Can Rhasspy's Larynx be used as the TTS for Home Assistant?

Help
,
1

So, I have Rhasspy running with the HASSos add-on with Larynx for TTS which sounds amazing, thank you @synesthesiam! Now, I’m wondering if I can set Home Assistant’s configuration.yaml to use Larynx for its TTS instead of google_translate. If so, what would I put in as the platform and base URL etc?

2

You can use the openTTS addons:

Install the addon with your language and use this as tts in Home Assitant:

tts:
  - platform: marytts
    port: 5500
    voice:larynx:harvard-glow_tts
1 Like
3

Awesome! Thank you @romkabouter!

4

It seems that HA still won’t play the TTS audio as there isn’t a media player element I can select without installing a VLC add-on and another addon for telnet to access it, apparently. It’s weird though. Rhasspy has no problem playing its TTS audio through the speakers without any of that.

5

I do not know what hardware you run. If you use TTS on Home Assistant, HA tries to play the speech on the audio device on the device.
I run HA on a vm in proxmox without speaker, so I have no use for TTS at all.

6

Yeah I’m using the Rhasspy Home Assistant add-on on an RPi 4. The challenge is that in Home Assistant, it seems to require a Media Player element for TTS to be played through. The OpenTTS add-on doesn’t seem to add one. I’m not quite sure what the simplest way is to add a media player element to just play the TTS through the Pi’s headphone jack, just as Rhasspy seems to do by itself without any further config.

7

Is possible to use whit HA core?

8

If you run the Docker image from the addon manually, yes.

9

Dear all,
I´m trying to setup the TTS and a media player. In my configuration.

  • Rhasspy is a in Pi 4 connected to a remote supervised HA instance on a different machine.
  • I don´t use SSL because I´m completely on local.
  • To make use of Larynx on PI4 had to setup rhasspy on a 64bit bullseye raspios (desktop version that comes with VLC, to be not confuse with TTS or media player). Larynx works well at low quality (no big delay) in this configuration.

I ´ve read a lot of staff either on this community than in home assistant. It seams to use Rhasspy as “media player Alexa or Google type entity” in HA we need to configure a custom “platform: notify” entity first. Did anyone try this?

Currently I use a rest_command, to sending as “service” from HA the text to Rhasspy API.

Would give it a try later on and eventually post the results, it is not really easy to found docs and/or guides on this.

10

I have 2 satellites i am playing with and 1 main node. Here is my setup

1 main satellite
2 dummy satellites
opentts running on a completely different machine

In HA I have a service created.

rest_command:
  rhasspytts:
    url: http://192.168.2.80:12101/api/text-to-speech?siteId={{ siteIds }}
    method: POST
    payload: "{{ payload }}"
    headers:
      accept: 'text/plain'
      Content-Type: 'text/plain'

Note the URL. That is the text to speech endpoint. So you send a payload to it, it sends it to the opentts container, and back. BUT if you look at the end of the url, you will see {{ siteIds }}

When I call the rhasspytts service from automations or whatever it looks like this.

      service: rest_command.rhasspytts
      data:
        siteIds: sat1
        payload: 'The washer is done , but not moved to the dryer. , , , Do you want mildew smelling clothes? , , ,  Because thats how you get mildew smelling clothes. '
        data:
          type: tts

By passing the name of the satellite I want i can tell the master rhasspy which satellite to playback on.

A strange note, I have seen in some stuff for alexa and what not that you can get the tts systems to do a pause by adding a hyphen - in the text. Seems Rhasspy didn’t really do anything with that, but it DOES add pauses via commas. So I just played around with this passive agressive washer notification until it sounded good.

Not sure if that’s what you were looking for or not.

11

Hello CrankyCoder,
Thanks a lot for the useful hint, TTS wise I did set up as per your advice and we are fine :slight_smile: it works like a charm, thanks again :slight_smile:

With this I guess will be possible to write a python script to use wikipedia and give Rhasspy the functionality to replay to some “curiosity request” or tell a joke… I mean some “personal assistant” skill… that honestly, once I figure out how to run slot programs or scripts I can easily adapt from something I developed in the past… (will post when getting some results a guide and the code used).

I did setup the main Rhasspy server on PI4 with 64bit Raspian and Larynx TTS works much better "low resolution mode) and PI4, got to say it sounds quite good :wink:

I would be interested to know how to get “Rhasspy” to play a songs (not create a playlist or something like this, as per it is well documented) or play the news from a url… like as a “media player”. I did understood that Rhasspy has no build in functionality for this, and I´m wonder what is the best known method to accomplish this task. I guesses it would be interesting to use HA media for “local stored” medias, and some script to grab medias from the internet, for instance this is the goal I would like to set here.

Thanks once again :slight_smile:

12

I am looking into the same thing. Not sure rhasspy can play the song, but I have a few pages and tutorials open for MPD + Rhasspy on the same box. I am also looking to see if there is a way to have rhasspy on something like picoreplayer (light weight logitech media server client)

But I have a feeling I will end up doing 2 pi’s in some rooms. 1 for rhasspy, 1 for music playback and whole home audio.

13

You can send wav file with HA Container and Rhasspy like this Play audio file from Home Assistant Container - #4 by DamienKoala - Configuration - Home Assistant Community