HeyGen's Professional Voice Clone uses 2 hours of audio to preserve breaths, pauses and laughs
rohanpaul_ai · x · 2026-10-21
HeyGen launched Professional Voice Clone: upload 30 minutes to 2 hours of audio and get a clone preserving breaths, pauses, laughs, accent, and vowel quality. Rohan Paul argues the data range is the point — two hours of speech contains hundreds of hesitations, recoveries, and pace shifts that a short clip can't capture, and these small irregularities are exactly what listeners use to decide if a voice is real. First 100 commenters get a free month.
Related event: HeyGen Launches Professional Voice Clone from 30 Minutes of Audio(2 posts)→
More from Multimodal
- Blender + Wan 3 recreates one New York street corner across 400 years in striking split-screen — angierova · 2026-10-21
- Prof. reveals full script of 'Gaussians Unite' trailer made with Codex + Blender — CSProfKGD · 2026-10-21
- Fully AI-generated café UGC video fools viewers, creator shares every prompt — socialwithaayan · 2026-10-21
- Fibo 2.0 LoRA Trained in 42 Minutes for $2 Puts Plush Dragon Into Famous Paintings — SirToki · 2026-10-21
- Ask an AI to draft architectural drawings of Google's Gradient Canopy — the results look straight out of a blueprint archive — fofrAI · 2026-10-21
- MiniMax H3 reference-to-video character swap tested, full ComfyUI workflow released — NextDiffusion · 2026-10-21