Play.ht: Ultra realistic text to speech voice (and demo!)
I've been meaning to make sure I wrote about what the team at Play.ht have been creating.
The company offers AI Text to Speech, AI Voice Cloning and a Voice Generation API – and I have to say, the demos I've tried are really, really good.
They offer over 800 natural-sounding voices that all come with surprisingly effective human intonation. I had to do a double-take earlier when I heard it ('her') taking a breath before it (she?) spoke! They also offer support for a whopping 142+ languages and accents.
The key thing for me is all of this can be driven by API, meaning the applications are limited only by your imagination.
Yes, you can log in to their web interface and play around – that's what I did – but I suspect the real magic happens when you're plugging in your existing applications and code to Play.ht.
On the Play.ht web interface, I used my free account credits to make you a little demo. Here we go - please do have a listen to the 22 second .wav file:
Definitely natural and realistic. That was generated in a second or so.
Pricing is very reasonable too. It's free to get started - as you can see from my demo. You get 12,500 of characters per month and one instant voice clone (which I haven't tried yet) along with access to all voices and languages. See below for the basics for their studio offering that I was playing with:
The API pricing is similar:
I love the fact they've got a 'Hacker' price plan at $5. That's a good incentive for me to investigate playing about with it myself. I might very well try that. What a good deal.
Their $999/month option, offering 10M characters of text to audio per month, means adding voice directly into your application or service is exceptionally accessible.
If you're at all interested in this space, I'd definitely recommend checking out the Play.ht offering.
If you're curious, do what I did, make a free account, type some text, generate it – and then send it to your team and blow their minds (or show them what's possible).
I also had a play with their Play.ai Voice Interface which was fantastic. I'll write another post about that shortly.
If you're looking for contacts at Play.ht, here are some suggestions:
Here is Hammad Syed, Co-Founder:
And here is Mahmoud Felfel, Founder:
A quick side note: My son Freddie really enjoys flipping through YouTube now and again. We do limit the time – and he's great about respecting those limits. However, he absolutely loves those mindnumbing AI-created-crappy-voice videos. I think the poor production quality is what annoys me the most. Then, after that, I'm annoyed that he's sitting watching something that ChatGPT threw together to monetise his attention. The ultimate author hasn't upgraded to using Play.ht yet. I think I'd have less of a problem if the quality was better.
