Hi all, I was just wondering how everyone is getting on with their satellites, and if you are having any problems when using multiple satelites?
When I started, I first set up a Jabra 510 conference speaker, plugged directly into my HA machine (HP elitedesk 800, running Proxmox), and using the Assist Microphone add-on, it worked really well, and was easily understanding my speech.
Today I received some components to make some small satellites, and just as they arrived, I found the Everything Smart Home video where he goes through how to make a satellite with an ESP32. After some soldering and 3D printing, I got a prototype working within a couple of hours. Yes, I can hear that the microphone’s not as good a quality as the Jabra, but it worked. However, not I’ve started getting problems where the openWakeWord add-on crashes when I have both the Jabra, and the ESP32 satellite active. I’ve tried making seperate pipelines for them each of them, but it’s still crashing.
I’ve disconnected the ESP32 satellite for now (4 hours and no more crashes of the add-on), so I was just wondering if others are having problems? Is it possible to run more than one satellite?
So far, from what I can see in mine is that if I use the Assist Microphone addon together with a USB conference speaker directly into the HA machine, and use a ESP32 device with a mic at the same time, then at some point, the openwakeword addon crashes. Restarting it is the only way to use Assist via the wakeword again.
I’ve disconnected the USB conf speaker for now, and am just playing with a single mic for testing, and so far no crashing. But let’s see what happens when I make an other ESP device with a mic, which will hopefully be later tonight.
Just checked OWW add-on and it had crashed. Restarted and the active Echo then responded (correctly) to WW. I then activated ‘use wake word’ switch on the 2nd Echo (different room) and OWW add-on crashed again with this error:
s6-rc: info: service s6rc-oneshot-runner: starting
s6-rc: info: service s6rc-oneshot-runner successfully started
s6-rc: info: service fix-attrs: starting
s6-rc: info: service fix-attrs successfully started
s6-rc: info: service legacy-cont-init: starting
s6-rc: info: service legacy-cont-init successfully started
s6-rc: info: service openwakeword: starting
s6-rc: info: service openwakeword successfully started
s6-rc: info: service discovery: starting
INFO:root:Ready
[18:43:24] INFO: Successfully sent discovery information to Home Assistant.
s6-rc: info: service discovery successfully started
s6-rc: info: service legacy-services: starting
s6-rc: info: service legacy-services successfully started
INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
ERROR:root:Unexpected error in wake word thread
Traceback (most recent call last):
File "/usr/local/lib/python3.11/dist-packages/wyoming_openwakeword/openwakeword.py", line 301, in ww_proc
ww_model.allocate_tensors()
File "/usr/local/lib/python3.11/dist-packages/tflite_runtime/interpreter.py", line 531, in allocate_tensors
return self._interpreter.AllocateTensors()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: /tensorflow/tensorflow/lite/kernels/reshape.cc:92 num_input_elements != num_output_elements (2 != 1)Node number 24 (RESHAPE) failed to prepare.
I then created a 2nd VA pipeline and assigned 1 pipeline to each echo (same WW though), restarted OWW add-on, reactivated ‘use wake word’ on both Echos and as soon as I tried to start a session on the 1st Echo the OWW add-on crashed again (same error).
Just changed the WW on the 2nd pipeline, and now it seems stable. I don’t want to have to have different wake words for each satellite though
I can probably cope with a different pipeline per satellite, and the Pi can probably cope with that if they all activate/deactivate based on room presence, but having a different wake word per device is not acceptable or workable!
Ive got 4 running atm with only one error in the past 3-4 days now.I had to reboot one when it went into the eternal blue light of death.That was a couple of days ago.
My HA server is on an I7 machine with 16gb RAM.
Connection to the network is huge with these things. I gave all of the satellites and the HA server priority on the network. All satellites are on the IoT wireless network with less traffic.
