No hotword detected and manual wakeup recording sounds interrupted

Hi everybody,

in version 2.4 I had everything working, but with 2.5 I was struggling to get my microphone working. Now it finally does work - kind of.

When the docker container is stopped I can simply record something with

pi@raspberrypi11:~ $ arecord -D plughw:CARD=MATRIXIOSOUND,DEV=0 -d 5 /tmp/test.wav
Recording WAVE '/tmp/test.wav' : Unsigned 8 bit, Rate 8000 Hz, Mono

And play it using:

pi@raspberrypi11:~ $ aplay -D "sysdefault:CARD=Device" /tmp/test.wav
Playing WAVE '/tmp/test.wav' : Unsigned 8 bit, Rate 8000 Hz, Mono

The sound quality is ok.

Using arecord as mic engine in rhapsspy I selected “hw:CARD=MATRIXIOSOUND,DEV=0”. I wasn’t able to get it to recognize any wake word (with multiple engines).
When using the manual wake up on the webpage to speak a command it’s kind of working. The accuracy of the recognized text isn’t very good. When I click the “Play the last recorded voice command” next to it it does play my input, but it sounds like it’s in some kind of slow motion with little interruptions in between.

Does anybody have a clue what that could tell me?