Good....

Confirms exactly what I assumed.  

5dB in 24Hz (which is about the same effective bandwidth of a good pair of ears)  is -15dB S/N in 2500Hz

 

Its interesting, too, that SSB is readable at far lower S/N ratios than might normally be expected.  My SSB contesting ears, which have had practice mostly with AWGN on the microwave and VHF bands, but nothing like as good as the experienced ops, can read callsigns and reports - ie preformatted data - at a S/N in the SSB bandwidth of around -10 to -5dB.   If we assume a S/N of +5dB is needed for incoherent detection, that suggests an effective aural bandwidth around 100 - 200Hz

 

Considering the whole human evolutionary process has been geared to listening to voice -ike sounds, its not really surprising that we apply considerable adaptive filtering to the voice channel to get an efcective bandwidth so much lower than the actual band.    So the human voice is a spread-spectrum system.

 

Andy
www.g4jnt.com