Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Reset nb_no_activity_frames when analysis DTX isn't usedanalysis_fix2 | Jean-Marc Valin | 2019-02-13 | 1 | -0/+2 |
| | |||||
* | Improve silence handling | Jean-Marc Valin | 2019-02-13 | 1 | -1/+5 |
| | | | | Copy the previous analysis data instead of flagging as invalid | ||||
* | Fixing initialization issues on small frame sizes | Jean-Marc Valin | 2019-02-13 | 1 | -3/+5 |
| | |||||
* | Fixes analysis buffering for silence and complexity changes | Jean-Marc Valin | 2019-02-13 | 4 | -20/+50 |
| | | | | The previous code would go out of sync in those cases. | ||||
* | Avoiding problems with x87 | Jean-Marc Valin | 2019-02-12 | 1 | -1/+1 |
| | | | | logE[b] can end up slightly larger than tonal->highE[b] due to rounding | ||||
* | Properly handle a bad stream_id in OPUS_MULTISTREAM_GET_*_STATE_REQUEST | Jean-Marc Valin | 2019-01-23 | 2 | -2/+2 |
| | | | | Thanks to Dmitriy for reporting this. | ||||
* | Refactoring: Isolating the matrix-vector product in gemm_accum() | Jean-Marc Valin | 2018-11-22 | 1 | -36/+39 |
| | |||||
* | Remove useless OPUS_SET_EXPERT_FRAME_DURATION() on CELT | Jean-Marc Valin | 2018-10-30 | 1 | -1/+0 |
| | | | | | | | frame diration is only handled at the Opus layer. There's no change in behaviour, just less puzzling code. Thanks to Dmitriy <dmilvdv@gmail.com> for reporting this. | ||||
* | Retrain RNN classifier weights to include reverberated speech | Jean-Marc Valin | 2018-09-25 | 2 | -627/+625 |
| | |||||
* | Fixes packet parsing for 16-bit CPUsv1.3-rc2 | Jean-Marc Valin | 2018-09-14 | 1 | -1/+1 |
| | | | | | | | Without that change, a very long (> 682 ms) illegal packet could trigger a wrap-around in the test and be accepted as valid. Only 16-bit architectures (e.g. TI C5x) were affected. | ||||
* | Avoiding get_demixing_matrix() symbol clash on unified build | Jean-Marc Valin | 2018-09-10 | 2 | -8/+8 |
| | |||||
* | OpusMSDecoder does not have an arch field | Jean-Marc Valin | 2018-09-10 | 1 | -4/+0 |
| | |||||
* | Apply equal bit allocation to ambisonic channels | Marcin Gorzel | 2018-08-31 | 1 | -51/+6 |
| | | | | | | Fixes issue #95 on GitHub. Signed-off-by: Felicia Lim <flim@google.com> | ||||
* | Fixing arithmetic problems for 16-bit CPUs in ambisonics_rate_allocation() | Jean-Marc Valin | 2018-07-28 | 1 | -5/+5 |
| | |||||
* | Fix bitrate allocation for channel mapping 2 | Felicia Lim | 2018-07-28 | 1 | -1/+3 |
| | | | | Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca> | ||||
* | Fixes C90 "mixed declarations and code" error | Jean-Marc Valin | 2018-07-26 | 1 | -1/+1 |
| | |||||
* | Remove ambisonics experimental flag | Jean-Marc Valin | 2018-07-26 | 6 | -223/+1 |
| | |||||
* | Use ambisonics families 2 and 3 instead of 254 and 253 | Jean-Marc Valin | 2018-07-25 | 2 | -5/+5 |
| | |||||
* | Validate multistream/projection decoder frame_size | Mark Harris | 2018-07-21 | 1 | -0/+5 |
| | |||||
* | Silence compiler warnings | Mark Harris | 2018-07-21 | 3 | -7/+19 |
| | | | | | | | clang -Wcast-align warnings with ambisonics enabled clang -Wnull-pointer-arithmetic warnings in test_opus_api.c gcc -Wimplicit-fallthrough warnings on arm msvc warning C4244 in celt_encoder.c with fixed point | ||||
* | Make bandwidth detection more conservative to avoid low-passing frames | Jean-Marc Valin | 2018-06-14 | 1 | -0/+16 |
| | |||||
* | Avoiding arithmetic on NULL pointerv1.3-rc | Jean-Marc Valin | 2018-05-29 | 1 | -2/+0 |
| | | | | | Causes a warning with -Wnull-pointer-arithmetic and the code wasn't doing anything anyway (the ctl call ignores NULLs) | ||||
* | Avoiding leaks when opus_demo exits with an error | Jean-Marc Valin | 2018-05-29 | 1 | -46/+36 |
| | | | | | This gets rid of false positives in static analysis (see https://trac.webkit.org/changeset/232236/webkit) | ||||
* | Only call isqrt32() with a positive argument | Mark Harris | 2018-05-28 | 2 | -6/+10 |
| | | | | Fixes test_opus_projection failure under ubsan, due to clz(0). | ||||
* | Switch from narrowband to wideband at 9 kb/s, don't use mediumbandexp_bandwidth_fix4 | Jean-Marc Valin | 2018-05-24 | 1 | -8/+12 |
| | |||||
* | Fix decision thresholds for larger frame sizesexp_bandwidth_fix3 | Jean-Marc Valin | 2018-05-24 | 1 | -1/+2 |
| | | | | Don't assume that frames larger than 20 ms save any overhead compared to 20 ms. | ||||
* | Fixes a SILK bandwidth switching regression | Jean-Marc Valin | 2018-05-24 | 1 | -2/+5 |
