Using my ME66 as a boom mic, there's often very low level noise picked up while recording dialogue. If I edit a scene where I have to raise some people's levels, then that hiss/noise gets raised with it. How do I eliminate that noise, so that when I raise levels, there's no hiss?

This guy who specializes in audio told me to use a notch filter. I assume that means taking an EQ (using Vegas 4, for me) and going through each frequency, and lowering it all the way down, until you find the frequency that the hiss/noise is on. Well, I tried that, but so far it's only worked once. Every other time, I spend forever moving the "notch" around, only to find there's hiss in every band!

Isn't there another way?

By the way, I still don't have a windscreen. Is this my problem? I'm not talking about wind noise. . .just basic noise at a very low level that only shows up when the volume is raised. Still though. . .

That is so weird, i don't hear about hiss so much with a xl1s.

Hiss is a funny thing, it is kinda like white/pink noise, it just invades a lot of frequencies.

How to get around this without re-recording all the audio again is next to impossible, but to rule some stuff out, what are you monitoring your audio with?

Are you sure it is not caused by the amplification or audio setup of your computer?

Or the sound card, before you come to the conclusion it is the xl1s?

Zac

I have used Sound Forge to "tweak" audio, and have had some success in removing odd noises like hiss. So, I will assume that Vegas, being made by the same people, will have the same filter. I haven't used Vegas in a while, but I know that even Premiere has a hum and notch. If not, you will have to go through it one frequency at a time. I forget what frequency hiss will normally reside in. When you find it, save those settings as a preset.

Sonic foundry have a Plug in called Noise reduction, steinberg also have a plug in of a similar vein...

I use noise reduction when im restoring old tapes as wella s removing unwanted environmental hums and hisses...

one thing is that your problem may not stem from the mic, it could be a weak line leading to the cam, or even the cams recording heads as well...

Noise reduction is the best way to refine the sound IMO...
Notch filters are ok, as do paragraphic eq's and noise gates (to a point) but which noise reduction, you can capture a second of the noise your wishing to eliminate and it will configure itself to remove those freq fluctuations.
also when EQing, it might be an idea to cut the high shelf.. running a cut off freq filter plugin might also help, but dropping gain usually wrks.

good luck

I tried using a shelf. I had to extend it from 20k all the way to around 300hz or so to get any good result! Isn't that noise reduction filter ridiculously expensive?


You're saying that ideally, I should be able to capture with my mic and not get any noise at all in the background, no matter how much I raise my volume?

Yes, the Noise Reduction filter from Sonic Foundry is great, but expensive. Sometimes you can get it bundled with Sound Forge for just slightly more than its original cost. Usually you are look at more than $200.

It could be the computer I guess. . no way to tell.

i actually think it might be your camera, i dont knwo what your using though, so who knows...

One thing to note is that if your using standard PC monitors the sound wil be different, obviously due to the quality and EQ settings etc etc...

if you use a good quality set, this shoudl help eliminate nuances in what your actaulyl hear and give you a more acurate rendition.

as for recording, it could also be that your gain is set too high...

I usually run a lil lower than needed, then normalise the sound during post.

I should record low? The thing is, the show that most of this stuff with the noise/hiss is for is on Public Access, and the equipment there totally sucks or something, because I have to have my peaks at -6db, instead of -12.

Hiss in post -- isn't that what they tie horses to? :-)

Not much can be done if the hiss is like "pink" noise, in which case it's spread all over the spectrum. An EQ filter can't take it out at all.

Attempting to bring levels way up in digital post might be a problem since there might be artifacts that add to the noise. Try acquiring through an analog route, using a mixer to bring up the gain to normal levels. Might be worth a try.

Good luck!
Dean Sensui
Base Two Productions

I don't own any of that stuff, mixer and all. I need a solution that doesn't cost 200 dollars, and can be found in Vegas, Sound Forge, or something cheap. "Get good audio during acquisition," you say, well I can't. I have no way of monitoring audio while also checking my viewfinder.

if ur using soundforge, (even in vegas actaully) u can run a paragraphic EQ preset for hiss removal

this might help...

the problem here could stem from many factors...
the solutions may be many and varied...

