VoxForge
Dear all,
Is there any way to take advantage of the multiple cpus in pc to train the acoustic model in Sphinx 3? I was able to do it by using Cygwin . How about without using cygwin ?
Thanks a lot,
Xinhui,
--- (Edited on 11/25/2009 5:40 pm [GMT-0600] by zxinhui) ---
suprisingly its the same as before. queue type should be posix and number of part should be the same as number of cpus.
--- (Edited on 11/26/2009 15:38 [GMT+0300] by nsh) ---
Thanks for your reply !
I did as what you said before posting this message. It worked in Cygwin, but not in Windows (without Cygwin).
When I run it outside of Cygwin, I got the followng error messages:
POSIX::setsid not implemented on this architecture at Z:\Utilities\Sphinx\tutorial\rm1\scripts_pl\lib/Queue/POSIX.pm line 66.
POSIX::setsid not implemented on this architecture at Z:\Utilities\Sphinx\tutorial\rm1\scripts_pl\lib/Queue/POSIX.pm line 66.
POSIX::setsid not implemented on this architecture at Z:\Utilities\Sphinx\tutorial\rm1\scripts_pl\lib/Queue/POSIX.pm line 66.
POSIX::setsid not implemented on this architecture at Z:\Utilities\Sphinx\tutorial\rm1\scripts_pl\lib/Queue/POSIX.pm line 66.
POSIX::setsid not implemented on this architecture at Z:\Utilities\Sphinx\tutorial\rm1\scripts_pl\lib/Queue/POSIX.pm line 66.
--- (Edited on 11/26/2009 10:59 am [GMT-0600] by zxinhui) ---
Well, you need to modify the script to spawn the processes on windows. Google is your friend here. See this for example:
http://www.perlmonks.org/?node_id=663272
--- (Edited on 11/27/2009 01:11 [GMT+0300] by nsh) ---