Sampling frequency on Radio programmes

[Vangelis forthnet]

On Mon Apr 27 11:09:31 BST 2015, Jim web wrote:

> I've been told that

 Many thanks for looking into this, Jim; 
much appreciated!

> Apparently the general wish of the device makers 
> was for this as a 'lowest common denominator' 
> to maximise the number of 'legacy' devices 
> that could accept the resulting shoutcasts.

 Fair enough... I would've thought a SR of 48.0kHz 
was compatible with most devices, but it appears it's not!

> the 48k -> 44.1k happens when acc is tapped out
> of the distribution cache and converted to mp3.

... and

On Mon Apr 27 13:26:21 BST 2015, Owen Smith wrote:

> They're converting from AAC to MP3,
> ie. from one lossy codec to another? 
> Surely the audio quality loss caused by that 
> will outstrip any sample rate conversion issues.
> Why can't they generate the MP3 from sample rate
> converted lossless?

My thoughts exactly! So, 

Uncompressed studio output (LPCM) -> 
(lossy conversion) AAC LC (bitrate?, 48kHz) -> 
(lossy conversion with downsampling) MP3 (128kbps, 44.1kHz)

looks like a far from perfect recipe... :-{ 
Have the AF people heard of generational loss?
https://www.youtube.com/watch?v=h1JjFKCQq7M

Add to that the fact the ShoutcastMP3 streams 
buffer like hell (despite me increasing the buffer 
memory in Winamp to 512KB (default 64 KB)
and the buffer memory in VLC to 10000ms), 
the ShoutcastMP3 streams do not deserve the 
ceal of BBC high standards...

Regards,
Vangelis.