2013-02-08, 22:38
Thanks a lot for your feedback.
"My C code" in terms of speech recognition is basically a continous listening example program provided with pocketsphinx. I then added a JSON-RPC layer on top of it and voila - you get xbmcvc.
Considering the above, you are right, there is not much I can do to improve recognition accuracy.
As for the segfaults, I admit I didn't test the PPA build well enough as I didn't have time to prepare a VM with proper sound integration. I did test the program extensively on Gentoo, though, and it's been some time since it segfaulted. Debugging would require creating a debug build and, as mentioned previously, I won't be able to describe and prepare that any time soon.
Same goes for your feature requests (thanks!). I'll consider them when I find the time to work on the application again.
"My C code" in terms of speech recognition is basically a continous listening example program provided with pocketsphinx. I then added a JSON-RPC layer on top of it and voila - you get xbmcvc.
Considering the above, you are right, there is not much I can do to improve recognition accuracy.
As for the segfaults, I admit I didn't test the PPA build well enough as I didn't have time to prepare a VM with proper sound integration. I did test the program extensively on Gentoo, though, and it's been some time since it segfaulted. Debugging would require creating a debug build and, as mentioned previously, I won't be able to describe and prepare that any time soon.
Same goes for your feature requests (thanks!). I'll consider them when I find the time to work on the application again.