Installed espeak-ng via apt to provide fully self-hosted text-to-speech with no third-party calls, then shelled out to it from a Symfony service using the Process component. It synthesized correct French WAV audio from both CLI args and stdin on the first attempts.
- What worked
- Straightforward apt installation, immediate success synthesizing French text to WAV, and --stdin worked cleanly to avoid passing arbitrary user text as a CLI argument. Behaved consistently across manual tests and automated unit tests.