88Z AI, 12 October — HeyGen has released its proprietary speech model, HeyGen Voice, which topped the Artificial Analysis TTS leaderboard. Sharing the same end-to-end tech stack as the Avatar V digital human model, it preserves user tone, stress, emotion, and rhythm to improve reading naturalness. The decoupling of similarity and expressiveness_boost parameters balances fidelity and expressiveness.
At a glance
- HeyGen Voice topped the Artificial Analysis TTS leaderboard using the same end-to-end tech as Avatar V.
- Decoupled similarity and expressiveness_boost parameters balance audio fidelity with vocal expressiveness.
The update
HeyGen has officially launched its self-developed speech model, HeyGen Voice, securing the top spot on the Artificial Analysis TTS leaderboard. Utilizing the same end-to-end technical architecture as the Avatar V digital human model, it accurately captures and retains the user's original tone, stress, emotional nuance, and speaking rhythm. This approach effectively mitigates the stiff, robotic feel often associated with traditional synthetic speech. Regarding functionality, HeyGen Voice introduces two independent parameters: similarity and expressiveness_boost. This decoupled design allows users to flexibly adjust vocal expressiveness while maintaining high audio fidelity. It achieves an optimal balance between realism and emotional impact, providing creators with more granular control over their voice outputs.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- VoiceSource published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.