Hacker News Reader

The major advancements in text-to-speech we've since in the last year have me planning on revamping Rad's voice (maybe basing it on Tom Sellek...who knows) but the fact we've come this far has really opened a world of opportunities for audio-first experiences.

View on HN

You could use unreal speech for this. Not a plug but i found it really cheap to make realistic tts using an api using it.

https://unrealspeech.com/

This looks (or sounds, rather) exceptional.

Long term I'm going to go with something self-hosted using Tortoise TTS etc but I also want to break away from Google's Cloud Speech as that can cost me hundreds a month when the user base has a peak so this looks great, thank you!

I’d be very interested to see what’s possible taking SOTA models and trying to get some version of them working offline, on-device.

Reply on news.ycombinator.com