19 Aug
2010
19 Aug
'10
6:38 p.m.
Hi guys,
Is it necc. for me to clamp my audio outputs to [-1.0, 1.0]?
Somewhere I got it in my head that I need to clip my output
signal (float*) to the range [-1.0, 1.0] before sending my
audio signals downstream.
However, this is a big time waster. When I review other
projects for efficient ways to clip... I can't find any.[1]
Looks like nobody's doing it.
Thanks,
Gabriel
[1] I found one in ardour, but it looks like it's just
for float->int conversion, and (possibly) not even
utilized within ardour.
19 Aug
19 Aug
7:11 p.m.
Hi,
On Thursday 19 August 2010 20:34:57 Gabriel M. Beddingfield wrote:
Is it necc. for me to clamp my audio outputs to [-1.0,
1.0]?
Somewhere I got it in my head that I need to clip my output
signal (float*) to the range [-1.0, 1.0] before sending my
audio signals downstream.
No, just send your data out as is. The [-1, 1] limit only applies for jack
when the float audio is converted to whatever bitdepth-integer your soundcard
uses. Which is why some people have a post-fader limiter in their master
channel in ardour during recording/mix (but not during mixdown/export).
You should keep in mind that your app sounds better when it stays within [-1,
1] under normal circumstances. But you don't have to force that. Their wont be
any losses when your signals go beyond 1 as long as the audio stays within
jack.
Have fun,
Arnold
7:29 p.m.
On Thu, 19 Aug 2010, Arnold Krille wrote:
No, just send your data out as is. The [-1, 1] limit
only
applies for jack when the float audio is converted
Thank you!
You should keep in mind that your app sounds better
when
it stays within [-1, 1] under normal circumstances. But
you don't have to force that. Their wont be
Yes, this is understood. :-)
Thanks,
Gabriel
9:26 p.m.
Arnold Krille wrote:
No, just send your data out as is. The [-1, 1] limit
only applies for jack
when the float audio is converted to whatever bitdepth-integer your soundcard
uses.
Not just Jack, any code path where float data is converted to
a finite width integer.
Regardless of where the float to integer conversion takes place
(Jack, Alsa etc) it is at the conversion stage that the clipping
needs to be done.
Erik
--
----------------------------------------------------------------------
Erik de Castro Lopo
http://www.mega-nerd.com/
5005
days inactive
5005
days old
linux-audio-dev@lists.linuxaudio.org
3 comments
3 participants
participants (3)
-
Arnold Krille -
Erik de Castro Lopo -
Gabriel M. Beddingfield