eSpeak NG
The compact speech synthesiser that gives Orca and other Linux tools their voice. Over 100 languages, and free.
Our verdict
eSpeak NG is the voice you will hear when Orca speaks, unless you change it. It is not something you use directly. It sits underneath, plugged into the system speech service, and the screen reader asks it to say things.
Two qualities make it the default almost everywhere. It covers over 100 languages and accents, which no commercial voice comes close to. And it stays intelligible when pushed to high speaking rates, which is exactly what experienced screen reader users do once the novelty wears off.
The price of that is the sound. It uses formant synthesis, which is a way of generating speech from scratch rather than stitching together recordings, so it sounds like a machine and makes no apology for it. Some people find that restful and precise. Others find it hard going and would rather have a warmer voice.
If the voice matters to you, listen to it in a live USB session before you decide anything. Other engines can be used instead, and the system speech service is where that choice is made.
What's good
- Over 100 languages and accents, which is far wider coverage than most commercial voices offer
- Stays clear at high speaking speeds, which matters if you read by ear all day
- Tiny, so it responds instantly and runs on very old hardware
- Already installed as the default voice on most systems that ship Orca
What to know first
- It sounds synthetic, because it builds speech from formants rather than recorded human speech
- Not the voice to pick if naturalness matters more than speed to you
- It is a voice engine, not an app, so there is nothing to open and click
- Last stable release was 1.52.0 in December 2024, so development is steady rather than brisk
The details
- Cost
- Free
- Open source
- Yes (GPL-3.0-or-later)
- How you install it
- Built-in app store
Windows apps it can replace
Last checked . Sources: