] [Thread Prev
Re: OT That 'sizzle sound' of Mp3s
Andy, thanks for putting some fish in the glasses....
The result of the process is not as bandwidth limiting as people
assume (except at very low bit rates the HF response is filtered to
prevent aliasing). I am surprised by how frequently people say that
they think MP3s have no bass. It simply isn't true.
The white noise test will show up artifacts because the process is
based on a theory of perceptual encoding which states that certain
loud sounds will mask certain quiet sounds and this is determined on
a frequency by frequency basis. (in FFT 'bins'). A swept sine wave
allows the process to deal with a single frequency at a time. Given
equal lengths, I wonder which test signal makes a smaller mp3 file.
>Charles Zwicky wrote:
>>You should record a 20-20k sweep tone and convert it to MP3. You
>>would be surprised.
>You're right, I was :-)
>Didn't expect it to reproduce a smooth sweep.
>It's a good test, but doesn't show up the limitations at all.
>(dunnow, maybe in a better listening environment, there certainly were a
> few artefacts revealed when mixing out of phase with the original)
>Next I tried with white noise,
>and the metallic sizzle appeared like magic.
>( tests done with Audition, and mid-fi speakers, sitting
> by noisey pc. mp3 at 128bps
> well, really it was 64bps mono,)