Formant (robotic) TTS for 100+ languages. Useful for placeholder VO, accessibility debug lines, and tiny footprints where neural models are overkill. The program is GPL-3.0 or later (COPYING). GitHub README also notes a BSD-2 getopt.c exception on Windows. Pages fetched on 2026-08-24 do not say the WAV you synthesize is yours. Shipping the engine inside a closed game is a GPL event. Shipping a bounced line may not be; that split is inferred. Prefer piper-plus / kokoro-82m when you need MIT/Apache TTS code.
At a glance
Licence
GPL-3.0-or-later
Commercial use
commercial OK
Credit
Not required
Status
active: License spot-checked at the source; the entry carries dated evidence.
espeak-ng-data (voices, dictionaries) lives in-tree under the same GPL. Do not treat voices as a separate Apache grant
This is why Piper Plus advertises "without espeak-ng"
Checklist: license file visible, commercial use of the program under GPL, no NC/ND, not a marketplace, named org, not blocklisted, engine vs WAV unquoted
The project states "GPL version 3 or later", so the SPDX identifier is GPL-3.0-or-later. The bundled getopt.c is BSD-2-Clause.
Evidence
Live GitHub README (2026-09-22): "eSpeak NG Text-to-Speech is released under the GPL version 3 or later license"
Live GitHub README, License Information (2026-08-24): "released under the GPL version 3 or later license"
Live COPYING (2026-08-24): "GNU GENERAL PUBLIC LICENSE Version 3, 29 June 2007"