Comparing A5 and A7 sound settings

Timo Jyrinki timo.jyrinki at gmail.com
Tue Mar 16 15:16:46 CET 2010


2010/3/16 Neil Jerram <neiljerram at googlemail.com>:
> I have to support Rui here.  I find this over-theoretical argument
> really annoying.

Sorry about that, and sorry about the term "misinformation" I used in
the first post.

> Do you understand 100% the operation of the Wolfson codec?

No. But I do believe Jöerg has better understanding about it than most
others, and therefore I would not like to dispute
http://docs.openmoko.org/trac/ticket/2121#comment:5 until there is
such empirical proof that, like I said, changes in these "wrong"
controls will create behavior that is better than is reachable by
lowering control.5 while keeping control.63, control.48 and control.12
in the "correct" settings. And note I do use quotes, I don't claim
there is one uttermost truth out there. It's easily seen in eg. that
Jöerg's ".new" file has control.12 at 7 (not to speak about the
original statefile suggested in the ticket which was quite bad), while
the conclusion in the thread (link at
http://wiki.openmoko.org/wiki/Neo_Freerunner_audio_subsystem#Alsamixer_channel_controls)
in Dec 09 pointed out that it should be 5. In the same post, btw, Al
Johnson also agrees that there should be no point in routing via PGA
(control.63).

> So, please could we have a bit more respect for people's empirical findings?

I don't mean to disrespect empirical findings, I just find that since
they are individual, they should not be generalized on a distribution
level. For example, Debian recently switched control.63 to non-Mic 2
because of an individual finding, while I think this should not be
done while the best knowledge is that control.63 should be kept at
"Mic 2", because other values mean extra routing. Even though it means
more noise, simply because it raises the mic level and control.5
should be lowered until the noise is gone.

I agree that being purely theoretical is not helpful, but I believe
it's a good foundation to having correct default settings. And by
correct default settings, I mean such that new owners of A7 phones
would have a good experience, in which case empirical findings from
A5/A6, buzz-fix or not, are not helpful when trying to make eg. SHR
and other distributions good for the occasional newbies we still have.

One problem is that mailing lists are not really optimal for
collecting experimentation results, and on the other hand when people
find something that works for them, they are probably not willing to
experiment if it would also work otherwise. But I'd be happy if
distributions would agree that proper default settings should be
directed towards A7 users, unless detection of the model is offered.

To give some more non-theoretical vibe to this thread, I started a new
empirical mic data table at
http://wiki.openmoko.org/wiki/Neo_Freerunner_audio_subsystem#Empirical_Data_for_Mic_Settings
- it's probable it will not be that helpful in searching any "one
true" defaults, but it might be helpful to have "known good
experiences" documented on a wiki, in a form of a table. I also added
an "experiment 1" there which is basically what I would think is
interesting knowledge, if people just have the patience and devices
(and subscriptions) to test it through.

-Timo



More information about the community mailing list