Speech synthesis
Speech synthesis is the broad field of generating speech rather than recording it. Text-to-speech is the most common form. It spans everything from early formant synthesizers to neural models that reproduce a specific person's voice.
Why it matters
Knowing that speech synthesis is a field, not a single feature, explains why quality varies so much. Early systems stitched or shaped sound with rules and sounded robotic by design. Modern neural systems learn from real speech and can carry the melody of a voice, which is why the best of them cross from novelty into something publishable.
In practice
When you compare tools, ask which era of synthesis they are built on. A concatenative or rule-based voice will always sound more mechanical than a neural one, no matter how clean the audio is. The approach sets the ceiling.