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

For our current usage, this doesn't matter, but is more consistent
 with the rest of the API.
We may want to reduce this to an unsigned char[], but I'd rather
 coordinate that optimization with SILK's planned reduction to
 8-bit CDFs, as we may be able to use the same code.

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

This ensures that the various alternative routines in the entropy
 encoder and decoder (e.g., ec_{enc|dec}_bit_logp()) really are
 just specialized optimizations of the same general ec_encode()
 and ec_decode() routines.
This is done by randomly picking one to encode with for each symbol,
 and randomly picking a different one to decode with.

This simplifies a good bit of the error handling, and should make it
 impossible to overrun the buffer in the encoder or decoder, while
 still allowing tell() to operate correctly after a bust.
The encoder now tries to keep the range coder data intact after a
 bust instead of corrupting it with extra bits data, though this is
 not a guarantee (too many extra bits may have already been flushed).
It also now correctly reports errors when the bust occurs merging the
 last byte of range coder and extra bits.

A number of abstraction barrier violations were cleaned up, as well.
This patch also includes a number of minor performance improvements:
 ec_{enc|dec}_bits() in particular should be much faster.

Finally, tf_select was changed to be coded with the range coder
 rather than extra bits, so that it is at the front of the packet
 (for unequal error protection robustness).

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).

Also dropped support for radices above 5.

The main encode call no longer takes a pointer for the optional resynthesis.
It's now done with a call to celt_encode_resynthesis().

Timothy B. Terriberry authored 15 years ago and Jean-Marc Valin committed 15 years ago

Accuracy for rsqrt, rcp, cos, and log2 is now at the level of truncation error
 for the current output resolution of these functions.
sqrt and exp2 still have non-trivial algebraic error, but this cannot be
 reduced much further using the current method without additional computation.
Also updates the fast float approximations for log2 and exp2 with coefficients
 that give slightly lower maximum relative error.

Patch modified by Jean-Marc Valin to leave the cos approximation as is and
leave the check for x<-15 in exp2 as is.

by POSIX. The other _t types that are not part of the API are still there
for now. Also, got rid of all that was left of the 64-bit types.

creating the mode. This means that the same mode can be shared for mono and
stereo.

Making it so all the information encoded directly with ec_enc_bits() gets
stored at the end of the stream, without going through the range coder. This
should be both faster and reduce the effects of bit errors.

Conflicts:

	tests/ectest.c

Gregory Maxwell authored 16 years ago and Jean-Marc Valin committed 15 years ago

"tandem-test" which tests the encoder and decoder in
many modes.

Timothy B. Terriberry authored 15 years ago and Jean-Marc Valin committed 15 years ago

Adds specialized O(N*log(K)) versions of cwrsi() and O(N) versions of icwrs()
 for N={3,4,5}, which allows them to operate all the way up to the theoretical
 pulse limit without serious performance degredation.
Also substantially reduces the computation time and stack usage of
 get_required_bits().
On x86-64, this gives a 2% speed-up for 256 sample frames, and almost a 16%
 speed-up for 64 sample frames.

problems when building shared libraries only.

Timothy B. Terriberry authored 16 years ago and Jean-Marc Valin committed 16 years ago

When I removed the special case for EC_ILOG(0) in commit
 06390d08, it broke ec_dec_uint() with _ft=1
 (which should encode the value 0 using 0 bits).
This feature was tested by ectest.c, but not actually used by libcelt.
An assert has been added to ec_dec_uint() to ensure that we don't try to use
 this feature by accident.
ec_enc_uint() was actually correct, but support for this feature has been
 removed and the assert put in its place.

Timothy B. Terriberry authored 16 years ago and Jean-Marc Valin committed 16 years ago

It is no slower with a large number of pulses, and as much as 30% faster with
 a large number of dimensions.

Timothy B. Terriberry authored 16 years ago and Jean-Marc Valin committed 16 years ago

This lets us encode and decode directly from the pulse vector without an
 intermediate transformation.
This makes old streams undecodable.
Additionally, ncwrs_u32() has been sped up for large N by using the sliding
 recurrence from Mohorko et al.
ncwrs_u64 could be sped up in a similar manner, but would require a larger
 table of multiplicative inverses (or several 32x32->64 bit multiplies).
Note that U(N,M) is now everywhere 1/2 the value it used to be.