[pulseaudio-discuss] Microphone DC-Offset compensation and noise filtering.

[IL'dar AKHmetgaleev]

На Tue, 11 Jan 2011 11:16:06 +0100
Maarten Bosmans <mkbosmans at gmail.com> записано:

> Whoa, that's some intense filtering, just to remove DC offset! Setting
> you highpass cutoff to 100Hz is a bit much, unless you are trying to
> also do some other filtering. Moreover, why are you expanding (se4)
> and then compressing (dysonCompress) again? There are ladspa elements
> just for removing DC offset, use one of those would be more
> appropriate.

Equalizer removes not only offset but low frequency noise. This is
enough to get good sound form my mic. Other plug-ins makes it even
better for VoIP chatting.

Se4 makes same thing as skype's auto gain control. My whisper and cry
becomes to have almost same volume and very comfortable for recipient.

But after increasing gain sound still has some over-gained values.
Actually limiter may be better then compressor to remove it. 


> What do you mean by the sound becomes very nice? Do you actually hear
> the difference between the original and the corrected signal, or is it
> just that in your sound editor the line now nicely sits in the center?
> I guess that's why the ALSA guys deem it so low-priority. Its an
> inaudible in your audio stream and if you really care you can just buy
> a slightly more expensive mic.

Perhaps I'll do it. But first I would like to try hack hardware I have.
Anyway I learned that autogain is very nice feature. But not all
VoIP applications has it.

-- 
Ильдар Ахметгалеев aka AkhIL
Втр Янв 11 16:24:02 NOVT 2011
Tue Jan 11 10:24:02 UTC 2011
----------------------------------
Visit my home page http://akhil.homelinux.org
jabber: akhil at jabber.ru
----------------------------------
Ничего не важно,
потому что ничто не может быть важнее чем что-то другое.
----------------------------------
Linux artstation 2.6.32-gentoo-r24 #1 SMP Sat Jan 1 20:29:36 NOVT 2011
x86_64 AMD Phenom(tm) 9550 Quad-Core Processor AuthenticAMD GNU/Linux
up 1 day, 19:44,  1 user,  load average: 0.01, 0.01, 0.00