It's certainly better than most TTS out there. We are developing AI in-house to make it as humanly sound as possible. There's always improvement need to be made, but we're getting there :) Your latter suggestion is definitely worth something to look at. Let me know once you've tested it out. We highly appreciate any feedback. Cheers!

I'm curious how the economics work out with building your own TTS vs. just paying Google to do it for you. Of course Google's TTS API costs money, but I assume that you also had to pay for a dataset of transcribed speech. After how many hours of audio is the breakeven point?

The actual text to voice synthesis isn't acutally bad, but the audio quality is terrible. The audio artifacts are so bad I could barely listen to the demo snippets. It's too bad they ruined a quite good product by skimping on the most important quality, audio quality.

I wonder how OP missed this? Audio quality has to be flawless when using headphones for extended times. It's not like you need a lot of bits for audio, so it's not something you save on.

Please fix!

Sounds good on my speakers

Headphones.

It's fine on my phone and laptop. I think you are the one with the issue.

I would delete my comments if I were you because you look like a jerk mentioning it 3 times in a post with 8 comments total. And it's not even true.

the audio sounds good to me too even with headphones, are you sure you haven't got broken headphones?

Laurel or Yanny test?

it sounds great for me too. you must have headphone issues.