What about track compression? Wouldn't this lower the difference between the louder and softer parts? I don't know much about compression, except what it does. Can anyone recommend a ratio (2:1, 3:1 etc.) and the threshold settings, as well as whatever else you might find relevant?

compression is used to confine teh freq distribution range of certain frequencies..

for speech a ratio of 2:1 is usually tight enough... basically what your doing with compresion is "tightening " up the freq fluctuations of the signal.. for basses a higher compression with a short envelope are usually good if you want to acentuate teh lower end roll off..
for speech its always an idea to chain a slight reverb after the compression to give it a fuller steadier sound.
You could also add vocal harmonisers or harmonic exciters for that added sparkle, but for video it might be a lil overboard.

compression it wont do much for noise reduction thou...

at least an uncompressed white noise can have some processing done to it.. if its compressed, itll make it harder to rectify later on as the freq are bought closer together, and it'll end up being a lot more work to segregate them again if you wnat to do so in teh future.

Yeah, but, what I was saying was that by compressing, don't you lower the difference between the low parts and the high parts? Don't push down the high parts and push up the low parts so that there's less between them?

yes it does, but that would then allow a steady stream of your distorted sound as opposed to a fluctuation which is easier to manage...

even with hiss, if u run it thru a stereo enhancer, you can probably make the hiss sound like part of the environment...

If you compress your sound, it might help a bit, but compresion will also raise the lower end of the freq field and give you an inaudible low rumble(you wont hear it as such, but if u look a your speakers you will see them jump)

in the end, i suggest you experiment to find what works for you.

<<<-- Originally posted by Peter Jefferson : compression is used to confine teh freq distribution range of certain frequencies..
... compresion is "tightening " up the freq fluctuations of the signal.. ... compression it wont do much for noise reduction thou...
... [a compressed signal'll...] make it harder to rectify later on as the freq are bought closer together, and it'll end up being a lot more work to segregate them again if you wnat to do so in teh future. -->>>

Forgive me for being picky, but words have meanings and people can be confused if terms are not used consistently. I believe you are correct but are using the term "frequency" when/where what is really meant is "gain" or "level".

That being said, compression does not work in the frequency domain, at least not unless one is using a sidechain and filtering or eq with a higher-end compressor in order to do frequency-dependent compression. But that's an advanced recording and pro sound technique and I believe is not what you are referring to.

Instead, compression works in the gain domain. The idea behind compression is to try and limit the overall dynamic range of a signal.

A compressor is basically a bandpass (in the frequency domain) amplifier that samples the gain level of the input signal and the rate of change (attack, first derivative of) the input signal. The output (gain) of the amplifier is modulated as a function of those two control parameters in such a way that the dynamic range of the output is lower than the dynamic range of the input.

A compressor can help keep noise out of a signal during recording by allowing one to run a signal "hotter" through the signal chain without running into headroom problems and clipping, since the dynamic range in the compressed signal is lower. In other words, since the gain level of the highest-gain transients and passages is lowered, the overall signal gain can be lifter higher above potential noise and the overall noise floor.

All that being said, you are correct that in general, compressing a noisy recorded signal in general will not help to ameliorate the noise, either in absolute terms or as a matter of perception.

As to recommending threshold, attack, release, hard/soft knee, and (in the case of a compressor/limiter) the limit etc, settings, it is best to either experiment, perhaps starting with manufacturer-recommended settings as a starting point.

BTW, by all accounts a really sweet analog stereo compressor is the RNC (Really Nice Compressor.)

Hi Lance,
the reason i kept it in very lamens terms was due to the fact that i didnt want to confuse the situation by goin on about compression and its variable uses.

in this case, compression was considered to lower noise.
I totally agree with your comments re: compression works on a gain level, however compression does indeed lower or lift the raw freq (as well as gain) even without a bandbapss filter or EQ or shelf, depending on how its used of course.
When i referred to "lift the freq", i should have worded it , "increased the gain of the frequency", so thats my mistake..

