>> On the contrary, Joe. I can prove humans can tell the difference
>> between certain letters, words, and sounds with 4k of audio that
>> you can't with 2.5k.
>
> For the sibilant sounds of speech, this is certainly the case.
There is a physiological reason for this and it also coincides with the
broad resonance seen at approximately 3.5 kHz on the classic Fletcher-Munson
family of loudness curves. The ear canal forms a pipe with one-open and
one-closed end, with the wavelength of its fundamental equal to
approximately four times the length of the ear canal. Accordingly,
intelligibility at 3.5 kHz is important for maximum speech recognition, but
see caveat below...
And "Jim Tonne wrote:"
> it a "Schedule C" - was really quite pleasant. And that is where I am
> about to stop. You do NOT NEED the region above 3400 Hz to
> make the "quality" any significantly better.
I would agree that 3 kHz represents the lower frequency limit of reliable
speech communication where bandwidth is at a premium -- like old Bell System
circuits and crowded HF bands as two examples. But it should not be stated
that 3 kHz represents diminishing returns in terms of audio quality or
intelligibility until the ear canal resonance frequency is reached -- that
frequency varies among individuals due to slightly different ear canal
distances between the outer ear and tympanic membrane.
One problem associated with extending transmitted SSB bandwidth is that as
BW increases, it takes more SNR to realize its benefit. An increase in
transmitted bandwidth when SNR is low is actually detrimental to voice
transmission. The energy in upper voice frequencies becomes masked by band
noise. The result is what oftentimes sounds like excessive low- frequency
response. So, the weaker the signal, lower BW actually results in greater
voice articulation.
Paul, W9AC
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|