Sync dubbing in 92 languages vs building a dub on Sume

Sync's August 31, 2026 changelog says its built-in dubbing now supports 92 languages, detects speakers automatically and returns lossless audio for the final dub. Sume has no single dubbing endpoint: you assemble transcription with an optional language hint, your own translation step, and text-to-speech.

Sync facts are from its changelog; Sume facts from the OpenAPI document and the Models and Avatar videos docs, read 2026-10-01.

What changed in Sync's dubbing?

The changelog entry says dubs can be created across the App, API, Adobe Premiere Pro and DaVinci Resolve. An earlier entry describes translating a video and lip-syncing it in one step, with translation, voice cloning and lipsync handled by Sync, listed there at 29 languages. The 92 figure is the newer one.

What do I assemble on Sume?

What does Sume not do here?

The Models docs say video models do not lip-sync to generated TTS or to a later voice-over. So a dub laid over an existing clip will not have re-synced lips; if you need a face that speaks the new language, generate the speaking shot from a still plus the new audio. Speaker detection is not described in the Sume pages I read, so split speakers yourself before transcribing. The speech transcription rate card lists $0.01 per audio minute.

What about Korean captions?

Avatar video captions have a rule: a Korean script with a Latin-only style such as slam, punch or tiktok-green is rejected with 400 caption_hangul_text_latin_style, so pick a Hangul style for Korean speech. For the full chain see build an AI dubbing pipeline.