After the wakework detection there is too less time for speaking the command sentence. It occurs quite often that before speaking the command or while speaking the first word I am interrupted with the wave file configured for RECORDED WAV, followed by the error message for NoIntentRecognized. Is there a timeout that can be set anywhere for a min. time for a command or min. time between recorded sounds. I also noticed that that in the Rhasspy web gui home page the text field beside the Recognize button remains empty, so not even one word was recognized. Possibly this happens if by chance there was a short noise immediately after the sucessful wake word recognition - at least it behaves much better if I close the window.
Thanks for any idea.
I think, I found a solution. Playing arround with STT settings Minimum Duration and Skip Before (1 and 0.3) brought much better results. Apparently STT heared WAKE WAV, which I which I replaced by a wave file saying the word ‘Ja’. I will test this, especially when it becomes a little more noisy. The thread ‘Latency after wake word detection and at start of text-to-speech’ brought my to this idea.