HeyGen launches Professional Voice Clone: 20-minute training, API preview

HeyGen · x · 2026-09-10

HeyGen announced Professional Voice Clone, now in preview on the API. The model trains on 20+ minutes of recordings to create a dedicated HeyGen Voice adapter per speaker, offering the highest-fidelity clone. It is in private preview, requires access approval, and is a paid feature: each voice requires a purchased professional voice clone slot. Users can create a voice via API, poll until active, and generate speech (completed or streamed).

Original post →

More from Multimodal

Multimodal channel →