AgentPlays
HeyGen Voice launches: voice cloning that matches the room and mic in your video
HeyGen released HeyGen Voice, a voice-cloning model built for talking videos. It learns your voice from about a 60-second phone recording, then uses the video's image and scene captions to set delivery and acoustics, such as room echo and mic distance, instead of copying the recording's sound. HeyGen says it ranked first on Artificial Analysis' controlled-voice leaderboard (Elo 1201 in the Oct 9 snapshot), and that in its own blind test people cloning their own voice picked it over ElevenLabs v4 in 62% of votes and over Cartesia Sonic 3.6 in 54%. HeyGen says it's included in every HeyGen video at no extra cost.
How you’d use it: Make a cloned narration sound like it was recorded where the video was shot.