Alan Cox wrote:
>
> > This is pretty much same situation as with usb speakers (my have
> > actually choice 44kHz or 48kHz). usb audio contains ugly routines to
> > do 8bit->16bit (and similar) in kernel, take a look.
>
> That is the sort of thing I want to avoid
I wouldn't object a patch to take them out 8-)
However, the much uglier thing is the mmap support plus FIFO
buffer afterwards, this would be a more worthwile goal to take
out, but mmap seems to be important for quake 8-))
> esd is not that good at 44.1->48Khz shifting however. And doing it right
> seems to be a science unto itself. I'd been assuming it was easy until I
> looked at the stuff folks do for this.
As long as the sampling rates are exactly known, it isn't really difficult
(in userspace, where you have access to the FPU/MMX units). The difficult
issues are measuring sampling rates with enough precision in
plesiochronous clocking environments, which we happily only have to
deal with if we want to copy an audio stream from one soundcard
to another. I have some code that does this, it is however not
really stable (has a pole on the unit circle) and has some 200ms
latency, which makes watching TV somewhat funny :)
Tom
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Apr 15 2000 - 21:00:20 EST