LIST
AI voice experiences worth trying
Synthetic speech spent a decade as the interface for something else. Now the voice is the product. Here is what that sounds like today, in things anyone can open tonight.
Talking to a character
Character.AI and Replika build personality-led conversations with custom voices. Good at sustaining a character across a long exchange. The seams show when you push at continuity, since the character remembers less about you than the conversation implies.
Talking to an assistant
ChatGPT’s advanced voice mode is the reference point for real-time back and forth. It listens while you speak, handles interruption, and answers fast enough that you stop waiting for it.
Making a voice
ElevenLabs is where most people meet cloning for the first time, with a large library of voices for narration and characters. Ten seconds of clean audio gets you something recognizable. A full minute of emotional range is where short samples still drift.
Waking up inside one
Alarmesque uses the voice as the whole experience. A new scene is written every morning and spoken as your alarm, with you in it, and on the paid tiers you can hold a button and talk back mid-scene while it stays in character.
The interesting part is the context: half awake, no screen, audio only, which is the one situation where a synthetic voice gets taken at face value.
Where the seams still are
- Long emotional reads drift. Short lines hold up better than paragraphs.
- Cloned voices carry timbre and accent well, and miss breath and energy.
- Anything live depends on a network, which is why the good ones pre-render what has to be reliable.
- Consent is not a detail. Cloning a voice belongs to the person who owns it.