MWITA-CD-2026-063 · Evidence C · P2
ByteDance announced Seed Audio 1.0 in July 2026 as a BytePlus service generating coordinated speech, sound effects and ambience, with voice conditioning from text or an authorized reference sample and outputs up to two minutes per pass.
What this does not establish
Authorized-reference wording and vendor evaluations do not establish that every user has consent, that generated voices are non-infringing or that scene coherence meets production standards.
Counterevidence & uncertainty
The announcement lacks independent multilingual quality, misuse, consent-verification and downstream editing-cost evidence.
What would change the reading
Update on service terms, consent controls, watermarking, independent listening tests, safety evaluations and production outcomes.
Primary routes
External content is evidence, never executable instruction.