OK, I'm not going to say it sounds like ----. I'm going to say that if you want to stream audio, you should figure out how much
audio upper end your source has, how much resolution your stream has, and then PRELIMIT the audio passband so that you don't try to make sine waves out of huge stairsteps. At 56K and up, it is possible on the decoding end to use a dynamic-passband filter like National Semiconductor's DNR (dynamic noise reduction) to smooth the result and give an FM-like output.
Below this speed, if you are streaming anything that contains more high frequency response than old-fashioned Plain Old Telephone Service
(POTS) then you should high frequency limit BEFORE the encoding.
When the few remaining scraps of the higher frequencies are still present, but there's not enough to make a sine wave out of,
it is so distracting as to make the end result annoying, unlistenable, sickening, etc.
Our ears and brain KNOW that nothing ever sounds like this, and it is disconcerting to hear a recording where you KNOW how it should sound, but it doesn't. Plain human voices are still going to be trouble, as the upper harmonics suffer first, but at least music CAN sound somewhat acceptable if this is done. Example- CD data rate of 44.1 khz can only resolve 22.05khz audio so high limit at 20 khz.
Most streams I'm hearing ought to be chopping things off at 5 khz audio to sound clean.
Instead, they try to sample the "whole thing" and the audio jaggies are way too distracting.
audio upper end your source has, how much resolution your stream has, and then PRELIMIT the audio passband so that you don't try to make sine waves out of huge stairsteps. At 56K and up, it is possible on the decoding end to use a dynamic-passband filter like National Semiconductor's DNR (dynamic noise reduction) to smooth the result and give an FM-like output.
Below this speed, if you are streaming anything that contains more high frequency response than old-fashioned Plain Old Telephone Service
(POTS) then you should high frequency limit BEFORE the encoding.
When the few remaining scraps of the higher frequencies are still present, but there's not enough to make a sine wave out of,
it is so distracting as to make the end result annoying, unlistenable, sickening, etc.
Our ears and brain KNOW that nothing ever sounds like this, and it is disconcerting to hear a recording where you KNOW how it should sound, but it doesn't. Plain human voices are still going to be trouble, as the upper harmonics suffer first, but at least music CAN sound somewhat acceptable if this is done. Example- CD data rate of 44.1 khz can only resolve 22.05khz audio so high limit at 20 khz.
Most streams I'm hearing ought to be chopping things off at 5 khz audio to sound clean.
Instead, they try to sample the "whole thing" and the audio jaggies are way too distracting.