eSpeak

eSpeak is a compact open-source text-to-speech (TTS) engine. It uses formant synthesis, generating speech by modeling the resonant frequencies of the human vocal tract rather than concatenating recorded speech segments. This design yields a small footprint and fast startup, making it suitable for embedded devices and screen reader applications, though the resulting voice often sounds robotic.

The project was originally developed by Jonathan Duddington and released as open-source software. It provides cross-platform support for many languages and variants, and can be used via a command-line interface or integrated as a library into other applications. eSpeak accepts plain text input and can emit audio in WAV format or pipe it to speakers. It also supports phoneme-level input, allowing finer control over pronunciation, and can apply language-specific spelling-to-sound rules.

Language coverage includes a wide range of languages, with separate data files for each language and accent. While it offers broad linguistic coverage, the naturalness of the voices varies by language and is generally less natural than concatenative or neural TTS systems. The software is distributed under an open-source license and is maintained through community contributions. A widely used fork called eSpeak NG continues development and adds new languages and features.

Users include Linux distributions, screen readers, and other accessibility tools, and the project is available on major platforms such as Windows, Linux, macOS, and Android.