:: Re: [DNG] Firefox nightly [now] req…
Page principale
Supprimer ce message
Répondre à ce message
Auteur: Adam Borowski
Date:  
À: dng
Sujet: Re: [DNG] Firefox nightly [now] requires Pulse Audio
On Sun, Oct 02, 2016 at 07:43:57AM -0500, Nate Bargmann wrote:
> I have better things to do with my time than spend it in a
> painful exercise trying to manage audio settings. I find PA truly
> useful these days. Yes, I'm using Devuan Jessie.


I for one have two[1] functional sound cards in my main desktop: on-board
and a PCI one, mostly because of falling for the 5.1 scam[2] in the past,
but even nowadays that I returned to plain stereo, that's useful because you
can reliably have headphones and speakers running at the same time.

What I want is the ability to tell programs which audio sink to use, and
occassionally switching that. PulseAudio would fit that nicely!

The usual downsides of PA don't apply for me:
* it's a desktop so I can afford constantly wasting 7% of a core (wakeups
are a murder on battery)
* I do no audio processing so PA's inability to handle high-resolution
high-frequency audio doesn't matter (remember, human ear can't hear above
16-bit 44kHz[3])
* I hardly play games so I can stand the sound lag

Thus, I'd be happy to use PulseAudio. Alas, I get:
* constant noise: quite quiet but loud enough to be infuriating, especially
when no audio is playing. Raw ALSA has no such problem.
* sound completely fucked up after a suspend/resume: it becomes high-pitched
and clipped until "killall pulseaudio" (it restarts automatically). Only
response to the bug report: "your sound card doesn't support suspend".
Which is bullshit as this problem happens with PA even if no sound was
playing at suspend time, while ALSA gets it right even if there _is_ sound
during suspend. And, since PA doesn't touch hardware directly but uses
ALSA, the culprit is obvious...

So, I don't use PulseAudio not because I don't want to, but because it's as
buggy as systemd or avahi, and the bugs are show-stoppers.


Meow!

[1]. And a HDMI pseudo-card that gets in the way.

[2]. There's no 5.1 sound sources, unless perhaps if you watch DVD movies (I
don't); also unless you can remodel your room specifically to place the
speakers right, the sound field is _worse_ than plain stereo or headphones.

[3]. Technically you _can_ hear more than undithered 16-bit, you just need a
lab-quality quiet room, high-quality speakers with volume set to near the
pain threshold, and a sample near the hearing threshold (ie, zeroes on all
but lowest bits); dithered 16-bit is ok even then. Oh wait, you _don't_
have such a lab or don't use pathological volume settings? Then 16 bits is
enough for you. As for 44kHz, Nyquist theorem says any frequency up to
22kHz can be reproduced exactly, and not a single human has been
demonstrated to hear ultrasound in a controlled test. Thus, anything above
is "audiophile" scam. Unless you do further processing, that is -- it's
easy to pile up errors, especially resampling at a different frequency
drastically reduces quality so you want to use much higher precision to
reduce the errors.

--
A MAP07 (Dead Simple) raspberry tincture recipe: 0.5l 95% alcohol, 1kg
raspberries, 0.4kg sugar; put into a big jar for 1 month. Filter out and
throw away the fruits (can dump them into a cake, etc), let the drink age
at least 3-6 months.