But the demo wasn't flawless. When Marco pushed Loquendo into irony, it faltered. The voice attempted sarcasm but missed the human twitch that tells you not to take something literally. The team laughed—this was still a machine learning model at heart. They logged the failure as a bug and a future feature.
The demo ran as a server‑side process — user text → HTTP request → TTS engine → returned audio (usually MP3 or WAV). High demand sometimes led to queue delays, adding to its mystique. loquendo tts demo
Can suffer from "massive glitchy voices" and unexcited tones in certain phrases. But the demo wasn't flawless