I'm sorry if this is a Lame question. I searched but didn't find.

Why the real bitrate in -q5.00 is much larger than in -q4.99?

Thx!

I believe that's the cut off point where Vorbis switches from lossy to lossless stereo coupling. ...Or was that at q6, I don't quite remember.

The -q3.99 to -q4 transient is funny as well.

Riggs

That cutoff point did exist at 5.0 before Vorbis 1.0, but the stereo coupling cutoff point doesn't really exist anymore.

http://www.hydrogenaudio.org/forums/index....showtopic=36084

However, that doesn't mean that interesting things don't happen at .0 boundaries.

I did a little test with a spanish song w/ lots of acustic guitar:



Note: the file in -q3.99 is slightly larger than in -q4.00

Heh, shows how little I've been paying attention to Vorbis development these past couple years...

I am pretty sure it's a -q 6 having read through the source-code before. You shouldn't be able to tell the difference it's just something to keep in mind, due to the fact that one of Vorbis most powerful "bitrate saving" mechanisms outside entropy coding, etc. It's also more sophisticated then "intensity stereo" and can be taken advantage for other purposes. I remember reading something in the source-code about multichannel coupling, a lot of people want to see this implement and it's going to happen, but some new mapping subroutines need to written and new residue books need to be generated, etc.



an acoustic guitar is transient, so it will more than likely use a lot ot padding/impulse short block's during encoding process thus spiking the bitrate.