That said, in this case we're dealing with post production compression, and my suggestion was to not compress the noise as it may be detrimental to future EQ and noise reduciton repairs.

thats the beauty of Audio these days, u can pretty much do anything with anything...

In this case for noise reduction, i was focussing on compression used for that noise reduction purpose.
I hope i didnt confuse anyone, i jsut didnt want to go too far into the depths of compression as i dont think it would help in this case.

If you are after a nice analogue style software solution, TRacks 24 is absolutely incredible...
Its a standale eq, compressor and limiter modelled on ye ol analogue workmanshit...

Hi,

A notch filter is useful to take out small groups of frequencies.
So, they're best used on hums (60 Hz from electrical group loop,
camera noise, fan noise) which are generally low-mid single
frequencies that are often acompanied by harmonics. When
there are a few harmonics or inharmonics, the sound is called a buzz. Hiss is generally spread across all frequencies, but is
most troublesome in the mid and high frequency regions. For
such wide-band signal, use a low pass filter or a high frequency shelving filter with the filter frequency at 5000 Hz or as low as you can get it without significantly dulling your material.
Most audio editors only offer 2nd order (2 pole, -12 dB/octave)
filters, so if you can manage to serialize two or more, you can
get a steeper rolloff and preserve the brightness of the
material you want to keep.

As for the source of hums, the 60 Hz group loop can often
be eliminated by plugging all of your eletrical devices into
the same outlet and eliminating extension cords. Often,
the 60 Hz hum has 120 Hz and other harmonics, so you need
to notch them out, too. . My Sony VX-2000 seems to add
a large 15217 Hz spike. While you often can't hear this,
these inaudible signals will mess up your level meters and
other signal processing like noise gating. You can identify
these hums by taking a respresentative sample of your noise and
using SoundForge, CoolEdit, etc. as a spectrum analyzer.
Increase the number of bins to the max. Then, sweep the
frequency a parametric equalizer with max boost, high-Q or smallest bandwidth. When you hear a single frequency
boosted, that's one your your hums. Then, decrease the
gain on that eq until you don't hear the hum. Use the minimum
amount of reduction because these filters will also attenuate
neighboring frequencies.

Most of the hiss is probably from your microphone preamps and
A/D circuitry, either on the camcorder or in your mixer.
Consider turning off or unplugging all extra audio inputs.
Position your microphone close to your source so you don't have
to amplify weak signals.

As for post production, the easiest is to use a low pass or high shelf filter on hiss and several parametric filters on hums.
I've tried the Sonic Foundry noise reduction. It's pretty good,
for any reasonable amount of attenuation, you will be left with
a dull soundtrack with robotic artifacts. A noise gate will
kill the hiss most apparent when your talent isn't speaking.

As for the the fellow with the high shelf/300 Hz, issue, were you
able to set the gain of the high shelf ? You should be able to
get at least -20 dB above the cutoff frequency. Stack two or more
if possible. I get a major reduction even with one if I set the gain to -20 dB and the frequency to 5000 Hz.

As for compression, it will amplify your ambient noise. So, use a noise gate/reduction wired *before* the compressor.

Ow, my brain hurts. Does Vegas 4 have a low pass filter? As you said, hiss is spread across most frequencies, and using the track eq effect, if I do a high shelf, I have to cut everything from 20,000 hz down to around 640 hz to eliminate most (still not all!) of the hiss, and of course then it sounds like crap.

Regarding the Vegas Video Track EQ, you made the right choice in using the High Shelf. Your goal is to assemble a filter with a steeper rolloff by using 2-4 high shelf filters all set to the same frequency. The It's a pain to adjust 4 filters, but you'll be able to save more of your material. Start by setting each of the four filters to: 5000 Hz frequency, gain = -Inf, rolloff= -12 dB.
Then, toggle them on one by one to hear the effect. Then,
drop the frequency in 250 and 500 decrements, band by band, auditioning each time.