Current effect implementation in record threads entirely delegates calls to
process() (and process_reverse() for AEC) to the audio HAL.
The reason why the EffectChain process_l() function must be called in the
thread loop is to update the effect state.
When EffectChain::process_l() calls EffectModule::process() (on each
effect), this call is a no op because the effect chain input buffer (and by
way of consequence the effect input buffer) is NULL.
See effect chain input buffer init
in AudioFlinger::RecordThread::addEffectChain_l() and the test
in AudioFlinger::EffectModule::process().
On Wednesday, April 29, 2015 at 8:04:13 AM UTC-7, Glenn Kasten wrote:
>
> The audio input HAL normally runs on the same thread as RecordThread (it
> is called by RecordThread),
> so I think this is OK. But I will ask others to confirm.
>
> On Tuesday, April 28, 2015 at 8:29:48 PM UTC-7, Hongyan wrote:
>>
>> I have a question about the preprocessing effects integration model in
>>
>> current audio framework, could anyone help on?
>>
>>
>>
>>
>>
>>
>>
>> Currently Android provided 3 kinds of preprocessing effects:
>>
>> AEC/NS/AGC and they’re wrapped into libaudiopreprocessing.so
>>
>>
>>
>> When App(Eg: Hangout) created such kind effects (type == what defined
>>
>> in libaudiopreprocessing.so), effects will be created and then handle
>>
>> will passed to HAL for further processing.
>>
>>
>>
>>
>>
>>
>>
>> SO maybe audio flinger should not call process_l function anymore for
>>
>> PRE or PROC effects, right? because they’re suppose to be handled in HAL.
>>
>>
>>
>>
>>
>>
>>
>> But in bool AudioFlinger::RecordThread::threadLoop()
>>
>>
>>
>>
>>
>>
>>
>> size_t size = effectChains.size();
>>
>>
>>
>> for (size_t i = 0; i < size; i++) {
>>
>>
>>
>> // thread mutex is not locked, but effect chain is locked
>>
>>
>>
>> effectChains[i]->process_l();
>>
>>
>>
>> }
>>
>>
>>
>>
>>
>>
>>
>> All of effects will be processed no matter if it’s PRE or PROC, so
>>
>> it’s possible both HAL and Audioflinger called the process function.
>>
>> It’s not expected. Is't a bug?
>>
>
--
--
unsubscribe: [email protected]
website: http://groups.google.com/group/android-porting
---
You received this message because you are subscribed to the Google Groups
"android-porting" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
