When it’s cold I feel cold, when it’s hot I feel ambitious

I hve several Inkbird Temperature/Humidity sensors running over bluetooth to a esp32 OpenMQTTGateway which as the name implies turns the readings into mqtt messages that are published to my HA server. That server is running FHEM and I used the built in mqtt broker. I got a couple more sensors, to save money these were temperature only. They didn’t register as temperature sensors, I did see mqtt messages though. hmmpf. After digging around in the code I could see there was a small change in the messages “name”:”tps” vs “name”:”sps”, I was thrown off a bit as both types of sensors say IBS-TH2 on the back, after looking at the source repo for this I found that a fix had been checked in that handles both message types. I has them marked as IBS-TH1 and TH2. Labelling problem on the device maybe? No matter, once I re-setup my environment for flashing, I was able to update the esp32 with the latest code, the devices are now seen as temperature sensors, and all is well.

Sometimes you go many times around the tree only to find what you dropped right there next to where you started.

Strangeness detectors
Collage charmers
Magnetic behavior
Quarks and order

hush hush, voices carry

Mycroft now does speech to text locally. Deepspeech is running on my local system, faster than goin tou to the internet, but maybe a small drop in recognition accuracy. Next up. try deep speech on a raspberry pi so it isn’t taking up all the resources on my graphics card.

Woke up with wood

I’m 3d printing a case for a project that hopefully will end up here eventually. I’m using “wood” filiament and print very thin layers. This combination when it gets to the interior parts were the infill is printing looks a lot like a popsicle stick. Oh and the first layer, it looked like elmer’s glue with brown dye in it. It should be finished printing in the next month or so 😉  I’ll let you know when it is done

 

Ch-Ch-Ch-Changes

The level of effort to make a completely standalone Mycroft environment is pretty big. I’ve decided to change my plans a bit. I’m going to use the standard tools and migrate towards self-reliance. The first thing I am going to do is make a doorbell skill, and in the process probably create an image that doesn’t use systemd.

Paranoia may destroy ya.

so you don’t want the cloud based misery of a Corlexagle , but yet you still feel like you are missing out, what to do?

There are a plethora digital assistants out there, most run on a Raspberry Pi, are named after some super hero’s computer, and use Flite for speech output and maybe pocketsphinx, for speech input, but more likely they use Google’s speech api, so where is the win in that?

Enter Mycroft AI, named I believe after the fictional AI in my favorite book “The Moon is a Harsh Mistress”, Mycroft “Mike” Holmes IV, who is in turn named after the fictional Sherlock Holme’s fictional smarter brother. Already a good start.

Mycroft is also cloud based :(, but all hope is not lost. The developers seem to put privacy forward in their thinking, and this all open source, and the rumor is that it can all be installed locally. …just don’t expect documentation.

You wanted get started and help write mad skills for your new helper. The fast, easy (and sadly systemd loving, freedom hating) way is to download their image for the Raspberry Pi, setup an account, get a sound card, speakers, and a microphone. Just follow the instructions on their site. Their instructions are way better than I can offer, so just go the Mycroft AI site and read, you’ll want to get many details, so read ALL of it, not just enough to find the link to the image.

Now, you’ve played around with that for a bit, wrote some useful skills that I can siphon off you, and you are still *unsatisfied because the log clearly shows all your words heading off to the internet for processing. What next?

First install your favorite virtual machine software (I use virtualbox), get a Gentoo Image. , boot it up. emerge-webrsync, emerge -uDN @world, and wait. Come back tomorrow, or maybe the next day.

Back? cool. now you’ll have to wait for the next episode as I am still compiling. Gentoo, oh Gentoo, my love, my chagrin, my time sucking friend.
Continue reading “Paranoia may destroy ya.”

Mr. therMistor – take these broken… …temperature readings.

MP Select Mini V1. I had it running pretty well, then suddenly the hot end reports 0°. Test, yup, thermistor is bad. I have one on hand! I pull things apart and replace it. Now at idle it is report about 24° where as before it (and the hot bed) reported about 19°,  ok, so worst case I’ll go 5° higher. The reading if I hold the thermistor between my fingers is 33° which is pretty close. I start a print and it comes out like a lace doily. hmmm. I point a non-contact thermometer at the hot end and I can only get about 188° when the temp is set to 200-215°,.

according to the MP Select mini wiki I have the correct 3950 thermistor.  After fiddling around a bit, I now seem to have other symptoms hinting at a clog. Time to pull apart the hot end…..

Syncronicity – With one breath…

So snapcast works. it works well. The only negative appears to be that a stream has to flow through the single server, so using multiple players as a source is not looking good. An example would be syncing 2 bedrooms while the main player is serving the rest of the house. I could probably do it, but then the individual rooms would always be served through the main box, and it is only a lowly nano-pi, but it does have gigabit ethernet. hmmmm.

In the meantime this will work, one main server that can sync to any or all rooms, and the individual players can either sync to that stream or play locally.

How is is all done? well mirrors of course.

Components:

Software

IF you use an R-Pi3 for the server you can use raspbian and the install will be more straight forward. In fact O!MPD has a rasbperry pi image you can use.

There are install instructions for all the software listed on their individual sites. It turns out there isn’t anything really special that needs to be done. O!MPD using mpd, which there are setup instructions for on the snapcast github page.

The issues I ran into are power for the Raspberry Pi 2, it put noise on the power bus that was transferred to the USB audio dongle. A USB hub solved that, the other is Gentoo, installing on a Rasberry Pi is a bit of a chore, but there are instructions for that and creating a 64-bit image that runs on the Nano-pi was a bit more difficult, but I used the Gentoo instructions and the kernel from the FriendlyArm linux distro.

Why you may ask use Gentoo when the Rasbian install is so much easier, well it is simple systemd ain’t free. I can’t abide it. If you can then more power to you.

 

Post 1, in which Doris gets her oats. Yes, music is on my mind. Whole house music that is.

Instead of quoting Charles Haltrey and the deaf-aids, I suppose the “house” referenece should have gone to Hugh Laurie, but not as musical, or maybe I should have just said “inst, inst, inst”. Any way on to the action.

I’ve been using O!MPD on a Raspbery Pi for my music server. I like it, but I want more. I want to have synchronized whole house audio. O!MPD can forward the playlist to other instances of MPD, but they freewheel and don’t necessarily stay in sync.

In comes SnapCast to rescue me. It can synchronize audio playback, and it uses MPD! Hooray!. Almost.

My first issue was ‘casting from one pi to another, noise, weird modulation, icky, but the music was there in the chaos. Turned to be a known issue with Raspberry Pis. There was noise in the power to the USB audio device. Powered up solved that.

Second issue, now the music is playing fine to the 2nd pi, but it isn’t in sync with the first. O!MPD was playing to MPD, which was sending traffic to the snapcast fifo and directly to the audio device.  I removed the audio device from mpd.conf and started snapclient. No audio at all to the main R-Pi. That is where it stands. I did see and error writing to device message from snapclient when I ran it in the foreground. I checked permissions and I can play a .wav file using aplay when I am su’d to the snapclient user. hmm… I’ll let you know.

 

**Edit – so I was using the wrong sound card index. changed -s 1 to -s 2 and voila!