Re: recommended Alsa version for an AWE64?

[Adam Myrow <amyrow@midsouth.rr.com>]

At 09:53 AM 10/16/02 -0400, you wrote:
>Hi,
>Well, with the2.4 kernels you aren't supposed to have to use isapnp
>tools.

I wasn't.  I'm using the native ISAPNP support.  So far, Alsa has given 
extremely mixed results.  As I stated, I had to go back to beta 12.  Some 
sounds sound better than with the stock drivers and some have a loud pop or 
squeal at the end.  Next, the whole thing about muting all channels by 
default is absolutely insane.  I had my /etc/modules.conf set up to 
auto-load Alsa and after a short time, the modules unload and when they 
reload, everything is muted again.  I got around this by manually loading 
the modules in /etc/rc.d/rc.modules so that they stay in memory.  I'll 
probably leave it that way anyway as it's faster.  Anyway, all this has 
been a let down when it comes to Speak Freely.  I've tried both the stock 
7.2 and the CVS version and had very unsatisfying results.  Both versions 
work fine in half duplex, but I may as well use the stock drivers if I have 
to use half duplex.  When trying full duplex, a person reported that I was 
sending dead carrier.  Further testing with an echo-back server reveals 
that what is happening is that I send out a long period of silence then 
audio.  Finally, it is as if the last packet gets held up until I close the 
connection.  When I kill sfspeaker, I will get a burst of sound like the 
last part of a word and then all sounds I play will play about half speed 
with a horrible clicking over the top.  It's like the old APH tape 
recorders that claimed to let you adjust the speed and keep the pitch 
sounding the same.  They produced a very similar sound.  In other words, 
Speak Freely is messing up Alsa.  I've tried every combination from the 
default flags to the -DLINUX_SMALL_BUFFER flag and removing the 
-DAUDIO_BLOCKING flag which produced noise instead.  With the stock driver, 
I could sort of get full duplex with the sflaunch front-end, but after a 
few minutes, incoming audio would start breaking up and get progressively 
worse until becoming unusable.  Sflaunch doesn't seem to work at all with 
Alsa most of the time.  Also, with the stock drivers, I got a log of 
recording overruns even when I wasn't talking.  The bottom line is I've 
never gotten satisfactory performance from the Unix/Linux version of Speak 
Freely.  I thought Alsa would be the solution, but I'm just about to go 
back to the stock driver and live with half duplex.  If anybody has 
actually gotten Speak Freely working with an AW64 in full duplex mode, 
regardless of what drivers they used, I'd like to hear about it!