[vorbis-dev] Can compressed music sound better than uncompressed?

[Robert Voigt]

On Thursday 10 May 2001 06:39, Mike Coleman wrote:
> My vote is for "interesting bullshit".  Pushing audio through a codec is
> arguably like pushing it through a very complicated dynamic filter.  The
> result, like the result of any filter, might sound subjectively better to
> someone, or even most people, but I can't see it sounding better to
> everyone for all kinds of input. 

Of all replies this sounds most convincing to me. That does not mean I wanted 
to hear it's bullshit.

I'm still asking myself: Can there be "inaudible signal content" in music 
that  makes the music sound worse if present and better if removed, just 
because it is hard for a playback system to reproduce this inaudible signal 
content (meaning it causes some kind of distortion)?

I think it's true for very high and very low frequencies and a low quality 
playback system. I once had a pair of very cheap speakers and tried them with 
a middle class amplifier. It sounded like shit. Then I tried them with the 
amplified output of a cheap soundcard, and I was amazed how good it sounded. 
I think that was because the very high and very low frequencies the middle 
class amplifier produced caused distortion in the cheap speakers. At least 
that's the only relevant difference between the two amplifiers.

But this is only the trivial solution to the problem. A perceptual coder 
removes not only very high and very low signal content, but a lot in between, 
whatever is masked. And we usually don't use low quality speakers.

I can imagine a faint inaudible click. A click, no matter how small, causes 
the impulse response in the playback system. In an ideal playback system this 
is a click again. But in reality it looks like 
http://www.headphone.com/ProductsHeadphones/RS1IRGraph.GIF
(this is the impulse response of the impressive Grado RS1!)
Can this cause audible distortion? Can the impulse response be audible if the 
impulse is not, because the response is in a different frequency range and is 
not necessarily masked? 
Can you think of anything else of this kind?

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.