| | | | | | | | The bug was triggered because f982b84d started using prefill for SILK bandwidth changes, which reinitialized the encoder state and prevented the variable lowpass from working properly. To fix the problem, we preserve the sampling rate and variable low-pass when prefilling. | ||||
* | Silk makes use of Opus VAD | Gustaf Ullberg | 2018-05-14 | 1 | -2/+11 |
| | | | | Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca> | ||||
* | Misc assertion cleanup | Jean-Marc Valin | 2018-03-27 | 2 | -3/+4 |
| | |||||
* | Some missing checks | Jean-Marc Valin | 2018-03-27 | 2 | -1/+3 |
| | |||||
* | Adding multistream decoder validation | Jean-Marc Valin | 2018-03-27 | 1 | -0/+16 |
| | |||||
* | Asserting on some ctl() calls that should never fail | Jean-Marc Valin | 2018-03-27 | 1 | -14/+14 |
| | |||||
* | Adds Opus decoder state validation | Jean-Marc Valin | 2018-03-27 | 1 | -0/+21 |
| | |||||
* | Fixing no-redundancy CELT->SILK and CELT->hybrid transitions | Jean-Marc Valin | 2018-03-27 | 1 | -20/+28 |
| | | | | | We make sure the CELT PLC we do in the transition uses the same bandwidth as the previous (CELT) packet and not the new bandwidth | ||||
* | Support for Ambisonics. | Andrew Allen | 2018-03-21 | 2 | -0/+190 |
| | | | | Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca> | ||||
* | Using a first-order filter for DC rejection | Jean-Marc Valin | 2018-03-12 | 1 | -32/+12 |
| | | | | | | | | | A second-order DC rejection filter is uselsss unless we have complex poles. However, complex poles means we have to compute the filter as a single pass (rather than two casdaded first-order filters), which has numerical issues that would require a higher complexity to solve. So rather than waste cycles with a second-order filter (with a longer impulse response), we just go with a first-order filter. | ||||
* | Fix comment to take into account previous commit | Jean-Marc Valin | 2018-02-21 | 1 | -1/+1 |
| | |||||
* | Increasing the CELT bit allocation slightly for stereo hybrid | Jean-Marc Valin | 2018-02-21 | 1 | -1/+1 |
| | | | | | | Fixes a "regression" from 1.2 to 1.3-beta that was especially noticeable on a Korean speech sample. (it wasn't a real regression since 1.2 only worked because it was using CELT) | ||||
* | Fix divide-by-zeros in opus_demo stats code | Jean-Marc Valin | 2018-02-20 | 1 | -11/+20 |
| | |||||
* | Scaling back the pitch filter when most of the energy is above 3.2 kHz | Jean-Marc Valin | 2018-01-26 | 1 | -0/+15 |
| | | | | That corresponds to the fundamental for the shortest pitch period allowed | ||||
* | Fixes NaN issues in compute_stereo_width() | Jean-Marc Valin | 2017-12-31 | 1 | -0/+6 |
| | |||||
* | Decouple OpusProjection* API from static matrices. | Andrew Allen | 2017-12-07 | 2 | -32/+68 |
| | | | | | Change-Id: I3db285875b6b5548decc436be00096b97be1be3c Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca> | ||||
* | Ensure mapping matrix size is always valid. | Andrew Allen | 2017-12-07 | 4 | -4/+33 |
| | | | | Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca> | ||||
* | Fix memory issues in Projection API. | Andrew Allen | 2017-12-07 | 7 | -190/+345 |
| | | | | | | Modified by Jean-Marc Valin Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca> | ||||
* | fix float constants | Jean-Marc Valin | 2017-11-16 | 1 | -1/+1 |
| | |||||
* | Tuning decision thresholds for low-bitrate music | Jean-Marc Valin | 2017-11-16 | 1 | -13/+20 |
| | |||||
* | Fix matrix export via CTL func. | Andrew Allen | 2017-11-10 | 2 | -6/+10 |
| | | | | | | | | Ensure matrix is properly exported from projection encoder. Change-Id: I5ede77e4f4d0dc82074e2230780777af542a416e Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca> | ||||
* | Retrained speech/music RNN | Jean-Marc Valin | 2017-11-09 | 3 | -205/+642 |
| | | | | Larger network, using 8-bit weights | ||||
* | Support for Channel Mapping 253 | Andrew Allen | 2017-11-07 | 7 | -44/+1104 |
| | | | | | | | | | OpusProjection* classes MixingMatrix class Projection tests Change-Id: I98644466abf4ffd36e48bdecad1204d69e1539b9 Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca> | ||||
* | Fixing (hopefully) bandwidth detection for 24 kHz analysis | Jean-Marc Valin | 2017-10-06 | 1 | -20/+31 |
| | | | | | | | | | The masking term was previously completely broken (even in 1.1). The bandwidth detection relies mostly on the noise floor and can only use masking to cut one extra band. The 12-24 kHz energy is now normalized properly but uses a higher noise floor due to the leakage in the resampler. Bandwidth detection is still mostly useless at identifying SWB speech (it usually says FB). |