Using Voice PE as a microphone and Google Nest Mini for audio output

Configuration Voice Assistant
1

Hello, I’ve been trying to solve a problem for several weeks—actually, even months.

Since my parents are old and speaking Turkish, I wanted to set up a Turkish Home Assistant in our house. I first tried using a Raspberry Pi, but its performance wasn’t enough. After reading online that many people switch to a mini PC, I did the same.

At first, I even thought the Google Nest Mini could speak Turkish, so I started with that because “ChatGPT” said it works well in Turkish—but it didn’t :D. That’s how I ended up with Home Assistant.

Currently, I can trigger many actions with remotes or buttons, but now I wanted to make it easier for my parents to use voice assistants.

My goal is to use a microphone for input and always output the audio on the Google Nest Mini.

For this project, ChatGPT and Gemini recommended the following:

  1. Connect a microphone to the mini PC and try it that way – didn’t work because the microphone wasn’t recognized.
  2. Use an ATOM Echo – didn’t work because I couldn’t separate audio output.
  3. Use an ESP32 with an IMP441 microphone – didn’t work because it didn’t recognize the microphones.

Every time, after many error messages, ChatGPT or Gemini told me that it wouldn’t work with that component and suggested I buy something else.

I’m almost at the end of my patience, and on ChatGPT’s advice I bought Home Assistant Voice Preview (Voice PE), which arrived today and has been set up.

But the same problem occurs: the audio output cannot be separated. I want to use the Voice PE only as a microphone and the Google Nest Mini for audio output.

I’ve seen some forum posts like these:

Since my English is not very good, I’ve been translating a lot, but I still can’t get it to work. I keep getting errors like:

[i2s_speaker] is an invalid option for [voice_kit]. Please check the indentation. i2s_speaker: False

or

Couldn't find ID 'internal_speaker'. Please check you have defined an ID with that name in your configuration. These IDs look similar: "internal_speaker_amp", "mixing_speaker", "internal_i2c".

Does anyone know how I can set it up so that this device is used only as a microphone, and the Nest Mini handles audio output?

ChatGPT keeps telling me to go to Settings > Voice Assistant and set where the pipeline should play audio, but I don’t have that option. I can only select the language and the services I use, like Google AI Conversation, STT Google Cloud, TTS Google Cloud, etc.

Thank you for any help!

2

What voice device are you currently using?

Setup esp device to work as normal voice device.

You will need to create action in esphome under voice_assistant section >> on_announce

The automation will send the received response url to homeassistant that generates announcement on target device.

3

I believe you can redirect replies from the Voice PE by changing it’s configuration. There’s docs about it here:

Edit: To be clear, this means changing the internal settings of the Voice PE itself, which is an ESPHome device. Apart from that, as far as I know, there’s no simple way of redirecting built-in responses in Home Assistant - voice is a bit of a black box. The only realistic option is to use custom sentences, which is far from ideal.

1 Like
4

right now, I am using home assistant voice preview, but cant change the target for media player

5 
voice_assistant:
  # ……
  
  on_tts_end:
    - homeassistant.service:
        service: media_player.play_media
        data:
          entity_id: media_player.your_home_speaker  # Change to your target speaker
          media_content_id: !lambda 'return x;'

I’m thinking something like this.
You can call HA service using api. Basically call the Media player service or similar and pass it the response url for the media for playback

6

If you install the ESPHome Device Builder app you should be able to access the configuration files of the Voice PE directly and edit them.

I don’t have Voice PE myself, but I’ve done this with other ESPHome devices.

7

GitHub - xiasi0/assist_tts_router · GitHub It’s currently just a beta version, hope it’s what you need.