Making sure we never waste bits due to band skip and also making
sure we don't skip bands "in the middle".

Now properly (I think) handling thresh[] and skipping

And fixed a post-filter bug for that special case.

This adds some side-information that can be used to change the
threshold freq arbitrarily.

Should be more robust to closely-spaced transients

This reduces waste at high bit-rate

celtenc now defaults to unconstrained VBR.

Also some code to select between constrained and unconstrained VBR

Left and right are coded independently.

Also, making per-band dynamic allocation less aggressive.

Bands that are intensity-coded also get less bits than the others

There's no explicit signaling for it yet.

Mimimum resolution is now 1/4 bit/sample instead of 3/8. The allocation
tilt now gives more LF bits for small frame sizes. There's also a new
"alternate" allocation that should soon replace the current one.

Timothy B. Terriberry authored 14 years ago and Jean-Marc Valin committed 14 years ago

This tunes the entropy model for coarse energy introduced in commit
 c1c40a76.
It uses a constant set of parameters, tuned from about an hour and a
 half of randomly selected test data encoded for each frame size,
 prediction type (inter/intra), and band number.
These will be slightly sub-optimal for different frame sizes, but
 should be better than what we were using.

For inter, this saves an average of 2.8, 5.2, 7.1, and 6.7 bits/frame
 for frame sizes of 120, 240, 480, and 960, respectively.
For intra, this saves an average of 1.5, 3.0, 4.5, and 5.3 bits/frame
 (for the same frame sizes, respectively).

Fixed an off-by-one in the handling of the IIR filter memory and
disabled "TDAC blending" at the beginning of a lost packet until it
can be made to work properly.