Commits · 017fa857754e15a91910f711c9d3c99d3f7d8752 · Alexander Traud / Opus · GitLab

Snippets Groups Projects

May 08, 2010
- Changing the encoder API to add the frame size · 017fa857
  Jean-Marc Valin authored 14 years ago
  
  017fa857
- Oops! Fixed the fixed-point build · fbd82eb8
  Jean-Marc Valin authored 14 years ago
  
  fbd82eb8
- Fix for very short frame size (M=1) · 811f21ab
  Jean-Marc Valin authored 14 years ago
  
  811f21ab
- Yet another step towards variable frame size · 073d0bc9
  Jean-Marc Valin authored 14 years ago
  
  073d0bc9
- More preparation work for variable frame size · 0475267e
  Jean-Marc Valin authored 14 years ago
  
  0475267e
- replaced instances of mode->nbShortMdcts by the dynamic number of MDCTs (M) · 3b918bac
  Jean-Marc Valin authored 14 years ago
  
  3b918bac
- Allocation table stored with BITRES accuracy based on the short block size · 7f1c9427
  Jean-Marc Valin authored 14 years ago and Jean-Marc Valin committed 14 years ago
  
  7f1c9427
- shortWindow hadn't been used for a while · 5f633543
  Jean-Marc Valin authored 14 years ago and Jean-Marc Valin committed 14 years ago
  
  5f633543
- Shrinking the MDCT's table by separately rotating by the 1/(8N) factor · f7547a4e
  Jean-Marc Valin authored 14 years ago
  
  f7547a4e
Apr 26, 2010
- Making the band definition the same at all frame sizes. · 65ee67ac
  Jean-Marc Valin authored 14 years ago
  
  65ee67ac
Apr 21, 2010
- Minor tweak to the band layout offset · be8d1259
  Jean-Marc Valin authored 14 years ago
  
  be8d1259
- Re-enabled special stereo code for N=2 · 6578ea0e
  Jean-Marc Valin authored 14 years ago
  
  6578ea0e
- Bands are now (again) a multiple of the number of short MDCTs · 4fb95687
  Jean-Marc Valin authored 14 years ago
  
  4fb95687
Apr 20, 2010
- Skipping transient shape renormalisation in the encoder because · d6b79eeb
  Jean-Marc Valin authored 14 years ago
  
  our search can now work on vectors that don't have unit norm
  d6b79eeb
- Bump bitstream version now that we broke everything · 281a63ea
  Jean-Marc Valin authored 14 years ago
  
  281a63ea
- Tuning the amount of spreading based on Koen's feedback · 8271dc8a
  Jean-Marc Valin authored 14 years ago
  
  8271dc8a
Apr 19, 2010
- Improved transient_analysis() by adding one frame of memory. · 9c30de57
  Jean-Marc Valin authored 14 years ago
  
  9c30de57
- Removed unnecessary calls to log2Amp() · b450ed48
  Jean-Marc Valin authored 14 years ago
  
  b450ed48
- Disabling resynthesis when not needed (need to remove folding for this to work) · b8ba70c9
  Jean-Marc Valin authored 14 years ago
  
  b8ba70c9
Apr 18, 2010
- Re-introducing the successive spreading rotations, but in a two-step · b8a06ee0
  Jean-Marc Valin authored 14 years ago
  
  scheme. Keeping Hadamard as an option (disabled for now) for transients.
  b8a06ee0
Apr 17, 2010
- Corrected some non-sensical code · 65a48706
  Jean-Marc Valin authored 14 years ago
  
  65a48706
- Converted a few double-precision constants to single precision · 628c0253
  Jean-Marc Valin authored 14 years ago
  
  628c0253
Apr 14, 2010
- Making sure compute_allocation_table() doesn't overrun at lower sampling rates · 3ff5e4c2
  Jean-Marc Valin authored 14 years ago
  
  3ff5e4c2
- version string set to 0.8.0-git · f080a151
  Jean-Marc Valin authored 14 years ago
  
  f080a151
- Changed compute_allocation_table() so it handles ebands that start and end in... · 137f3366
  Jean-Marc Valin authored 14 years ago
  
  Changed compute_allocation_table() so it handles ebands that start and end in the same allocation band. Also fixed a minor C89 issue.
  137f3366
Apr 07, 2010
- Fix small mode leak (forgot to free the new logN array). · 31bec963
  Jean-Marc Valin authored 14 years ago
  
  31bec963
- Caching log2_frac(N, BITRES) in the more data to save some CPU · f400a3c5
  Jean-Marc Valin authored 15 years ago
  
  f400a3c5
- A few minor optimisations (compute_allocation, denormalise_bands, · 64209a3e
  Jean-Marc Valin authored 15 years ago
  
  deemphasis)
  64209a3e
- A bit more tuning on the pseudo-frac-Hadamard. Also Trying to improve · 9a92d61e
  Jean-Marc Valin authored 15 years ago
  
  transient spreading.
  9a92d61e
Apr 03, 2010
- Doing the spreading with a "pseudo-fractional-Hadamard" transform · 354bf60b
  Jean-Marc Valin authored 15 years ago
  
  354bf60b
Mar 20, 2010
- fixed-point: improving accuracy of the energy prediction · 8974f00d
  Jean-Marc Valin authored 15 years ago
  
  8974f00d
- fixed-point: slightly increasing the decoder accuracy again · c7b01bad
  Jean-Marc Valin authored 15 years ago
  
  c7b01bad
Mar 16, 2010
- Update the log2 approximation to accomodate the increased bit precision from... · dc8a6eee
  Timothy B. Terriberry authored 15 years ago and Jean-Marc Valin committed 15 years ago
  
  Update the log2 approximation to accomodate the increased bit precision from bd0610d2. 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.
  dc8a6eee
- fixed-point: Further reduction in mismatch · 3a4a463f
  Jean-Marc Valin authored 15 years ago
  
  3a4a463f
- fixed-point: reducing the mismatch in the folded part · 70bdf9bf
  Jean-Marc Valin authored 15 years ago
  
  70bdf9bf
Mar 13, 2010
- Increasing log energy resolution · bd0610d2
  Jean-Marc Valin authored 15 years ago
  
  bd0610d2
- Using the mean energy even when start>0 · 5503f11a
  Jean-Marc Valin authored 15 years ago
  
  5503f11a
Mar 12, 2010
- Improves fixed-point precision for quant_coarse_energy · 210ccafa
  Jean-Marc Valin authored 15 years ago
  
  210ccafa
Mar 01, 2010
- Fix quant_bands() for start != 0 · e45ccb16
  Jean-Marc Valin authored 15 years ago
  
  e45ccb16
Feb 26, 2010
- Allowing CELT to skip the low frequencies · 3a0bc3d9
  Jean-Marc Valin authored 15 years ago
  
  3a0bc3d9