Timothy B. Terriberry authored 13 years ago and Jean-Marc Valin committed 13 years ago

Although the current encoder uses does not use them, the decoder
might have tried to use offsets beyond the end of these tables.
Inserting values here ensures that the stream can be decoded in a
well-defined way.
The particular values of these tables were chosen to match a
Gaussian distribution with sigma=3.93.

Ralph Giles authored 13 years ago and Jean-Marc Valin committed 13 years ago

This removes the configurability of the OPUS_TEST_RANGE_CODER_STATE
consistency check, in favour of always running it. The performance
penalty is low, and it's an important check on correctness for the
reference encoder to provide.

Ralph Giles authored 13 years ago and Jean-Marc Valin committed 13 years ago

Fix a warning on gcc 4.6. 'internal_sampling_rate_Hz' was
replaced by 'bandwidth' in a10e8796 but the local variable
holding the setting wasn't removed.

Ralph Giles authored 13 years ago and Jean-Marc Valin committed 13 years ago

Automake version 1.11 and above provide a build mode where
each build command is replaced by a single-line annotation.
Since autotools build lines for libraries typically each fill
an entire terminal screen, this mode makes it easier to keep
track of build progress, and warnings stand out more clearly.

For debugging the build process, the old behaviour can be
recovered by running 'make V=1' at bulid time, or by passing
--disable-silent-rules to configure.

Since Automake 1.11 is still very recent, we only invoke
the AM_SILENT_RULES macro if it is defined, allowing fallback
to the old behaviour under previous versions of automake.

Ralph Giles authored 13 years ago and Jean-Marc Valin committed 13 years ago

A comment in opus_decode() said, "bit 6 is ignored" when in
fact bit 6 is the padding flag, parsed in the previous block.

The comment, and the previous one, are modified to reflect
the current code.

Ralph Giles authored 13 years ago and Jean-Marc Valin committed 13 years ago

I've changed the grouping of the options to place the performance-
related ones after the more serious thread-safety options. This
is still somewhat confusing as with C99 on wants -DHAVE_LRINTF
but NOT -Drestrict=

I've also added leading whitespace to the comment lines to better
distinguish them from the commented-out code lines.

- Merged the LPC stabilization from NLSF2A_stable.c into NLSF2A.c
- The bandwidth expansion in NLSF2A() now operates on int32 LPC coefficients in
Q17 domain (instead of int16 Q12 coefficients)
- The function bwexpander_32() has a more precise way of updating the chirp
variable (round to nearest, instead of round down)
- Changed a few variables in NLSF_stabilize() from int16 to int32 to avoid signed
wrap-around (no difference in results as the wrap-around would always be reversed
later)
- The LSF codebook for WB speech has a quantization stepsize of 0.15 (was 0.16).
This doesn't break the bitstream, although it slightly limits quality of signals
encoded with the old version and decoded with the new one (I can't really hear it
and PESQ gives high scores as well).  I does improve handling of tonal signals.
- As discussed: the Q-domain of the poly function is now in Q16 (was Q20)
- As discussed: limiting the LSFs in NLSF_decode() to 0...32767
- The silk_NLSF_DELTA_MIN values were lowered to deal with a possible future situation with less or no input HP filtering.

Also fixes two stereo mode switching issues.