Raimund Specht wrote:
> I agree, that such quiet ambient recording situations are much more
> demanding than recordings of relatively loud calling animals at
> close distances. However, I believe, that it does not make much
> sense to amplify extremely soft sounds artificially for playback.
> You should listen to these recordings also in a quiet environment
> (e.g. using headphones) at low sound levels with the playback volume
> adjusted not much louder than at the original recording site.
I think here we have to realize we don't have control on how a
independent listener will play back our recordings. I believe there have
been a number of discussions of this elsewhere, the bottom line being
that listeners tend to have a dB level they prefer, regardless of what
they are listening to. I expect on average we can predict that listeners
will amplify quiet ambiance recordings by many dB above their original
levels. So, the more dynamic range we can manage to obtain the better.
The specs of the equipment are only one part of this.
Music tends to adjust dynamic range via compressors or expanders to try
and handle this problem. Nature Recording tries to be more purist and
not adjust. Which makes our job harder and puts higher demands on the
quality of our equipment.
I've also experimented with another form of filter, a spectral dynamics
filter. One way to use this is to trim sound that falls below some dB
setpoint. By extremely careful use of this filter it's sometimes
possible to trim off the mic self noise. All depends on what the quieter
parts of the recording contain as it's also going to trim the ambient
sound that's in that sound level. Very tricky to use, but some results
have been encouraging.
> The logarithmic sound level scale is based on the dynamic range of
> our human auditory system. 0 dB corresponds to the lowest sounds
> that we are able to perceive. A microphone noise floor of 10 or 5 dB
> would be very close to our hearing threshold. However, when
> considering these few remaining dB's, mother nature still seems to
> have the more advanced know how in designing low-noise
> omnidirectional 'microphones' than any other human microphone
> manufacturer ;-).
As one who has done hearing tests on many thousand young adults
(military recruits and draftees from the central valley of California) I
can say this is a very simplified view of what folks hear.
Even in people with fairly undamaged hearing only a fairly narrow
frequency range will approach 0dB. A few will reach hearing levels in
voice frequencies of -10dB or so (I did back then). Even back when I was
doing it, the early '70's, the average young adult had a lot of fall off
in frequencies above voice (4khz and above in the case of our testing
profile), and even in voice 10-20dB minimums were very common. Some were
only able to hear in the voice range at 40-50dB. And remember that's
folks mostly in their late teens or very early 20's who consider their
hearing ok. It's worse the older you get.
Mother nature may have designed very good 'microphones', but they are
very easily damaged. I also believe that they are not omnidirectional.
More impressive is the sound filtering capabilities of our brain. This
is both good and bad as part of what the brain takes as input is our
attitudes or expectations of what we will hear. It will modify the sound
to meet our expectations, even ones we are not aware of. On the good
side it will act as a extremely good noise filter.
> As Walt pointed out recently, the absolute sound levels in natural
> environments are usually very low. Imagine, you recorded a soft
> sound, lets say with a peak amplitude of 40 dB by using a microphone
> having a self noise level of 10 dB. When you then playback that
> recording at a higher volume in your living room (e.g. 20 dB louder
> than at the recording site), the original microphone noise of 10 dB
> would be reproduced at 30 dB (10 + 20) and the peak sound level at
> 60 dB (40 +20). The amplified microphone noise of 30 dB can then be
> very annoying. However, if the original sound level at the recording
> site were louder (say 60 dB), and you set the playback volume also
> to 60 dB (20 dB less gain compared to the first example), so that
> you have the same sound level in your living room as at the original
> recording site, the microphone noise would be reproduced at 10 dB
> only. That soft noise would be nearly inaudible. Therefore, the
> subjective quality of a field recording would be better if the
> original absolute sound level at the microphone was higher.
The actual specs of a mic as designed into it never change as long as
the mic is functioning properly. But what all this discussion is really
about is relating those specs to our field recording, to the sound
levels we find in the field. We don't really have to be calculating
exact numbers, but being able to judge how our gain adjusts will effect
it all is important. We have probably already wandered far too much off
track in number babbling.
Listening while out in the field is a important part of this. Not just
to the signal from the mic, but just with our ears. Often you can tell
all you need just by listening. But beware of your brain's ability to
ignore noise it's not interested in.
The truth is that all this can be about the choice to even record or
not. If we know because of the conditions in front of us and our mic's
specs that we are going to end up with a tiny dynamic range, we probably
will not record. (or we will hunt a better mic for the conditions) All
depends on the purpose of our recording. The scientific survey
recordings that are a big chunk of my recordings I'll make even though I
know they will not be pleasant to listen to. In those it comes down to
recording identifiable calls. (the main problems there are not mic self
noise, but manmade noise) But, if I'm recording for pleasant listening
I'll take a quite different view. (And if I find a Pine Barrens Treefrog
calling in Georgia I'll record regardless and try to filter him out
later if necessary)
I carry, as part of my kit, a pretty good sound meter. I routinely
measure the sound levels of most recording sites now. As well as
attempting readings from individual animals. I do not do this to set mic
gain, but just part of general scientific curiosity. It's pretty rare
that meter gives me a reading below 35dBA, and more common would be
several 10's of dB above that. Even without manmade noise there are a
lot of background sounds. We mostly tune them out, but the meter does
not. Note for bird recordists, I'm mostly recording at night, so don't
do a lot of daytime readings.
> This seems to be the case in Evert's recordings he made with the
> SoundMan microphones. The ambient sound level in a train is usually
> higher than in a quiet natural environment. Additionally, the broad
> noise spectrum in a train will mask out the microphone noise (by
> psycho-acoustic effects).
There are times in the summer when I could record with a mic with a
70dBA self noise in the 8khz band and it would not be apparent. The mic
could not outshout the seeming millions of insects calling. Once the
summer insects start down here my reach as far as recording frogs is
greatly shortened. I've also had frog sites where it's so loud it was
continuously painful to listen. Where putting on the headphones was
necessary just to protect my hearing. So, it's not always quiet out
nature recording. But, when it is we really test the limits.
One other little thing on self noise specs. The sound chambers used are
generally at "office environment" conditions of humidity and
temperature. At least the few I've been around were. We know that some
mics fail at higher temp/humidity conditions. Often this failure may be
in the form of severely excessive self noise. It would sure be nice to
see self noise profiles related to temperature and humidity. I'm sure
the self noise of some mics varies as these vary. Even before we declare
it failed. A self noise test at, say, 95% humidity and 95 degrees F
would be interesting to me. Even better would be graphs.
Walt
________________________________________________________________________
________________________________________________________________________
|