I am trying to use 2 approaches:
1) forward processed sound through playAudio node to julius,it is working but i don’t have improvement of recognition compared to pulsaudio with Webrtc echo-cancel module,
2) saveHTKFeatures node and -input mfcfile at julius,
I have some questions:
– i have problem with saveHTKFeature. I am getting Segmentation fault error after few seconds of processing. Do you have any idea where the problem can be? It is both for offline (from wave file) and online from mic.
– Do you have any checked set-up (configuration) for 1) solution?
– Which mic array is better to work with sound recognition (julius) playstation eye or Kinect?