Change TTS model

[JuanJose2325]

Hello, this is my first time here, and I would like to ask a question. I hope this is the right place.

I have OVOS installed on my Raspberry Pi 4, and I would like to change the voice model to one that is in Spanish.

I tried everything, but I didn't get it. I don't even know which voice plugin is being used, as both mimic3 and piper have the ‘apope_low’ voice model with different names, but they sound the same.

I tried to look at ‘.config/mycroft/mycroft.conf’, but for some reason always rebooting the raspberry pi deletes its content, and only the ‘PHAL’ part remains:
{
‘PHAL": {
ovos-PHAL-plugin-wifi-setup": { “ovos-PHAL-plugin-wifi-setup”: {
first_boot": { “first_boot”: false
}
}
}

I tried to install three different OVOS images (the latest multilingual version, the preconfigured English one and the preconfigured Spanish one), and all of them on first boot failed trying to start ‘ovos-listener.service’. The only one that worked was the Spanish preconfigured one, once almost all the contents of ‘.config/mycroft/mycroft.conf’ were deleted, as I mentioned before.

In advance, thank you very much for the help.

[timonvanhasselt]

Hi, for hands-on support check this active community forum of OVOS:
Matrix Chat rooms

[Swotboysandy]

Welcome 🙂

Yes, this is the right place to ask.

On Raspberry Pi OVOS images, a few things can be confusing at first, especially around TTS selection and config persistence, so you’re not doing anything “wrong”.

1. Which TTS is actually being used

On the current OVOS Raspberry Pi images, Piper is the default TTS, not Mimic3, even if both are installed.
That’s why the apope_low voices sound identical — they are effectively the same underlying voice in Piper.

You can confirm this by running:

systemctl status ovos-tts-piper

If that service is active, Piper is what OVOS is using.

2. Why mycroft.conf keeps getting wiped

On the Pi images, ~/.config/mycroft/mycroft.conf is not meant to be edited directly.

On boot, OVOS regenerates it from internal defaults and device provisioning.
That’s why everything disappears except the PHAL block.

This is expected behavior.

Instead, custom settings should go in:

~/.config/mycroft/mycroft.conf.d/

Create a new file there, for example:

~/.config/mycroft/mycroft.conf.d/tts.conf

OVOS will merge this on every boot and it will not be deleted.

3. Setting a Spanish voice (Piper)

First, list available Piper voices:

ls /usr/share/piper/voices/

Then create tts.conf with something like:

{
"tts": {
"module": "ovos-tts-piper",
"ovos-tts-piper": {
"voice": "es_ES-sharvard-medium"
}
}
}

(Replace the voice name with one that actually exists on your system.)

Restart OVOS:

sudo systemctl restart ovos.service

4. About the ovos-listener.service failures

This is a known issue on some Pi 4 setups, usually caused by:

missing or misconfigured audio input

ALSA device mismatch

first-boot race condition

That’s why the Spanish preconfigured image “worked” — it ships with fewer services enabled by default.

If listener fails again, check:

journalctl -u ovos-listener --no-pager