flite

lightweight text-to-speech synthesis

TLDR

Speak text

$ flite -t "[Hello world]"

Speak from file

$ flite -f [text.txt]

Save to audio file

$ flite -t "[Hello world]" -o [output.wav]

Use specific voice

$ flite -voice [rms] -t "[Hello world]"

List available voices

$ flite -lv

Speak input as phonemes

$ flite -p "[pau hh ax l ow w er l d]"

Read SSML input from file

$ flite -ssml -f [input.ssml]

flite (Festival Lite) is a small, fast text-to-speech synthesis engine developed at Carnegie Mellon University. It converts text to audio using concatenative synthesis, producing speech from recorded fragments.The engine is designed for embedded systems with limited resources, providing reasonable quality without large runtime requirements. Multiple voices are available with different characteristics.flite works offline without internet connectivity, making it suitable for accessibility applications and audio generation.

PARAMETERS

TEXT

Text to speak. If it contains a space, it is treated as a literal text string rather than a filename.

-t TEXT

Explicitly set input text string.

-f FILE

Explicitly set input filename.

-o FILE

Output audio to file (WAV format). If omitted or set to "play", audio is played on the default audio device. Set to "none" to discard output.

-p PHONES

Synthesize input as phonemes.

-voice NAME

Voice to use (name, filename, or URL).

-voicedir DIR

Directory containing voice data.

-lv

List available voices.

-ssml

Read input text/file in SSML mode.

-b

Benchmark mode.

-l

Loop endlessly.

-s F=V

Set feature to value (guesses type).

-v

Verbose mode.

--version

Display version number.

--help

Display help information.

INSTALL

sudo apt install flite

sudo dnf install flite

sudo pacman -S flite

sudo apk add flite

sudo zypper install flite

nix profile install nixpkgs#flite

CAVEATS

Limited voice naturalness compared to neural TTS. Few voice options. Output quality varies by text type.

HISTORY

flite was developed at Carnegie Mellon University as a lightweight version of the Festival speech synthesis system. It's used in accessibility applications, embedded systems, and offline TTS scenarios.