HeyGen speech

Use the registered tools, not the retired /v1/audio speech endpoints.

• heygen_tts: native HeyGen Voice1 / Turbo / Turbo Free. Native ACTIVE public, instant or professional voices from /v3/models/audio/voices. Professional voices require Voice1 (not Turbo). Output is44.1kHz WAV; text limit5,000 chars.
• heygen_catalog_tts: catalog voices from /v3/voices; explicit engine selection on /v3/voices/speech (Starfish, HeyGen Voice, or ElevenLabs). These voice IDs are not interchangeable with native clone IDs.
• heygen_voice_convert: timing-preserving conversion with an ACTIVE instant workspace voice via /v3/models/audio/sts.

Read the October23 provider guide for current request fields, output paths, timestamp/SSE handling, pricing, and retry behavior. Do not silently fall back between native models, catalog engines, or hosts. Turbo Free is promotional through November30,2026 and has shared capacity limits.