[Elecraft] NR settings at AD4C K3 now available at my website

[Matt Zilmer]

David,

Thanks for your input on this.  I understood pretty much correctly,
but appreciate your more complete explanation.

By "WAV", yes, I meant linear-coded whatever resolution.  I think
windoze Sound Recorder records files in this format.  They're huge...

As a function of my group's charter (at work), we sometimes run
distortion analyses on MP3 vs WAV files for plain voice phrasing.
They're used in turn-by-turn directions on vehicle navigation units.
We'll run the WAV file on the X (ref) axis, and the MP3 on the y axis
and snapshot different parts of the phrases with the display on
short-term average.  We've found differences of up to 10% distortion,
though whether it sounds that way to a human ear I have no idea.  I
suppose it depends on the recording format of the MP3 (resolution
again).  A 12-bit sample looks the worst - easily naked eye visible.
In practice, we use 16-bit sampling at 16 KHz for all non-Text To
Speech pronuciations.  This seems to work best and whatever distortion
there is can't be heard by anyone I've tested.

I was only questioning whether or not this would be a valid
comparison, and it sounds like MP3 is probably OK for plain voice.
Thanks for that answer.

73,
matt W6NIA

On Sun, 30 Aug 2009 19:30:14 +0100, you wrote:

>Matt Zilmer wrote:
>
>> Does anyone have more input on MP3 vs WAV?
>
>MP3 has a certain amount in common with noise reduction algorithms, 
>although what is is trying to reduce is weak sounds so close to the 
>signal that they won't be heard. With very good noise reduction it may 
>completely blank noise that was only reduced by the noise reduction.
>
>Whether it is valid to test noise reduction with a human ear, after the 
>signal has been MP3 coded, depends on whether or not you believe that 
>MP3 only removes what cannot be heard, anyway.
>
>One should never use MP3 on signals that are going to be processed by 
>something other than a human ear, and I'd personally be wary of using 
>for any comparison of audio quality.
>
>By WAV I'm assuming you actually mean 16 bit linear audio, as WAV is a 
>wrapper format that can wrap MP3.
>
>Unfortunately, in the same way that JPEG seems to be used for all images 
>on the web, MP3 tends to be used for all sounds.  Both make assumptions 
>about the nature of the signal and human perception.  (PNG is often the 
>best format for screen shots, for example.)