aboutsummaryrefslogtreecommitdiffstats
path: root/core/mixer
Commit message (Collapse)AuthorAgeFilesLines
* Remove an unnecessary subtractionChris Robinson2022-04-012-2/+2
|
* Invert a check to put the first taken path firstChris Robinson2022-03-311-11/+11
|
* Make and use a processHfScale function with a separate outputChris Robinson2022-03-311-6/+7
|
* Remove the reversed all-pass trick in MixDirectHrtfBaseChris Robinson2022-03-302-42/+8
| | | | Given the minimum phase HRTF, it's not going to stay linear phase anyway.
* Merge pull request #559 from tatokis/split-sse-sse2kcat2021-05-131-3/+2
|\ | | | | Allow enabling SSE without SSE2
| * Allow enabling SSE without SSE2Tasos Sahanidis2021-04-301-3/+2
| |
* | Decode UHJ buffers to B-Format for mixingChris Robinson2021-03-311-6/+1
| | | | | | | | | | This should also have an adjustment for the shelf filter. Although it's not clear what the appropriate adjustments should be.
* | Adjust the bsinc filter table packingChris Robinson2021-02-233-9/+9
| | | | | | | | | | | | | | | | Now each scale's filter and phase delta are interleaved for each phase index, followed by the scale and scale+phase delta for each phase index. This ensures no holes in the filter coefficients for the fast bsinc resampler for a given run, while keeping the scale deltas in the same vicinity for the non-fast bsinc resampler.
* | Use shifts instead of shuffles with SSE where possibleChris Robinson2021-02-181-4/+4
| |
* | Add some optimization hintsChris Robinson2021-02-163-18/+24
| |
* | Use spans instead of references to arraysChris Robinson2021-02-066-14/+17
|/
* Return non-const from the resampler functionChris Robinson2021-01-026-37/+36
|
* Use more constexpr variables instead of macrosChris Robinson2020-12-255-30/+31
|
* Increase direct HRTF delay to 256Chris Robinson2020-12-251-1/+1
| | | | This should improve the error from the all-pass on the partial input.
* Avoid compiling different sources for different targetsChris Robinson2020-12-134-0/+17
| | | | | | | | Simplifies configuration and fixes a potential problem with inline functions. Inline functions that fail to inline will have a callable body generated. If such a body is generated with the SSE4 source, for example, it can generate SSE4 instructions. Calls for that function in other sources can then end up calling the SSE4-generated body outside of any CPU capability check.
* Be more consistent with size_tChris Robinson2020-12-124-10/+10
|
* Move the mixer functions to coreChris Robinson2020-12-129-0/+1253