people are speaking, and make a cut on the main audio between the person speaking. Fade in/out the background audio at this point. If you are lucky, there will actually be a few points at which the yelling kid is taking a breath at the point where the talent is talking. Those segments may actually sound pretty good. At the points where the kid is yelling at the same time as the person is speaking, you may be able to adjust your fades to make it sound like a natural, but fairly short squawk. An occasional squawk, rather than continuous wailing, has a different effect on people, and perhaps will be more acceptable.

Of course, if the talent is talking with very few pauses, then this won't work very well. I use this technique all the time to remove coughs, things dropping (usually by the stupid camera operator -- me), etc.

vicmilt wrote on 5/11/2006, 7:35 PM
A few assumptions and a suggestion -

1 - for all intents and purposes the original track is irreparable
2 - it's a daycare video, so we are not really listening to carefully scripted lines
3 - parents will love to see their children, and with good cause - these years pass so quickly
4 - we can assume the video itself is darling, to say the least

It's your responsiblitly as a film-maker to present an entertaining end product. Since we are not trying to salvage a presidential address here, I suggest the following:
Get a cute musical track that reflects what's going on On-Screen. If they are dancing, get some music in a similar rythmn, or mode. If they are acting, get some light humorous music, or some sentimental music. Put that under your video and see how it looks without any of the original ambience. Then write a simple voice-over tribute to the day. Music continues under this Voice Over at all times.
"What a day in the life of the South Central day care. It was filled with laughter and tears, excitement and joyous celebration. Little Timmy Smith sang his heart out, (fade up timmy for a second, crying and all, then out). Patsy Mulligan danced beautifully and the entire chorus line was a delight to watch. Mrs. Smithers watched with delight and was rewarded with the finest performance ever in the history of South Central day care".
You can't lose - forget your original track - it's gone - get over it, and get back to the "real" busines... entertaining and delighting your audience.
best,
v
sheraj wrote on 5/12/2006, 9:00 AM
Thank you very much for all the responses. I have not yet tried 'johnmeyer's suggestion yet. If I succeed in removing the crying, I will let you all know.
Since this a common problem that I face with many of my projects(ie. to remove voices from video) I wanted to make an attempt with my day care project.
JohnnyRoy wrote on 5/12/2006, 9:43 AM
> Since this a common problem that I face with many of my projects(ie. to remove voices from video) I wanted to make an attempt with my day care project.

For next time, you might try getting an MP3 player that also records (e.g., iRiver 700 or 800 series) and plug it into the tape out of the PA system (I assume the speaker was using a microphone). This will give you much better audio.

If there is no PA system or they won't allow you near it, get a Giant Squid lavalier mic ($25) and place it close to the speaker (maybe on a podium) with the MP3 recorder. It should be small enough to be inconspicuous. Maybe even have them wear it (best scenario) and place the iRiver in their pocket. For about $100 investment you will have much better audio.

For a little more money ($180) get an Audio-Technica Pro 88W wireless mic and plug that into your camera and again have the speaker either wear the lav mic or place it close to them while they are speaking. The bottom line is to get better audio by recording better audio rather than trying to make better audio out of a bad recording.

I tape school plays all the time and crying babies were always a problem. Then I wised up and took a feed off of the sound mixer (first with my iRiver 790 and now with my Marantz PMD660) and my audio quality skyrocketed. Getting better audio is well worth the trouble.

~jr
B.Verlik wrote on 5/12/2006, 12:21 PM
I believe somebody here mentioned that by using two identical mics, one for the speaker and one for the room, that they can be placed out of phase, during mixdown, and eliminate background noise. The speaker, being basically on only one mic, will be heard, but all the noise bleeding into both mics can be cut back. (phased-out, so to speak.)
I'm going to have to try this sometime.
Serena wrote on 5/12/2006, 10:52 PM
In principle the 2-mike system could be used to discriminate between wanted and unwanted sounds (our hearing does it all the time), but a simple summing system can't do the job. The phase difference is frequency dependent and people are distracted by even small levels of crying baby. As an aside, a countermeasure against fighter pilot communications was to inject a "crying baby" loop into their radio frequency channels, although current systems are much less susceptible to this technique. Bagpipes were another, but useless against Scottish aircrew.
jrazz wrote on 5/13/2006, 2:09 AM
The Scottish have an aircrew? : )

j razz
Serena wrote on 5/13/2006, 3:25 AM
Sure thing. The Scots are everywhere.
Grazie wrote on 5/13/2006, 6:00 AM
If you have an ongoing "friendly" working relationship with the sound-person, fine if you don't/won;t see them ever again, I'd suggest you paddle your own canoe and invest in a stand and a mic clip & a mic that WILL pick up the speaker, stick a Tx up its bottom and have a Rx to your camera. I also attempt to cover my tracks with something that will "reach" either the speakers mouth OR is pointing AT the loudspeaker.

