[fonzie]

Is there any way to have this running on 2 different Kodi installations with 2 different wake words?

Yes. That's actually the way you're supposed to do it. You'll set up two different servers and make two different skills.

[newoski]

Is there any way to have this running on 2 different Kodi installations with 2 different wake words?

Yes. That's actually the way you're supposed to do it. You'll set up two different servers and make two different skills.

Same Heroku app or unique per setup?

Sent from my HTC6535LVW

[jingai]

Two separate deployments.

[iophobia]

Guys, thanks for creating this skill. My Echo Dot is on its way and I was looking up stuff to control it with. Kodi was an obvious choice. I like how much you're able to do with this skill already.

What I found annoying from the demo vids, though: Do you have to use the wake word and invocation with every command? I saw that guy on YouTube that solved it very smooth and keep Alexa listening which can be done by

...to keep asking for the next command, use response.ask() instead of tell(), that keeps the session alive.

His setup seems to be an individual one as he couldn't link me a How-to just this:

I'm sorry I don't have a tutorial for that setup. But there is also a community projects available to control kodi using alexa. In my case, I submit a MQTT message to AWS IoT Hub from the lambda function invoked by the voice skill, and at home in the local network I have a nodejs based listener connected to the AWS IoT hub, which forwards received commands to the Kodi REST api.

Is it possible to incorporate this or to make the difference between "ask and tell Kodi to"?

Keep up the good work, otherwise.

[jingai]

Two things. First, there needs to be context, since the user (like myself) might have more than one copy of the skill (for more than one instance of Kodi). So, "Alexa, ask <skill>" is how we provide that context.

Second, it appears from the video that what he's doing is keeping the session open. This is a decent idea when you're by yourself and have control over other sources of audio in the vicinity, but if you have say, family in the room chatting, this will cause misfires.

Basically, the reason we don't do this is the same reason Alexa doesn't do it for any of the built in stuff she can process: without first hearing the wake word, it's impossible to determine if what she heard was simply humans talking to other humans (or other noise), or if it was a command directed to her.

I would love for it to work as you see in the video, but ultimately, Alexa has no idea of context. This is something that would have to be addressed by Amazon to get right.

[iophobia]

Understood. It's probably another 3 years or so until she'll be able to recognize context but I'm pretty confident with all the neural netword development in recent years. Also I was wondering why Alexa doesn't recognize different pitched voices as different users as this should be easy to implement by frequency.

I'll try this skill nevertheless

[Jagsta]

None of the MQTT integration is required for that to work, although that's the way I run the skill and from a security perspective it's preferable to port forwarding to either your Kodi box or the skill if you're running it locally. I have to admit watching that video that using an ongoing session to browse your library is pretty attractive.

[jingai]

FWIW, for me at least, it's not as cumbersome as you might be imagining. Unless you're just playing around with it, you're likely to issue only one or two commands for each video you watch.

Admittedly, UI navigation is something I don't do by voice, partly because of the requirement for the invocation phrase. But even without it, navigating is a pain via voice.. I can't imagine scrolling through a list by repeatedly shouting "down down down down" at her lol.

Ultimately though, I'd rather solve this problem with more direct commands. Rather than navigating the UI like you would with a remote, I'd like to see more commands that directly take you to the media you want smartly. A lot of these types of commands are already present: "Alexa, ask Kodi what new movies do I have?" then "Alexa, ask Kodi to play the movie Finding Dory" for instance would be a sequence I commonly use.

@Jagsta, you could pretty easily modify the skill if you wanted to play around with keeping the session open. I think you'll find that the misfire rate is pretty high though unless you're the only one in the room.. so it's unlikely we'd make this default behavior even if we ever did support it.

[CaptainMoody]

Following edru's guide on page 10 to setup Apache on my home server and have some questions:

Where should my port forwarding point to?

Code:

KODI_ADDRESS
KODI_PORT
KODI_USERNAME
KODI_PASSWORD

Should be my kodi box details I guess?

What should my endpoint URL be in Alexa?

Thanks in advance![/php]

[jingai]

@CaptainMoody, endpoint is your external IP address (so, you'll need DDNS if you don't have a static IP from your ISP). Note that HTTPS with a valid cert (from a CA that Amazon 'recognizes') is required.

The KODI_* environment variables should be the LAN IP (since you're self-hosting the skill locally), port, username, and password you set up in Kodi.

[CaptainMoody]

Thanks.

OK - I'm getting there.
Kodi details are correct so it's the endpoint I'm struggling with. I already have a dynDNS address setup and in use.

So if my dynDNS address is "blabla.dynDNS.com" my endpoint would be "https://blabla.dynDNS.com/kodi-alexa"?
Do I need to setup any port forwarding?

Also, I've followed the instructions in the post for setting up a SSL certificate, copy and pasted into the skill setup page. Should the little tick go green? It's still grey.

[iophobia]

@Jingai: So what about PVR related tasks? AFAIK it's not possible to directly ask Kodi to start a certain channel from the home screen, or?

[jingai]

Yes, you need port forwarding so you can access your Apache webserver from outside your LAN. It should be port 443.

I do believe it should go green as well. I assume you're using a self-signed cert if you're talking about copying it into the skill. It might stay grey for that.. not sure. I use certs signed by letsencrypt (free).

[jingai]

@Jingai: So what about PVR related tasks? AFAIK it's not possible to directly ask Kodi to start a certain channel from the home screen, or?

Not in yet, but there is an initial effort up as a pull request. I think it will need a little tweaking before it's ready (some stuff is hard-coded that maybe shouldn't be, etc), but it'll be in eventually.

The biggest problem for me is simply that I don't have a PVR set up, so if I'm the one reviewing it will take a little longer.

[iophobia]

@Jingai: So what about PVR related tasks? AFAIK it's not possible to directly ask Kodi to start a certain channel from the home screen, or?

Not in yet, but there is an initial effort up as a pull request. I think it will need a little tweaking before it's ready (some stuff is hard-coded that maybe shouldn't be, etc), but it'll be in eventually.

The biggest problem for me is simply that I don't have a PVR set up, so if I'm the one reviewing it will take a little longer.

If that is in, I agree with you wholeheartedly that just with a single command or two, one should be able to go everywhere in Kodi.