| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
| |
Add back the capability to change the deep buffer size according to screen state.
This solves various issues related to audio focus, volume and pause control
that arise with large audio buffers.
Those issues should be ultimately addressed by changes in the audio framework.
Change-Id: I6889ecf0e5d8740745152261f27343e1ff533e7b
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixed 2 issues with media volume:
1 - since we use mm port for music and tones port for other use cases
the digital volume should be applied to both "DL2 Tones Playback Volume"
and "DL2 Media Playback Volume".
2 - the total gain applied to audio originating from the AP is the
combination of digital gain in ABE and analog gain in codec. Some use cases
like telephony have a higher priority than media and apply a different (higher)
analog gain. As this analog gain is common to all sources, digital media gain
should be adjusted accordingly to avoid volume bursts while in call and playing
music. This is particularly important in speaker phone mode.
Change-Id: I90200282edca7098603edca2d56821290988cb20
|
|
|
|
|
|
|
| |
Fixed memory leak introduced by commit 4e7a573f in case
of error in adev_open_output_stream().
Change-Id: I4acc070d748cea228da846f95c7826160e0196a5
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Implement one output stream with short buffers and
one output stream with deep buffers.
The stream with short buffers is selected for most use cases and
provides short latency. It uses TONES_DL port and IOCTL write mode.
The stream with deep buffers is used for music playback.
It uses MM_DL port and MMAP NOIRQ write mode.
The deep buffer stream is not used when the device selection is
BT SCO, HDMI or SPDIF.
The echo reference is only taken from the short buffer stream.
Change-Id: I60ef720e52e96970b8b6618f9f43f24baadce60b
|
|
|
|
|
|
|
|
|
|
|
|
| |
Use 4 buffers of 96 frames each = 4 ms at 48 kHz.
Keep the 44.1 kHz -> 48 kHz up-sampler in HAL.
Disable mmap mode and non-IRQ mode; this gives better variance for cycle times.
Reduce number of buffers from 4 to 2, works OK in non-mmap mode but not mmap mode.
Update comments based on code review.
Tested with audio input.
Not yet tested with echo cancellation.
Change-Id: I69db00ab408cd2aad5788d602eb01fc0c7e4e78b
|
|
|
|
|
| |
Change-Id: Ia6b6caf67f3c2e53431d7b65c3a30c57975faa2a
Signed-off-by: Mike Lockwood <lockwood@google.com>
|
|
|
|
| |
Change-Id: Ia2d0f55fc065e7071d9f5207e0dc91b63f554759
|
|
|
|
| |
Change-Id: I1169d279b4a59355cf4362a7128b053bf940c158
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Added support for audio pre processing libraries
implementing dual mic solutions.
When a pre processor is enabled, its multi channel capabilities are
queried and compared to capture channel combinations supported by the
device and other enabled pre processings.
The most favorable configuration is chosen and pcm capture driver is
restarted with the appropriate channel config.
Also made various capture and process buffers naming and allocation more
consistent.
Change-Id: I90be4798951d0a34dc77d6bdc93ef15cad3ff5af
|
|
|
|
|
|
| |
Added audio policy manager configuration file.
Change-Id: I62163e203a42596ac69b2971c5c0fa99817b33b3
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 78a7609d fixed audio drop at the start of ringtone.
This commit fixes another similar issue with camera shutter sound being
dropped while in call over headset.
There was a workaround for this second issue in audio policy manager but this was
not satisfactory as it was impacting all devices for a problem that is
Prime specific.
Change-Id: I42b37c7da4a232323b520a8a55ac5b3086b5a230
|
|
|
|
|
|
|
|
| |
Fix error in get_capture_delay() that was not taking into account
the fact that frames in in->buffer are at driver sampling rate while
frames in in->proc_buf are at requested sampling rate.
Change-Id: I09e627bd316daedab5ffea3dd638254eaa270a5b
|
|\
| |
| |
| |
| |
| |
| | |
selected
* commit 'd28a1a802b1965ab4b9014c658240faafe219994':
audio: fix audio drop when speaker is selected
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
When changing audio path to speaker while playback is active,
several hundred ms of audio are dropped. This is mostly noticeable
when a ringtone starts playing.
