It's one of the most predictable Linux‑ecosystem traps: anything that starts out mediocre gets defended with "you just need to configure it", which is how people end up spending three evenings tweaking espeak-ng flags like they’re tuning a race car that was never meant to leave the driveway.

The pattern: The default experience is objectively bad. Users ask why it's bad. Someone replies with a 14‑step guide involving PulseAudio modules, obscure environment variables, and a GitHub repo last updated in 2019 (that they have no experience with but just searched it for you). The user tries it, gets marginal improvement, and is told "well you didn't configure this part." Hours vanish, and the result still sounds like a Speak & Spell.

Linux TTS specifically sucks because it has No unified audio stack. Every TTS engine interacts differently, so "fixes" are never universal. Most engines are research prototypes Festival, espeak, flite; they were built for academia, not consumer polish. The "better" options require GPU inference Coqui, Piper, VITS, etc. -They can sound good, but only after you configure models, sample rates, vocoders, device backends, and sometimes build from source... And no one agrees on defaults: Windows and macOS ship with curated voices.

Linux advocates love saying: "It's bad by default, but you can make it amazing!" -And that "amazing" is hidden behind compiling, editing config files, manually installing voices, tweaking latency, adjusting phoneme tables, and debugging audio routing. -And it's still worse than the default voice on a $40 Android phone!

Linux culture treats time as an infinite resource. If something wastes your time, the community reframes it as a "learning opportunity."