Hi, everyone!
I’ve upgraded the OpenAI TTS addon.
New voices have been added for broader expressiveness and variety.
A new speech model has been integrated for better quality and realism.
An instruction field has been added to the voice settings — you can now customize tone, accent, emotional style, intonation, and more.
You can try out the voices here: https://www.openai.fm/
Here’s a compiled JAR for testing.
Hi! Great stuff! 
How to specify openai tts in the say command? Is it possible to specify other language than english?
Hi. Unfortunately, the voice can only be configured once, and there’s currently no way to change its parameters with each request to OpenAI. The service understands almost any language — that’s the advantage of this binding. You can speak in one language and get the response in another (when used together with the ChatGPT HLI). So you don’t need to switch or select a different voice for each language separately.
Ok, thanks, that sounds good. But if I would like to output “say” to two different sinks, how do i do that?
This question isn’t related to this binding. Please check the forum for existing answers or open a new topic if needed.
Agree, but I could not find the name of the service to use with the say command. After listing voices in karaf I found it.
openhab:voice voices
OpenAI TTS Service - engelsk - Alloy (openaitts:alloy)
OpenAI TTS Service - engelsk - Echo (openaitts:echo)
OpenAI TTS Service - engelsk - Fable (openaitts:fable)
OpenAI TTS Service - engelsk - Nova (openaitts:nova)
OpenAI TTS Service - engelsk - Onyx (openaitts:onyx)
OpenAI TTS Service - engelsk - Shimmer (openaitts:shimmer)
So the name that has to be specified in the say command is openaitts together with the name of the voice. E.g. “openaitts:nova“
You can set the default TTS and default voice in Main UI → Settings → Voice. After that, you can use say without any parameters.
As long as i need to output to different sinks and not the default one i could not find a way without specifying voice.
Is this really a speech to text (STT) addon ?
It get the impression that it’s a text to speech (TTS) one.