Sequencial sentences sent to a speaker in 1 automations does not work. All help is appreciated

Configuration Voice Assistant
,
1

I am trying to send a sequence of texts to my speaker in 1 automation.
I notices that only the last sentence is said. All previous texts are somehow skipped.

All help is appreciated!
Below the yaml code of the automation

alias: test speaker
description: ""
trigger: []
condition: []
action:
  - service: tts.cloud_say
    metadata: {}
    data:
      message: This is the first sentence
      entity_id: media_player.esp32speaker1_esp32_speaker_1
      cache: true
  - service: tts.cloud_say
    metadata: {}
    data:
      cache: true
      message: This is the second sentence
      entity_id: media_player.esp32speaker1_esp32_speaker_1
mode: single
2

Create a Script to queue the messages.

eg:

  tts_queue_kitchen:
    alias: TTS Queue Kitchen Speaker
    mode: queued
    sequence:
      - service: media_player.volume_set
        data:
          entity_id: media_player.googlehome3019
          volume_level: 0.35
      - service: tts.cloud_say
        data:
          entity_id: media_player.googlehome3019
          message: "{{ message }}"
      - wait_template: "{{ is_state('media_player.googlehome3019', 'idle') }}"

Then send your message to the Script via an Automation.

eg:

      - service: script.tts_queue_kitchen
        data:
          message: >
            "Good morning, the weather forecast for today will be a maximum of {{ states('sensor.morley_temp_max_0') }} a minimum of {{ states('sensor.morley_temp_min_0') }} and it will be {{ states('sensor.morley_extended_text_0') }}"

Since I have set the Script mode to ‘queued’ it will play any new message after the previous, because at the end of the Script it waits for the media player to be idle.

1 Like
3

Than you for helping out. I tried using a script and then had sent multiple sequensive sentences to the speaker from an automation, but this also did not work. I also set the script to “queued” but with the same results. However I saw you also check or the speaker is idle. I will try this as soon as i get my hands back on my keyboard :smiley:

4

I tried again and added the
“{{ is_state(‘media_player.esp32speaker1_esp32_speaker_1’, ‘idle’) }}”
but still the automation just skippes all commands and just speaks the last (second) sentence.

alias: test speaker
description: ""
trigger: []
condition: []
action:
  - service: tts.cloud_say
    metadata: {}
    data:
      message: This is the first sentence
      entity_id: media_player.esp32speaker1_esp32_speaker_1
      cache: true
  - wait_template: "{{ is_state('media_player.esp32speaker1_esp32_speaker_1', 'idle') }}"
    continue_on_timeout: true
  - service: tts.cloud_say
    metadata: {}
    data:
      cache: true
      message: This is the second sentence
      entity_id: media_player.esp32speaker1_esp32_speaker_1
mode: single
5

Add a delay of two seconds between the first tts and the wait template and try again

6

You can try the Chime TTS integration which has its own queue and supports making back-to-back service calls.

7

Thank you for this valuable hint. This did it!
However, since I am still learning, could you explain why the delay is needed?

8

Thank you for this valuable hint. Chime TTS looks like a very well worked out integration. Congratz. For now, I will start with using the delay of 2 sec and try the Chime TTS later.

9

Because creating the tts mp3 files is faster than sending them to the speakers and receiving then updated state of playing.
So what happens is that the first tts is sent and then because the state has not changed to playing yet it will send the second message.
You probably can use 1.5 seconds also but it really doesn’t matter, in my home 1 second or less does not work.
Perhaps it’s different for others.