Welcome to mirror list, hosted at ThFree Co, Russian Federation.

github.com/mumble-voip/speex.git - Unnamed repository; edit this file 'description' to name the repository.
summaryrefslogtreecommitdiff
path: root/libspeex/ltp.c
AgeCommit message (Expand)Author
2015-04-01remove trailing whitespaceTristan Matthews
2013-12-07Tidy the speex/*.h includes some moreRon
2008-09-06removed some if()s from an ltp inner loopJean-Marc Valin
2008-08-13Added an option to implement iir_mem16() and fir_mem16() in terms ofJean-Marc Valin
2008-08-11Making it more explicit that the nb filter size is constantJean-Marc Valin
2008-08-06Making it possible to disable either the encoder or the decoder using macrosJean-Marc Valin
2008-08-05Sorting out #include "" from #include <> and using relative paths for includingJean-Marc Valin
2007-11-23Removed some warningsjm
2007-11-23Patch by Thom Johansen: define and use SPEEX_MEMSET, SPEEX_MEMMOVE andjm
2006-12-27fixed-point: making better use of memory -- saved 512 bytes of stack.jm
2006-12-26merged some fixed-point and floating-point code in open-loop pitch search. jm
2006-12-08Now using only the 16-bit version of the filters.jm
2006-11-26Removed the generic PSHR/SHL/SHR operators and changed them to either the jm
2006-11-19should no longer complain about unused var with the float build.jm
2006-11-17Enhancer should no longer overflow on insanely loud signals.jm
2006-11-17Think this should prevent overflows in the pitch search when the input isjm
2006-07-29copyright details, more info about the mdf learning ratejm
2006-06-15removed unused varjm
2006-06-15Patch by David Rowe to split the 3-tap search inner loop in a separate functionjm
2006-06-14Swap code/variable declaration to be pre-C99 compliant.tterribe
2006-06-12removed a float/fixed split in pitch_gain_search_3tap()jm
2006-06-10Better control of the pitch gain to prevent potential instabilities fromjm
2006-06-07Patch by David Rowe: sending the stack parameter to *_mem16() filters.jm
2006-06-06Minor optimisation -- should help the open loop pitch search for N=1jm
2006-05-30Cleaned up the pitch constraint a bit. Should be better and faster.jm
2006-05-28oops, declarations go before the codejm
2006-05-27C89 fixjm
2006-05-21Got completely rid of the 64bit type (was only used in the adaptive codebook)jm
2006-05-20excitation and target are now 16-bit (were 32) in the encoderjm
2006-05-16decoder excitation now in 16-bit precision (was 32), which saves quite a bitjm
2006-04-21Fixed a major regression in the 4 kbps mode by adding the target update tojm
2006-04-08Got rid of ~64 divides per sub-frame and saved 512 bytes of scratch space.jm
2006-04-08minor RAM tweak: delayed stack allocation to reduce max scratch memory usejm
2006-04-08weighted signal now in 16-bit precision, saving 500 bytes in the encoder statejm
2006-04-07oops. Fixed an off-by-one that could (at least theoretically) cause stackjm
2006-04-01cleanup and saved 480 bytes of RAM in the decoderjm
2006-04-01bit of cleaning up. More to come.jm
2006-03-19Getting rid of 25600 divide operations/sec in the pitch search. jm
2006-03-18new experimental comb filter codejm
2006-02-14fixed potential problems with float rounding causing NaNsjm
2005-06-07Added MAX16 operatorjm
2005-06-07Blackfin assembly for compute_pitch_errorjm
2005-06-07Separated time-critical portion of the pitch gain computation so it can easilyjm
2005-06-06Removed unnecessary call to open_loop_nbest_pitch (which in turn fixesjm
2005-06-05Packet loss concealment converted to fixed-point. Added hooks for blackfinjm
2005-05-28Cleaned up arch-dependent optimizationsjm
2005-05-10removed some // commentsjm
2005-05-02Added some control on the aggressiveness of the pitch predictor in the form ofjm
2005-04-29Think I fixed the PLC slowdown due to denorm/underflow. Also don't re-jm
2005-04-25convert codebook data (signed char) to spx_word16_t in a cleaner mannerjm
generated by cgit v1.2.3 (git 2.25.1) at 2024-06-02 02:42:29 +0300