- 3ff5e4c Making sure compute_allocation_table() doesn't overrun at lower sampling rates by Jean-Marc Valin · 15 years ago
- f080a15 version string set to 0.8.0-git by Jean-Marc Valin · 15 years ago
- 137f336 Changed compute_allocation_table() so it handles ebands that start and end in the same allocation band. Also fixed a minor C89 issue. by Jean-Marc Valin · 15 years ago
- 31bec96 Fix small mode leak (forgot to free the new logN array). by Jean-Marc Valin · 15 years ago
- f400a3c Caching log2_frac(N, BITRES) in the more data to save some CPU by Jean-Marc Valin · 15 years ago
- 64209a3 A few minor optimisations (compute_allocation, denormalise_bands, deemphasis) by Jean-Marc Valin · 15 years ago
- 9a92d61 A bit more tuning on the pseudo-frac-Hadamard. Also Trying to improve by Jean-Marc Valin · 15 years ago
- 354bf60 Doing the spreading with a "pseudo-fractional-Hadamard" transform by Jean-Marc Valin · 15 years ago
- 8974f00 fixed-point: improving accuracy of the energy prediction by Jean-Marc Valin · 15 years ago
- c7b01ba fixed-point: slightly increasing the decoder accuracy again by Jean-Marc Valin · 15 years ago
- dc8a6ee Update the log2 approximation to accomodate the increased bit precision from bd0610d21b28b76095a49e601df44fb792b96369. This requires another term in the power series to reduce the error to the level of truncation (peak absolute error 0.621785). Also refactor the rounding bias term so that further changes to DB_SHIFT will leave the result unbiased. by Timothy B. Terriberry · 15 years ago
- 3a4a463 fixed-point: Further reduction in mismatch by Jean-Marc Valin · 15 years ago
- 70bdf9b fixed-point: reducing the mismatch in the folded part by Jean-Marc Valin · 15 years ago
- bd0610d Increasing log energy resolution by Jean-Marc Valin · 15 years ago
- 5503f11 Using the mean energy even when start>0 by Jean-Marc Valin · 15 years ago
- 210ccaf Improves fixed-point precision for quant_coarse_energy by Jean-Marc Valin · 15 years ago
- e45ccb1 Fix quant_bands() for start != 0 by Jean-Marc Valin · 15 years ago
- 3a0bc3d Allowing CELT to skip the low frequencies by Jean-Marc Valin · 15 years ago
- 5378bf8 Updating CELT_RESET_STATE to clear some fields that were "recently" added. by John Ridges · 15 years ago
- 6c79a93 Getting "make dist" to work again with the files that got removed by Jean-Marc Valin · 15 years ago
- 619a968 Misc changes for 0.7.1. by Jean-Marc Valin · 15 years ago
- df72655 Change name of libcelt to libcelt0 by Thorvald Natvig · 15 years ago
- 7f18229 Updates celtclient to the new API. by Stefan de Konink · 15 years ago
- 74128be fixed-point: converted PLC filters and LPC parameters by Jean-Marc Valin · 15 years ago
- 303b3b6 fixed-point: starting conversion of the new PLC by Jean-Marc Valin · 15 years ago
- 07fed1b Adding a safeguard against unstable LPC, so now there's no way (that I can by Jean-Marc Valin · 15 years ago
- d69c1cb Making new PLC code work in fixed-point even though it's still using float by Jean-Marc Valin · 15 years ago
- aec0ee4 C89 friendliness by Jean-Marc Valin · 15 years ago
- d5f9930 This fixes a bug in stereo PLC (offset wasn't changed between channels) by Jean-Marc Valin · 15 years ago
- 5a0fae5 PLC: Added lag windowing and constraint to synthesis energy by Jean-Marc Valin · 15 years ago
- b8002a0 Varlength arrays fix for C89 by Thorvald Natvig · 15 years ago
- e14fe90 New LPC-based PLC code by Jean-Marc Valin · 15 years ago
- 1ccfd3c Oops, forgot to free the pitch bufer by Jean-Marc Valin · 15 years ago
- 6fd0270 Fixing (read) array overrun for 1024-sample frames. by George de Vries · 15 years ago
- aebb6fd Some work on the pitch search weighting filter (still disabled) by Jean-Marc Valin · 15 years ago
- e465c14 Reorganised the pitch code to simplify the PLC case and fixed a buffer overflow by Jean-Marc Valin · 15 years ago
- 7a7c42a Fixing a few memory errors by Jean-Marc Valin · 15 years ago
- 065dafd Use stack_alloc for pitch.c by Thorvald Natvig · 15 years ago
- e05e7e2 minor code cleanup by Jean-Marc Valin · 15 years ago
- 6d584ac Renamed MDCT functions to avoid symbol clashes with Vorbis by Jean-Marc Valin · 15 years ago
- 9319e3e Removing original freq-domain pitch code by Jean-Marc Valin · 15 years ago
- 294863b Temporal pitch search by Jean-Marc Valin · 15 years ago
- 5ed24f8 celt_ilog2() is only used in fixed-point mode by Jean-Marc Valin · 15 years ago
- 8c7bb4c Expose the normalized range for reciprocal square roots in fixed-point mode. This allows subsequnt calculations to use the full precision of the result. by Timothy Terriberry · 15 years ago
- 630ee44 SMALL_FOOTPRINT preserves the special n=1 cwrs case by Jean-Marc Valin · 15 years ago
- c2f7dff Added a SMALL_FOOTPRINT option by Jean-Marc Valin · 15 years ago
- a3803a9 misc minor changes by Jean-Marc Valin · 15 years ago
- f81edea removing more unused code by Jean-Marc Valin · 15 years ago
- 789fc14 Removing unused code in the entropy coder by Jean-Marc Valin · 15 years ago
- 954fb11 doxygen fix by Jean-Marc Valin · 15 years ago
- 7a047ea prevent busting at ridiculously low bit-rate VBR by Jean-Marc Valin · 15 years ago
- 45f1110 Making sure the VBR controller never busts the number of bytes allowed by Jean-Marc Valin · 15 years ago
- 25767d1 Better computation of the VBR rate upper bound and reducing the coarse energy by Jean-Marc Valin · 15 years ago
- a9ffc14 Enhancements the fixed-point approximations of non-linear functions. by Timothy B. Terriberry · 15 years ago
- ab4dcc5 Allow coarse energy to take almost all the bits. Also, fixed a some issues with by Jean-Marc Valin · 15 years ago
- 73bf633 Making the multiplication in normalise_residual() produce a 32-bit result just by Jean-Marc Valin · 15 years ago
- 4a275d4 Accuracy improvements to the fixed-point celt_rsqrt(). by Timothy B. Terriberry · 15 years ago
- a3bba38 This should prevent a rare divide-by-zero in the pitch gain code by Jean-Marc Valin · 15 years ago
- bd5d54a New VBR rate controller that doesn't allow more than one frame worth of excess by Jean-Marc Valin · 15 years ago
- 328b8bd fixed a few trivial bugs: exporting celt_strerror(), changed DB_SCALING to a by Jean-Marc Valin · 15 years ago
- 234969c Removed the _t from all the celt*_t types to avoid clashing with POSIX by Jean-Marc Valin · 15 years ago
- 8b2ff0d Updated copyright notices by Jean-Marc Valin · 15 years ago
- 30f7f81 Changed all the celt*int*_t types to remove the _t suffix, which is reserved by Jean-Marc Valin · 15 years ago
- d35d677 Constrain the pitch gain to prevent potential instabilities. by Jean-Marc Valin · 15 years ago
- 163b02e Now using a unique ID for CELT_ALLOC_FAIL by Jean-Marc Valin · 15 years ago
- ece94a0 Improved error handling, and implemented celt_strerror() by Jean-Marc Valin · 15 years ago
