v0.10.2-beta2

[nimroddolev]

✨ Features

• The chime_tts.say and chime_tts.say_url services have a new Audio Conversion parameter to convert audio to work with Alexa speakers. It can also be utilized to specify an arbitrary list of arguments for an FFmpeg conversion job.
• The duration of the generated audio (in seconds) has been added to the chime_tts.say_url service response (duration), eg:

  url: https://{YOUR_HA_ADDRESS}/local/chime_tts/x3u5vr88.mp3
  duration: 4.883
  

🧰 Improvements

• Added support for setting file type in FFmpeg conversion
• Added drop down UI to convert audio to Alexa speaker specifications, or to set custom arguments to run an FFmpeg conversion
• Increased volume of whistle chime

🐞 Bugfixes

• Fixed bug where chime_tss.say queue would become stuck until Home Assistant was rebooted
• Fix for constructing folder paths
• Potential fix for external url
• Potential fix for stuck queue

What's Changed

• Update README.md by @nimroddolev in Update README.md #61

Full Changelog: v0.10.2-beta1...v0.10.2-beta2

[KEV-github]

Good morning and thanks for all your help.
I have updated to the latest beta. Some nice changes in there.
Unfortunately mine is still generating HTTP local address.
I bet it's something simple that's stopping it. 🤦🏻‍♂️

[nimroddolev]

Thank you, i'm sorry you're still having issues.

I remembered that the configuration.yaml can include your internal and external addresses.

Can you please try adding the following to your configuration.yaml, restart Home Assistant and try again?

homeassistant:
  external_url: "https://_____________.nabu.casa"
  internal_url: "http://192.168.1.97:8123/"

[KEV-github]

Yipeee. We have lift off. That worked. Now to work out the correct yaml config to announce on Alexa.

[nimroddolev]

Excellent!

It should now just be a simple matter of using the url result from the chime_tss.say_url in the notify.alexa_media service:

sequence:
  - service: chime_tts.say_url
    data:
      chime_path: bells
      tts_platform: google_translate
      audio_conversion: "Alexa”
      message: "Hello world!"
    response_variable: chime_tts
  - service: notify.alexa_media
    data:
      message: <audio src= "{{chime_tts.url}}"/>
      target:
        - media_player.echo_speaker
      data:
        type: tts

[dbullendsl]

looks good, no issues

[nimroddolev]

Just to clarify: do you mean the above YAML sequence works for you and you have playback from on your Alexa speakers?

[KEV-github]

finally got it working

[nimroddolev]

So what was different?

[KEV-github]

I have replied in the other thread. I was copy and pasting the config so I decided to create it using the visual editor and it worked.
Doesn't look any different from what I can see🤷🏻
Thanks for all your help

[dbullendsl]

First my apologies for not being thorough, it was early and I didn't get much sleep.
I was testing without actually pushing the output to the Alexa's. First I mentioned somewhere that there was a problem not translating the [chime_tts.url] - that was my mistake there is no problem there, I mistakenly defined the response_variable incorrectly.

But I am having an issue with beta 2, Alexa doesn't like the URL generated - I tracked down the problem - with beta 2 the URL generated has the port number attached to the server name and in beta 1 it does not. My theory is this, I have an SSL certificate and use duckdns.org to create my fqn, beta 2 creates an output of https://blahblah.duckdns.org:443/local/chime_tts/faser3.mp3 when I push stuff to Amazon for the Alexas Amazon already knows my port number and somehow tacks it on resulting in https://blahblah.duckdns.org:443:443 etc.

The output from beta 1 does not add the port number and the playback works perfectly.

I see that there's a new version 0.10.2, I haven't tried it but I assume it'd have the same issue.

[nimroddolev]

Hi @dbullendsl, I pushed out v0.10.2 to all users based on your positive feedback that it was working, so please in the future take your time to test well before reporting back to me (I don't have any Alexas so I am relying on your feedback 👍)

Regarding the duplicated port issue you reported: I don't understand why you're saying the latest beta introduced the issue, as the MP3 output from the service call to chime_tts.say_url is valid: https://{domain}:{port}/local/{mp3_filepath}, and notify.alexa_media receives that valid path.

Can you please run the same sequence (chime_tts.say_url --> notify.alexa_media) on both v0.10.2-beta1 and v0.10.2 and report back if they're different, and if so the debug logs for both attempts?

[dbullendsl]

Again I'm sorry, I can't duplicate the problem - release v0.10.2 works perfectly -- I have done some cleanup of the tts output file as it was getting full of test files so maybe that was the issue all along. Again it works perfectly and I apologize for raising a false issue.

BTW I've been experimenting with Piper and your integration works perfectly with it as well. Even changing the voices - now I can use custom voices for different announcements.

[nimroddolev]

👍 I’m glad it’s working for you!

How are you specifying the different voices for Piper? I am currently working on improving the options parameter parsing.

[dbullendsl]

you have to use the voice profile in the Piper integration, so right now it's a bit awkward - look up the voice and copy it's full name, but then in your integration check the options box and paste in the info, like this:

options: "voice: en_GB-southern_english_female-low"

the voice names can be found here: https://github.com/rhasspy/piper?tab=readme-ov-file

scroll down to the download voices link, pick your language, in my case "en" then you can choose your dialect. For english either en_US or en_UK , once on either of those pages you'll find a list of the available voices. Pick one and click, some will offer only one quality choice others more. Once you've clicked thru to the voice qualtiy page there'll be a short list of files - and they'll be in the format en_US-voice-quality.onnx . Just copy the name portion, ie en_US-voice-qualtiy and use the options field in chime_tts.say or chime_tts.say_url as shown above.

Piper loads the voice into the system the first time it's used so it may take a short delay to get a response, but after the first time it's preloaded it is very fast.