naturerecordists
[Top] [All Lists]

Re: noise reduction techniques

Subject: Re: noise reduction techniques
From: Walter Knapp <>
Date: Sun, 04 Jan 2004 02:03:21 -0500
From: "Rich Peet" <>
> 
> In answering an off line question regarding my post I guess I should 
> clarify part of my prior post.
> 
> Editing for Consumer MD effects for me involves two things.
> 
> 1. When there is no audio that is important above 15,000 cycles I 
> drop the gain in this region by 50% to 75% (6db to 10db) on a slope.  
> It does not effect the overall recording and hides the ragged 
> compression of this area that can be seen on a spectral display which 
> is the signature to many that the recording was a MD.  It really does 
> not change the heard audio any, but for those that pay attention it 
> fools a few of them into thinking they may be looking at a 24 bit 
> recording with >100db s/n.  I use it with restraint but hey if it 
> makes someone think that I am better and have higher tech equipment 
> then I do then it was worth the effort.  This part of the edit is 
> vanity. 

Unless they use sonograms, few would find the MD effect you speak of. 
Nor is it consistently there. If it is in your equipment, then there may 
be some other fault with the equipment. It is certainly not consistently 
there in all the sonograms I've done on MD recordings.

Perceptual encoding works by choosing where quantization noise will be 
least likely to be heard. Not, as some may think, by choosing what part 
of the sound to remove. The effect can be most easily seen in the 
highest part of the frequency range. Where, as I've noted there are few 
samples to spare and they need to be right on. But that does not mean 
it's not in all other frequencies as well. Even in the high frequency 
range it's not produced from nothing, you have to have sound in that 
range to show the noise. Silence is rather easy to encode.

Perceptual encoding is not the only way to produce this, so it's not 
really a telltale for MD. In fact by posting the sample as mp3 you added 
another layer of quantization noise from it's encoding. And could have 
used that as cover.

I could easily see that you had filtered the high frequency down on 
sonogram. That could also be the signature of using something like 
cassette, however. I would not confuse it with 24 bit by sonogram. And 
definitely would not confuse the result with a higher sampling rate, 
which would show the higher frequency signals more clearly.

I sometimes apply a high cutfilter to my recordings. But it is mostly 
done when I have unwanted sounds out of the environment I'm recording. 
Like the huge insect noise in July here.

I really don't care what level of equipment a person uses. If I judge 
them by their recordings it's in how well they did with the equipment 
they used and the soundfield they were trying to record. Only way I 
might judge someone about equipment is in the equipment choices they 
make. Is their outfit balanced, is it well chosen for the type of 
recording they want to do and the money they had, that sort of thing. 
But that would be far second to the recordings themselves. Surprisingly 
the most room to impress is given by the lowest end equipment. As you 
progress up the scale it's simply easier to get a good recording. That's 
the real truth, we buy high end stuff so we can be lazy and sloppy in 
our technique ;-) But also to give ourselves more room for our own 
creativity to enter the recording.

> 2. The real edit for consumer MD effects is to cover a less than pro 
> mic preamp.  The mic pre noise is pretty uniform over everything.  If 
> there are portions of the sound spectrum that are not used at all it 
> makes sense to drop the gain 50% in those areas as you are 
> effectively increasing the signal to noise level. Keeping in mind my 
> own bias that bass, even if uniform, is needed in a natural sound 
> recording for use as a reference.

Below your cutoff frequency, I believe you overdid it a little. All the 
noise I see is not city noise, or even pre. I would have trimmed the 
bass a little more so the ducks stood out more, probably run my active 
noise filter on it. They sound pretty distant for a Telinga. But each of 
us has our own style. They do sound appropriate for a bunch of ducks 
being fed. I could close my eyes and visualize them.

I do, however, agree that a consumer MD pre does at times need help. 
This is true of any recorder at that quality level. This is particularly 
true if higher gain settings were used. Removing broadband noise so it 
sounds right is tricky, if it can be done at all. Especially if mixed 
into the calls you want.

> So to make it clear, I am not trying to hide artifact or any real 
> defect of any kind in editing Consumer MD recordings.

There is too much of this strange idea that we are even capable of or 
have the equipment for making perfect reproductions of what we hear 
onsite. We have no such ability. It's that idea which leads to a lot of 
witch hunts. At best we can make a flawed reproduction to give folks 
some idea what we heard. Removing flaws we perceive, real or not, is all 
part of the business. A recording is the creation of the recordist. And 
our recordings are more than the sum of our equipment, they include our 
personality, our thoughts, even our very soul if you wish.

You did not represent this recording as being more than the first of a 
new year full of potential recordings, a uplifting idea, which is how I 
took it. So looking for hidden motive seems a little strange, especially 
as you made it clear you did filter it. We all like to get our first 
recording, though to speak truthfully, they are rarely the best of the 
year. More like a launch on another recording season. The season does 
not launch until the first recording for some of us. It is a important 
ritual. I kind of pity folks who think it's a technical ritual, that 
hardly matters.

May everyone's new year of recording be filled with good sounds to 
record. A new season is upon us. Though some of you up to your armpits 
in snow and ice may take a little longer to notice.

Walt







________________________________________________________________________
________________________________________________________________________

<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU