It’s been a busy week already – I’ve been saving up a load of “Can we write articles for your blog?” emails and this morning I decided to take a look. The sample articles were ATTROCIOUS and now I know why so many articles are bad out there – they are written to order by people who just write stuff. Well, no thanks. I’ve asked the lot of them to remove me from their databases.
In the course of the post coming in I noted a Github notice of a new version of ALEXAPI – now for those of you who don’t live in the USA or America, Amazon Alexa is a box that you can talk to and control your home, play music (theirs, not yours) and generally buy Amazon products without needing to be able to read and write. It’s a wonderful thing. ALEXAPI claims to do the same on a Raspberry Pi or similar.
I have two of the Amazon DOT devices (which are very reasonably priced at £49 as against the full on unit which IMHO is, like the Google unit and others, way OVER priced). Very happy with them. Well, I’d be happier if they had an RJ45 connection instead of just WIFI – but that’s ok.
So what do we do with these? Well, we have shopping lists, we listen to music (on Maureen’s account as she has Prime, I don’t) and we control house heating, lighting etc. All in all they’re part of the family – and the BIRDSONG skill drives our cats NUTS which is a lot of fun.
And so it was that I found this link interesting in the email this morning. With this software, a usb mic and speaker of some kind, not only can you emulate an Alexa on a Raspberry Pi but on all manner of other devices.
Now, I long ago gave up on using a Pi, £32 for the Pi, plus powered speaker, plus decent mic plus case plus power supply comes to WAY over the cost of a ready-built Echo complete with REALLY pretty lights and a microphone array.
But this morning for a moment I was distracted – one could use an Orange Pi Zero which is only a few quid!!! And off I went and followed the instructions. So – before I start, if you want to play with this – you CAN install this on an Orange Pi Zero (Armbian is good), it DOES work and you CAN completely uninstall if. Been there, done that.
I stuck in my USB audio adaptor, powered speaker, 3.5mm stereo microphone and installed the software. I filled in the details on Amazon which gave me the necessary keys to update the config on the AlexaPi. I rebooted the Orange Pi Zero and…
Nothing. As usual, audio settings. I talked to some VERY helpful people on their forums who guided me through changing the settings to the USB audio device. I tested it – audio output – perfect. Audio input – perfect. I thanked them, rebooted and…
Well, I nearly fell off my seat. “Alexa, what time is it?” – after quite some delay (this is NOTHING like a DOT which responds almost instantly) the unit came back with the time. I tried a radio station and LO, it played the radio station. By now I was getting excited… I could put up with the delay.
Sadly, and this is no reflection on the obviously hard work done by the designers, things went downhill from there.
“change accounts” I said, wanting to switch to my wife’s account as she has Amazon music. No other people in the account setup. That could have been just a technicality so I ignored that. remember I know exactly how to use these things as I have two DOTS.
“Set a timer” – turns out the unit does not do lists – most likely API-related.
“Find my devices” – this was the killer. It took some hunting down but it was actually in the UNINSTALL instructions that I found … “AVS API doesn't have a way to connect the devices to Amazon.”
That last one was the killer. That and the mic – the problem is – Amazon put a lot of work into the mics – and they handle speech over background music reasonably well. I can blast my music and shout over the couch “ALEXA – SHUT UP” – and it promptly obeys. Not so the AlexaPi – while playing the radio could I HELL get it to take any notice of me until I turned the music down – and I was only sitting 2ft away.
I uninstalled the lot – and my Orange Pi Zero seems to be as good as new. So if you want something new to play with – or if you have a very limited set of needs – have a play. For me – I’d rather fork out £49 at this point.
It was suggested that one thing this software could to that the real thing can’t – is respond asynchronously – i.e. use it to output speech without a command – that is an annoying omission on the Amazon system – though if you write a special skill you can have it say what you want in reply – but then you have to call the skill and have a secure connection etc.
But then it occurred to me – I have a Raspberry Pi running the house and HABRIDGE – and it has Node-Red and I already have IVONA speech on it – all I need is a speaker and when the DOT talks to HABRIDGE to fire an MQTT control message to my lighting and heating – it can also send off some speech itself. Don’t know why I didn’t think of that before.
So there it is.