I've learnt NOT to rely on the audio person. If I'm working with them regularly - fine. If not, just file/stick this kit in the Save-Ass bag and feel more confident. When you are more flush with $$$ consider a Marantz solid recorder.

Grazie
John_Cline wrote on 5/13/2006, 7:21 AM
"I believe somebody here mentioned that by using two identical mics, one for the speaker and one for the room, that they can be placed out of phase, during mixdown, and eliminate background noise."

The chance of this technique ever being acceptably successful is virtually zero. There are certain circumstances where this can be made to work at a specific frequency, but simply putting another mic in the room, matched or not, and expecting it to cancel the audio when mixed in out of phase is defying the laws of physics. In fact, it could just as easily make the problem worse.

John
B.Verlik wrote on 5/13/2006, 10:11 AM
This is where I saw that:

Reply by: BobGreaves
Date: 4/13/2006 10:22:59 AM

In the future, it is possible to add one additional microphone in a room with the idea that the second microphone containes the room sound of the first microphone. It is helpful if that second microphone is within a foor of the main microphone. In production reversing the phase of the room mic and mixing it in judiciously can help to reduce room sound.

Did you ever notice that the grateful dead often appeared on stage with two microphones on every stand but they only sang into one of them? One microphone was in the reverse phase of the other. The two microphones when combined canceled stage sound in the mix.
Serena wrote on 5/13/2006, 5:14 PM
This is similar to the technique employed in "noise cancellation" earphones. Those use a microphone to pick up external sound and insert it antiphase into the signal played. This works because there is complete discrimination between the wanted sound and unwanted noise and noise phase differences (between outside and inside earphone shell) are practically zero. To match this you use something like the set-up that Grazie described (getting full amplitude signal with good S/N) and use another mike to get room noise without the speaker. Of course the latter is the problem, because often it will include the speaker and the noise recorded will not be in phase with noise received by the speaker's mike and therefore will interfere with that when signals are mixed. But, give it a try -- the phasing problem appears rather tricky but sometimes that which shouldn't work, does. I'm not sure that "The Grateful Dead" stage recordings are a good standard for evaluation; rolling phase errors might be an enhancement. Och Aye!
farss wrote on 5/13/2006, 5:48 PM
As Grazie would say "like a moth to a candle":

Sound travels at 1000'/second, in rough figures.
At 50Hz that gives a wavelength of 20', we can ignore that if the two mics are 1' apart. Cancellation will work.
At 500Hz the wavength is 2' i.e. there'll be a 180deg phase difference between what the two mics pickup. However we've already added a 180deg phase shift so instead of cancelling the two signals will ADD!
Plot the results over the audio spectrum and you find what you've created is a comb filter aka flanging.

Quite possibly the trick works for the band by limiting it to low frequencies only and that's what they'll have lots of on stage. Note that changing the distance between the mics has a dramatic effect on what's going to happen.

Babies don't cry at low frequencies, the spaces in question (school rooms) are highly reflective also. Forget this idea.
Well no, don't take my advice, try it. At least when it doesn't work you'll understand why. And please use decent monitors to listen to what's happening.

There is a much simpler solution. Do what has to be done to do it right up front.

I've told this story before but it's worth repeating.

One of my first, totally mine, jobs was to shoot a demo for a young ladies scholarship application in the clients home.
Young lady had a night out on the tiles and got home at 3 AM, she hadn't rehearsed and the entire extended family was on her case. She was breaking down in tears, in short the whole shoot was going south big time and I was feeling this was more like family feud than anything else.
I lost it, big time. Pretty ugly so they say. I threw everyone out apart from the girl. She performed brilliantly and went on to win the place.
I still have that client and he recommends me to everyone he can.
Don't be intimidated, there is a time to simply take charge. Be polite, be considerate but let them know who's in charge, know what you're doing and explain why you need certain things to happen. That doesn't mean becoming a drama queen either, in a reality setting forget the arty stuff, that you can add in post but the basics like good clean audio and vision are vital, neither of those can be fudged in post. Get those upfront and then you can easily add so much in post. Nothing kills the creative spirit like spending hours just cleaning up a mess before you can get started.

Bob.
Serena wrote on 5/13/2006, 7:35 PM
You can see why I said that "simple summing won't work". But I think also that people should be encouraged to experiment when easily done and the effort involved is small. And obviously the topic has wandered from the original query and into the problems of dealing with noisy recording environments. If you really want to make the proposed technique work you only have to digitally process the signals making phase corrections according to frequency. Digitise the signals, take narrow frequency bands, process phase error for the specific arrangement, subtract noise proportional to amplitude as recorded on speech mike and output. This requires a little bit of powerful processing, but will work (in principle). Agree that the cheaper option is to do as specified by others.