Festival
An older text-to-speech system from the University of Edinburgh. A warmer alternative voice to eSpeak NG, with narrower language coverage.
Our verdict
Festival is the other voice. It came out of the Centre for Speech Technology Research at the University of Edinburgh and it predates most of what it sits alongside, which shows in both directions.
Its voices are generally warmer than eSpeak NG's, because they draw on recorded speech rather than generating sound from scratch. If eSpeak NG grates on you after an hour, this is the first thing to try.
The trade-offs are real and you should know them before you rearrange anything. The project's own site documents version 2.4 as current, and there has been little visible movement since. Language coverage is much narrower, with English best developed and Spanish and Welsh also present. Better voices are separate downloads, which is why a fresh install sometimes sounds thinner than people expect.
This is a choice you make inside the system speech service rather than an app you open. If speech is how you use a computer, try both voices in a live USB session and pick on your own ears, not on a description.
What's good
- The voices sound less mechanical than eSpeak NG to many ears
- Free for any use, personal or commercial, under a permissive licence
- Still packaged by the main Linux distributions, so it installs from the app store
- Can be selected as the voice behind the system speech service that Orca uses
What to know first
- Development has been slow for years, with version 2.4 the documented current release
- Far fewer languages than eSpeak NG, with English by far the best served
- Voices are downloaded separately, so the default install can sound worse than it needs to
- Slower to respond than eSpeak NG, which is noticeable when you read by ear at speed
The details
- Cost
- Free
- Open source
- Yes (X11-style permissive licence)
- How you install it
- Built-in app store
Windows apps it can replace
Last checked . Sources: