The question of 192 (or 96) sample rate benefits will probably
never be resolved until the technology is so cheap that lower
rates are forgotten. But even with my aging ears I can hear
important differences between 192 and 96 and 44. The question is
whether a specific audience (artist or customer or whatever)
If you create certain waveforms that are predisposed to audible
aliasing, such as a square-wave frequency sweep, it's obvious
that the Nyquist frequency is at least two octaves ABOVE the
onset of significant audio artifacts. This is not subtle -- it's
gross and obvious to the untrained ear. You can hear aliased
"tones" start sweeping into the output in the midrange.
Although "natural" sound doesn't contain much, if any, of these
problematic kinds of signal, nevertheless it is unsettling to
have heard such gross and bizarre distortion. It makes it hard
to dismiss the benefits of higher sample rates as being "beyond
normal hearing." However, it's equally obvious that we're
accustomed to full-spectrum music and nature recordings at 44.1
kHz, and nearly all of us barely notice any distortion at all.
A subtler issue arises in my own area of work -- spatiality. Our
spatial perception depends on both gross loudness differences
and on subtle phase relationships between the signals at both
ears. Binaural recordings produce a far more realistic sense of
space than simple stereo recordings, because they permit the
listener's brain to interpret small timing differences (on the
order of 1-7 milliseconds) in certain frequencies. The brain
builds a convincing volumetric model of the listening space from
these differences, even though we cannot subjectively perceive
them in their own right.
This point -- that IN-audible differences produce substantial
experiential differences -- again makes it unsettling to
consider an approach to recording that is blind to these (and
other) phase relations. My own experiments over the years have
convinced me that a reasonably simple signal (with only a few
partials), arriving at the ears with a 3-4 millisecond delay,
probably needs 24 bits of amplitude resolution and at least
96kHz sampling to even approximate the original shape. When you
consider that this 3-4 msec delay is often slowly changing,
perhaps sweeping from 0 to 7 msec over several seconds, the
aliasing possibilities are significant even at 96kHz.
Taking all this into account, from the point of view of maximum
useful recording quality, 24 bit 192 kHz is obviously the way to
Having said that, I must return to the initial issue -- what do
your listeners (including yourself) actually CARE about?
Binaural reproduction is rarely used, although probably more so
in nature recording than anywhere else. Few audio systems are
good enough to do justice to 24x192. Most listening rooms are
inherently noisy. Most people have damaged hearing. And the
focus of listeners is, in many contexts, only superficially
focused on the material.
Personally, I like to record at 24x192, mix in the 64-bit
domain, and play back through a 24x192 sound card into bi-amped
Magnepan MGIIIa speakers in a large soundproofed room. Or listen
to binaural 24x192 source on Sennheiser HD-600 phones.
But I also like to play my 8-channel recordings through 8
affordable amps into 8 affordable speakersystems in an
accessible listening room; in such a venue there's a good chance
that 16x44.1 is going to sound very close to 24x192.
I also like to make my recordings available for download, and
although a few people choose the higher resolution options, the
vast majority of downloads are MP3 -- which at its
(psychoacoustic) best is (by definition) damaged beyond repair.
So the conclusion I've reached is quite simple. For best
quality, record at best quality. For selected audiences and
venues, reproduce at an appropriate quality. But always take
into account that's BEST may also be what is AFFORDABLE, or what
can be HAULED AROUND in the field. So -- out of context -- there
is no best. But IN context, clearly it's worth aiming for the
higher word sizes and bit rates, without going crazy or broke in
Hope this helps...
Behalf Of Dan Seven
Sent: Wednesday, February 22, 2006 12:07 PM
Subject: [Nature Recordists] Questions for audio gurus
Hello Audio gurus:
A quick question or 2 ..If a recorder is capable of a frequency
range of 20
to 20kHz such as a new Tascam HD P2 and it can utilize192 kHz
frequency, can you actually hear the upper end of the audible
i can hear to 18.5 Khz or so and then it gets "iffy"
If a good pair of speakers and a CD player is what everyone
uses, then why
would CD quality 16 bit 44.1 playback need to be exceeded for
consumption. Do we all have these playback resources and do we
not all have
ears that hear the same limited range?
The Portadisc records 10 to 20 kHz and the lower frequency is
easier to hear
for people, as in Ruffed Grouse drumming and many many other
sounds. Is this
more useful in actual practice or do most mics simply not
achieve those low
frequencies..my subwoofer does..
The search for meaning here is to achieve an understanding as to
practical recording qualities and capabilities necessary to
playback resources with an audio result that we can actually
HEAR, and not
imagine MUST be there.
I am sure that most of you can contribute a better viewpoint
than I have,
and much appreciated.
Thanks beforehand, Dan7
"Microphones are not ears,
Loudspeakers are not birds,
A listening room is not nature."
Yahoo! Groups Links