Fast Facts
- Read text out loud using an AI voice
- Clone voices using 2-3 hours of speech
- Create custom pronunciations of key terms or acronyms
- Hundreds of off-the-shelf voices, ready to use
- Covers 130+ languages
- Voice cloning requires lots of audio
- Voice quality is not best-in-market; feels robotic
Free plan; $30-50/month
Our Review
Play.ht is another option for synthetic voices. While we like ElevenLabs better, it's also worth looking at Play.ht as another option. Here's why:
Presets and international
ElevenLabs is fun if you're a tinkerer and willing to play around with generating a voice, finding one that you like and seeing it read a script that you write. It's very much an "experience AI" moment.
But if you just want to make a decent voice say something, Play.ht has 800+ prebuilt voices that you can explore and use off-the-shelf.
The quality is good, but not the best. We think the voices don't have enough emotional / tone inflection to sound truly real - or maybe our AI radar has just gotten too good recently.
Beyond this, Play.ht also offers 130+ languages, while ElevenLabs currently has 8 languages. We haven't verified the quality very extensively, but this is helpful if you need lots of voices in aparticular non-English language.
It's still early
We'd be lying if we said there's a clear winner for AI voices. Play.ht is building all the useful tools you might want related to voice cloning (although it requires 2-3 hours of audio), like speech synthesis, voice cloning, etc.
We'll stay tuned to how Play.ht develops from here.