It's been a while now but once out of curiousity I tested the text to speech functionality (with my eyes closed) on an iPhone and another device (a laptop maybe?), and found iOS to be more advanced but still lacking. I don't know much about this but it's nice to see improvements. GrapheneOS seems to be a high value project in terms of dollars-to-output.
Does it handle multilingual scenarios? For example 'Your line Сливница - Летище София is arriving'? Don't need perfect pronunciation, need it just to not spell it out letter by letter or hallucinate... (Matcha-TTS' demo on HF is broken and couldn't test for myself)