You can try the realtime version now in AI Studio. I'll be taking a look at it later today and if I can download the output will share here of course. Google's new TTS model just came out yesterday as well and sounds great, you can control how the model says what you type. For example, you could ask it to read in a really happy sounding voice or have it sound verry neutral with little expression.

@ZBennoui Native Audio, that is!

Yeah exactly. They have two versions, one for normal TTS and another for voice interaction, the same as Open AI's GPT4o.