This change is a workaround forcing the output in standby when speaker
is selected.
The root cause must still be indentified and fixed.
Change-Id: Idef8dc1cdbf2da499a414d0b60244f91ef66e73b
|
| |
| |
| |
| | |
Change-Id: Ib1d5af6687479c8d189a3407c229a6ac0ed5c03b
|
| |
| |
| |
| | |
Change-Id: Ifb68db236cb6b9e039eadf573e177add1de62d8c
|
| |
| |
| |
| | |
Change-Id: If9c95a4808785e58ee4595e5c762d01d87f1936d
|
|\ \
| |/
| |
| | |
Change-Id: Id432e901f8107a00a7f371e5882b1290a1154961
|
| |
| |
| |
| | |
Change-Id: I5179699b22224473bd158e90f864e4e73895b5dc
|
| |
| |
| |
| | |
Change-Id: I2e2a5f625956dc5d09dbdc3f6f2d9a010ecc7bad
|
| |
| |
| |
| | |
Change-Id: I711920dde1560ca202ef878ee93a2af61545524b
|
| |
| |
| |
| | |
Change-Id: I7a30fe3f66933aed8b5a6185553112575b4de1a7
|
| |
| |
| |
| |
| |
| |
| | |
See https://android-git.corp.google.com/g/#/c/157220
Bug: 5449033
Change-Id: I2e1c43800c19b718cc7ee94ec299c62bc14873b4
|
| |
| |
| |
| |
| |
| |
| | |
See https://android-git.corp.google.com/g/157065
Bug: 5449033
Change-Id: I02cfaca251935e4a50ad4302a72c4273be41db22
|
|\ \
| |/
| |
| |
| | |
* commit '31688e73c947845cea86079aefa2dfab68b56c93':
audio HAL: release audio pre processing buffers.
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Buffers allocated for audio pre processing are not released
when an input stream is closed.
Issue 5753047.
Change-Id: Ie8fd5f49d97e9bebc70fc38de0844a79074ac526
|
| |
| |
| |
| |
| |
| | |
The clock sync func is unused for both HSPA and LTE device.
Change-Id: Ia9f369a0151cb3bb15242544e5f5442b893253bc
|
|\ \
| |/
| |
| |
| |
| |
| | |
ics-mr1
* commit 'ec429c1320e97145b42a8f334ed5506d316bb412':
audio: force speaker route for call when docked
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
As we did for the HDMI audio case, force the speaker route for
calls when in a digital dock because we cannot directly route
the modem audio output through the S/PDIF output because it is
a McASP device.
Fixes bug 5434090
Change-Id: I52ff7877a8be778b9e74eebb3ad2c9f13b634bca
|
|\ \
| |/
| |
| |
| | |
* commit '56e8b292cb2aa15804eb436d48c71e1a98b36550':
audio: decrease headset gain by 14dB for ringtone mode
|
| |
| |
| |
| |
| |
| | |
This is to prevent audio shock in AUDIO_MODE_RINGTONE.
Change-Id: Ic21c347a64ee0e2668dbff49dc6addcb93e4d82f
|
|/
|
|
|
|
|
| |
See https://android-git.corp.google.com/g/#/c/143865
Bug: 5449033
Change-Id: Iab0aa050fba57491f5cb7ed928f44a0fda7d1ea4
|
|
|
|
|
|
|
|
|
| |
Mute and unmute VX_UL gain to avoid pop noises in the tx path
during call switch to the modem during the switch it means when
audio path changes(Example: Analog path switches from EAR<->HS<->HF).
Change-Id: I567d4156a5b9aa7b51d068fe279f942376a5a40c
Signed-off-by: venkappa mala <venkappa.m@samsung.com>
|
|
|
|
|
|
|
|
| |
- new gains for toro and maguro devices for various use cases.
- use of DL2 digital gains to compensate for lack of range in
codec speaker volume.