- 80ed147 The number of channels is now set when creating the states rather than when by Jean-Marc Valin · 15 years ago
- bf2398b first step for removing the number of channels from the mode by Jean-Marc Valin · 15 years ago
- 6a97802 Adjusting the allocation for stereo directly in the alloc function rather than by Jean-Marc Valin · 15 years ago
- a9947c4 merged the code for quant_bands() and unquant_bands() by Jean-Marc Valin · 15 years ago
- f7a1e16 Renamed mix_pitch_and_residual() to normalise_residual(), after minor by Jean-Marc Valin · 15 years ago
- f6dc1eb Further simplifications to mix_pitch_and_residual() (which no longer does by Jean-Marc Valin · 15 years ago
- 8861955 Doing the mdct shaping in the normalised domain, so that energy is preserved by Jean-Marc Valin · 15 years ago
- 185cabf stereo allocator tuning by Jean-Marc Valin · 15 years ago
- 39f68ac making compute_ebands() a tiny bit less stupid. by Jean-Marc Valin · 15 years ago
- e12017e moved the de-emphasis code to a separate function (shared between encoder and by Jean-Marc Valin · 15 years ago
- de67858 Simplifying the IMDCT by Jean-Marc Valin · 15 years ago
- aa4f58b Simplifications to the code by Jean-Marc Valin · 15 years ago
- 40fc679 more stereo simplifications by Jean-Marc Valin · 15 years ago
- abd6723 No longer interleaving channels of the normalised spectrum. Also fixed the by Jean-Marc Valin · 15 years ago
- 676c5fd simplifications to the stereo code by Jean-Marc Valin · 15 years ago
- 1e3263a Fix for some test program compat and an assertion that didn't make sense anymore by Jean-Marc Valin · 15 years ago
- 4213a72 Demoved the divisions from the inner pitch prediction loops, bumped the version by Jean-Marc Valin · 15 years ago
- d5e5436 Fix for folding_decision() in stereo mode and more cleaning up of the code by Jean-Marc Valin · 15 years ago
- 7013db7 Fix stereo mismatch problem by Jean-Marc Valin · 15 years ago
- 6db9e6e Bringing back min_bins=3 for now by Jean-Marc Valin · 15 years ago
- 75732de Tuning the spreading rotations by Jean-Marc Valin · 15 years ago
- 4834c92 More simplifications from denorm pitch by Jean-Marc Valin · 15 years ago
- 095c178 Removed code that is no longer necessary with denorm pitch and spreading by Jean-Marc Valin · 15 years ago
- 92ae370 enabling denorm pitch only at low bit-rate by Jean-Marc Valin · 15 years ago
- 00fb6b0 fixed-point: denorm pitch converted by Jean-Marc Valin · 15 years ago
- ab220aa denorm pitch works in fixed-point (though there's still some floats left) by Jean-Marc Valin · 15 years ago
- 4c77ea9 doing pitch in denormalized domain by Jean-Marc Valin · 15 years ago
- 515136a removed unused #ifdefs by Jean-Marc Valin · 15 years ago
- 3aa3afe Only calling intra_fold() when there's no pulse allocated by Jean-Marc Valin · 15 years ago
- a7750b9 Re-introducing the successive rotations as a way to control low-bitrate by Jean-Marc Valin · 15 years ago
- 0f0da99 Better fading for PLC: no fading for the first loss, muting after 6 by Jean-Marc Valin · 15 years ago
- 9714f66 Disabling some checks for the C55 by Jean-Marc Valin · 15 years ago
- bb275c4 Making it possible to use the C64x FFT by Jean-Marc Valin · 15 years ago
- 79b3619 Forcing side to be orthogonal to mid for N=2. This saves one degree of freedom. by Jean-Marc Valin · 15 years ago