[pulseaudio-discuss] [RFC PATCH 0/6] LFE filter

David Henningsson david.henningsson at canonical.com
Fri Feb 6 03:41:21 PST 2015

On 2015-02-04 18:45, Alexander E. Patrakov wrote:
> 29.01.2015 03:14, David Henningsson wrote:
>> Hi!
>> Hui and I have been working on some LFE filter patches lately, and this
>> is our first draft for review/feedback.
>> First, I have greedily stolen the math from CRAS, because CRAS is BSD and
>> as I understand we don't have a problem with merging more liberal
>> licenses.
>> The LFE filter is implemented in the resampler, which means it is done
>> for
>> every sink-input rather than every sink - this might mean some additional
>> CPU processing if many different streams play back at the same time, but
>> putting it on the sink side would disable the possibility to mix a 2.0
>> stream
>> with a 2.1 stream.
>> The rewind part is very drafty and untested, and I'm not sure I choose
>> the
>> best design here. But at least this is something that could act as
>> base for
>> discussion.
> Sorry for a possibly-stupid question, but...
> Which part of PulseAudio is supposed to disable the effect if the user
> plugs headphones in? Or is it yet to be written?

Hrm, that is actually a good question. In theory, I would expect 
module-switch-on-port-available to switch profiles between 2.0 and 2.1 
as headphones are plugged in and out, but in practice,

  - I'm not 100% sure if our "don't switch to HDMI" might prevent 
switching from 2.1 to 2.0 when headphones are plugged in, and

  - As the 2.0 profile is available on speakers, that will continue to 
be selected when headphones are unplugged.

So, while this is not directly related to whether there is an LFE filter 
or not - we already have a 2.1, 5.1, etc, profiles - indeed the problem 
might become worse with the LFE filter.

David Henningsson, Canonical Ltd.

More information about the pulseaudio-discuss mailing list