Change-Id: I4ff1ebe79aa53934720389fbef5f60b9c0cc2138
|
|
|
|
|
|
|
|
|
|
| |
Mono mixer is only strictly required for downmixing stereo media
content to the mono speaker, so only enable it then. This works
around an issue with modem rx mute when using handsfree.
Fixes bug 5481245
Change-Id: I8e4c5400241a0d8bb8d74966b6f612b7bab56301
|
|
|
|
|
|
|
|
|
| |
Defined new audio buffer sizes to help increase periods
of idle CPU with new scaling governor settings.
Related to issue 5486806: mp3 playback power re-regressed...
Change-Id: I5f0f54d0ef8e189c2e3ac84bf8eed4bafece9111
|
|
|
|
|
|
|
|
|
| |
Some metalic noise is happened on headset, earpiece voicecall.
Especially, The noise can be felt easily in woman voice.
If we use 4Khz LPF, the noise is gone.
Change-Id: I106efd89af2b84fad40314c8c07b5f0aa7901c8b
Signed-off-by: Changoh.Heo <changoh.heo@samsung.com>
|
|
|
|
|
|
|
| |
Disable low power audio playback when audio capture is active
even if screen is off to avoid high latency during SIP calls.
Change-Id: Ib559bf2877b0cf89731e039b1bfab2bc3806f56a
|
|
|
|
|
|
|
|
| |
Since the speaker is only connected to the DL2 left channel,
downmix all DL2 audio from stereo to mono to avoid losing
information.
Change-Id: I8f536d3373b5517682722422df648d9d8050b840
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
At the first incoming call, wb amr callback time is faster
than ril-connecting time so wb status is not updated.
To update wb amr status get it at ril-connecting time.
HSPA supports getting wb amr status,
but LTE does not support it.
Change-Id: I477cb19f8ef72d5461c2800e09958f504ae733e5
|
|/
|
|
|
|
|
|
|
|
|
|
|
| |
Implement a mechanism to dynamically switch between short and long
buffers in kernel pcm driver. Using long buffer significantly
decreases power consumption at the expense of latency.
Therefore a hint is given to audio HAL by AudioService indicating
when the screen is off and low latency is not required any more because
neither video playback, VoIP/video chat or any user interaction is expected.
This mechanism relies on the support for MMAP and NO IRQ write modes in
tinyalsa.
Change-Id: Ida9216a141750137a0592187e24a68f263ef3fbe
|
|
|
|
|
|
|
| |
ALSA period sizes must be a multiple of 24 frames to match
ABE requirement.
Change-Id: I52ac1d5d4a2588a1b66100bfecab6d35339fc718
|
|
|
|
|
|
|
| |
Native 44.1kHz will be used for HDMI audio since the output
device supports it.
Change-Id: I60eebf2556c0384e2a4c21150bee2fbbbd5ca6fd
|
|
|
|
| |
Change-Id: I03ba325b613aef21dba8d16187aaccca08d2a328
|
|\
| |
| |
| |
| |
| |
| | |
ics-factoryrom
* commit 'fcb204e9329241047ed7564c4808440f62a5c580':
Fix issue 5415809: increase HP volume for TTY.
|
| |
| |
| |
| |
| |
| |
| | |
Increase headphones volume to -2dB when TTY mode is full or VCO
as per Samsung's request.
Change-Id: I92da179b487c87d07bc363f7344c20cc8779abd6
|
| |
| |
| |
| | |
Change-Id: Ia571fca8e0ce384283a15024b6b271231bf86479
|
| |
| |
| |
| |
| |
| |
| |
| | |
The number of frames written to the echo reference buffer in out_write() was wrong.
As we write frames at the audioflinger sampling rate we should write the number of
frames passed to out_write(), not the number of frames passed to tynialsa after resamopling.
Change-Id: Ia6a1c7e090c73e1566634a17b720e1e6049b22fe
|
|/
|
|
|
|
|
|
| |
In case of an error when opening the modem pcm driver in start_call(),
the order in which the tinyalsa pcm streams were relased was wrong and
could cause calling pcm_close() on a null pcm stream.
Change-Id: Iad7149997d3993561f4a3ed4b2005f5867b51c56
|