Amazon Echo (Alexa): Laggy responses

The Alexa team is investigating this. We’ll have a resolution as fast as possible.

It’s the avoidance of the cloud that Amazon are trying to tackle. The actual sale of Amazon Echo, Dot etc are not revenue generators for them. It’s the traffic that they can potentially drive to their preferred partners is the area they are trying to exploit. Clearly the registered Amazon Skill is the conduit.

I have no doubt that Mios, having a registered Amazon skill won’t be affected in the long term but clearly the recent Amazon firmware updates have had unintended consequences to third party developers.

Alexa Gen1 would work on “non port 80” setups (e.g. docker running on 81).

Alexa Gen2 and Gen3 do not work on other ports than 80. I can imagine Amazon has aligned their firmware to only work with port 80 setups. But that’s only guessing.

I had to change from a docker setup on my nas (port 81) to a ubuntu vm (port 80) about 6-8 month ago due to havin bought 2 new gen 3 alexa’s and else it would not work anymore. My older gen1 still works on the port 80 setup.

There’s quit a lot of information on internet about this “feature”…

Yes quite a bit of technical information out in the wild that the ordinary user wouldn’t be aware of.

As Gen 1 devices approach end of life firmware changes will become more relevant as these devices are replaced.

Some users have more than one Alexa device of mixed generations and are not aware of this so the situation is not as straightforward as it appears. I imagine that’s why we are hearing the likes of “it’s working for me” and “it’s not working for me” scenarios.

Either way it’s getting interesting now as more and more users are becoming more reliant on Alexa devices.

I have a mix of devices. Not sure if my earliest is Gen 1 or 2. The same issue seems across all of them

C

As an experiment you could remove all but Gen 1 devices from your setup and run a test on that basis.

I imagine that Amazon did little or no consultation with skill designers prior to the recent round of firmware updates. That would explain the chaos that ensued.

The intent in general is very true. But unlike GH, they did not shut the door to locally command devices like hue and harmony (yet?). I have all 3 generations of devices and have had no problems with any of them (maybe because I have at least 1 gen1 device… not sure) using them with the hue emulator on port 8080. My echo firmwares are very much up to date. I think a lot has to do with the version I use of the habridge though…

1 Like

Yes you’ve identified the most likely reason why you haven’t experienced any problems so far. You have a Gen 1 on your network.

I don’t use the HA Bridge but am watching closely it’s progress. The change to port 80 for some to resolve their issues is not proof positive that the problem is resolved for everyone.

Those users would need to report exactly what Alexa devices they are using and what generation these devices are before a conclusive solution is agreed upon.

I have one Gen 1 among other devices and run HA Bridge and had the problem.

Hi everyone, we’re still looking at this but at first sight it might be something “in-house” that is triggered by the latest mms deploy. We’re not ready to jump to any conclusions yet but we’re on this.

1 Like

For me it stopped working and I had to move habridge to port 80.
I am also into the idea of cloudless and the only reason to keep the device connected to cloud services is to also have google home workin. Even though there would also be other options to do it, these are costly from a time manner (need development and some board running and a skill set up in firebase).

mms Sorin?

FWIW seems better here but could be totally perceptual

C

It is better indeed, I forgot to mention that, but we’re still detecting losses here and there, which we’re investigating and addressing.

1 Like

It takes roughly 2 seconds from the end of the voice command ‘table lamp off’ for the light to turn off here in the UK. I’m occasionally getting responses that the command has failed but this generally works the second time around.

I just had a 4 second delay. Seems about normal…

Had a 15 second + earlier.

C

It would be pretty awesome indeed to incorporate the habridge emulator concept within the firmware and have a simple interface page for it within the vera UI. That would be pretty brilliant.

1 Like

Agreed. There is always going to be some sort of cloud interaction with the echo products, but keeping the TTS path as local as possible would mean much faster response times. I have my HAbridge set up on docker, using openLuup as the source and the response times are excellent.

Also, I wrote a rough plugin that uses my echoes as a TTS endpoint via Home Assistant. It works well, but I imagine it would work much better, and more reliably if the Vera Amazon skill allowed you to do something similar. I would suggest a plugin on the front end, so that those who don’t want the functionality could opt out and would not incur the machine overhead of the service (or privacy issues for that matter) and then a connection directly from the plugin to the Amazon portal with limited involvement of any Vera servers. Or perhaps staging the Vera server on Amazon’s cloud. A plugin could broker and queue messages so that you could have a truly back and forth interactive experience, which would be a first in the automation world.

Have the same slow response with both Alexa and Google home.

I continue to get failure messages routinely from all my alexa devices. At this point they seem to fail more often then they work. I have gen 1, 2 and 3 devices and cant seem to identify any trend between the gen of the device and the failures.

Probably 10% failures at the moment. Maybe a little less…

C