Skip to content

GitLab

Switch branch/tag
  1. 12 Jul, 2016 1 commit
    • Glenn Kasten's avatar
      Update the Wilhelm unit tests · 34ea7482
      Glenn Kasten authored 
      Some of the unit tests had experienced bit rot, so fixed the tests to
run again.  Except "xa" is not completely fixed yet.

Change-Id: I97ebc87cd859bd2b9893f0b315d94b1a09c85c09
      34ea7482
  3. 11 Jul, 2016 2 commits
  5. 08 Jul, 2016 3 commits
  7. 06 Jul, 2016 2 commits
    • Eric Laurent's avatar
      Add tests for pre processing interfaces · 14e6affe
      Eric Laurent authored 
      Modify slesTestRecBuffQueue test to allow control
of performance mode and pre processing effects:
- option -mX (X=0,1,2,3) sets the performance mode
- option -xY (Y=a,e,n) enables AGC(a), AEC(e) or NS(n)

Bug: 4999001
Change-Id: I7975cf424c43e82db976963ee25924f18ad55400
      14e6affe
    • Eric Laurent's avatar
      Fix pre processor interfaces implementation · 677c7960
      Eric Laurent authored 
      - Fix SL_IIDs for AEC, NS and AGS interfaces to match actual
effect type GUIDs.
- Remove IsAvailable() method which did not work because the interface
could not be exposed anyway if the effect was not present
on the device. Also the implementation was linking directly against
the effect factory which is not allowed in non priviledged processes.
- Make pre processor interfaces EXPLICIT instead of OPTIONAL.
- Exit early in AudioRecorder realize if the native AudioRecord creation
fails.
- Add native AudioTrack cleanup in case of error in AudioPlayer realize().

Bug: 4999001

Change-Id: I6235248234491eaa9723cde9de85b965c031571d
      677c7960
  9. 09 Jun, 2016 1 commit
    • Eric Laurent's avatar
      Add configuration option for performance mode · 8c9071f4
      Eric Laurent authored 
      Performance mode tells the framework how to configure the audio path
for a player or recorder according to application performance and
functional requirements.
It affects the output or input latency based on acceptable tradeoffs on
battery drain and use of pre or post processing effects.

Bug: 20825626
Bug: 17981873

Change-Id: I0e79cf2da03b4bfa2080fe6d4ea9d8753ffe3497
      8c9071f4
  11. 13 May, 2016 2 commits
  13. 29 Apr, 2016 1 commit
  15. 28 Apr, 2016 1 commit
  17. 25 Mar, 2016 2 commits
  19. 24 Mar, 2016 3 commits
  21. 21 Mar, 2016 1 commit
    • Glenn Kasten's avatar
      Fix copy/paste error · 6994a522
      Glenn Kasten authored 
      See OpenSL ES 1.0.1 section 9.2.19 SL_METADATA_FILTER

Bug: 27658100
Change-Id: I0a82d27142454f73a1ee41d569b08bc558f3c43e
      6994a522
  23. 17 Mar, 2016 1 commit
  25. 14 Mar, 2016 1 commit
    • Paul McLean's avatar
      Native routing phase 4 · 15f1e492
      Paul McLean authored 
      Allow Acquire before Realize

Bug: 23899814

Change-Id: I27682ecb5259fc6ebbe3eca5c6aba5d956aa42bb
      15f1e492
  27. 10 Mar, 2016 2 commits
  29. 09 Mar, 2016 1 commit
    • Paul McLean's avatar
      "Refinements" to Native Audio Routing API · 053b3c26
      Paul McLean authored 
      Change method names to more generic "AcquireJavaProxy" to support
(potentially) multiple proxy types.
Add "Release" method.
Remove JNIEnv* parameter from Acquire and Release methods.
Add Type ID parameter to support (potential) multiple proxy types.
Enforce "singleton" allocation for proxies.
Disallow proxy acquisition for innapropriate players/recorders.
Refactor to make the code clearer.

Bug: 23899814
Change-Id: Iff354ba46547781e0998a9032de96f445d1474b8
      053b3c26
  31. 01 Mar, 2016 4 commits
  33. 29 Feb, 2016 1 commit
  35. 22 Feb, 2016 1 commit
  37. 19 Feb, 2016 2 commits
  39. 18 Feb, 2016 1 commit
  41. 16 Feb, 2016 1 commit
    • ilewis's avatar
      Emulating old channel mask behavior for both input and output · 28340d1d
      ilewis authored 
      This adds the "ignore positional channel masks" behavior back
into output as well as input.

Bug: 27042146
Change-Id: I51fc4a4494cd9477fdcddb6901cd353b88ea2c2e
(cherry picked from commit f4f89da331441ec4693c8ebb6ae3cf5460a22328)
      28340d1d
  43. 12 Feb, 2016 2 commits
  45. 10 Feb, 2016 3 commits
    • Ian Ni-Lewis's avatar
      Merge "Document default behavior when channel mask is 0" into nyc-dev · 452d288e
      Ian Ni-Lewis authored
      452d288e
    • ilewis's avatar
      Restore behavior of channel masks on input · 33f46b8c
      ilewis authored 
      In the past we ignored positional masks on input 
and used the defaults based on channel count
instead. Changing this behavior to validate positional
input masks breaks apps that had relied on earlier
behavior, without adding much benefit. Restore
the original behavior on input.

Output channel masks have always been defined, so
no change to that behavior.

Bug: 27042146
Change-Id: I91544c719f450f2790b839981243451bc767933f
(cherry picked from commit deb4ccb9556d25dc0ad20a84efadf9a2bf62b20a)
      33f46b8c
    • ilewis's avatar
      Document default behavior when channel mask is 0 · 9b006ad6
      ilewis authored 
      The behavior has not changed substantially; the implementation
has always assigned a default mask based on the number of
channels requested. However, this default behavior does not
carry over into the new indexed channel masks, and that seems
worth noting.

Change-Id: I84b45560dd6988d41e9e6b2cbc7073e03a3a66d3
(cherry picked from commit 91823ae89a82dcd57f8a1cc9711215418bc30503)
      9b006ad6
  47. 09 Feb, 2016 1 commit