Where to find a voice ID for Sume text to speech: voices_list and voi_
Sume TTS wants a voice UUID or a voi_ library id, never a voice name like alloy. Where to copy it from, and the error you get for a name.
To find a voice ID for Sume text to speech, open Assets then Voices and copy the id, or call the voices_list and voices_get tools through the hosted MCP server. A valid voice.id is either a UUID (8-4-4-4-12 hex) or a Voices library id that starts with voi_ followed by 32 hex characters. A name such as alloy or ko-KR-SunHiNeural is rejected with invalid_voice_id.
What a valid selector looks like
The tool schema accepts exactly one way to name the voice. Either send voice: { "id": "..." } or send a ready avatar's avatar_id or avatar_handle, in which case Sume uses that avatar's ready TTS voice. When you send both, they must resolve to the same voice. Details are in MCP tools and gates.
| Selector | Form | Notes |
|---|---|---|
| voice.id (UUID) | 8-4-4-4-12 hex | Copy it verbatim |
| voice.id (library) | voi_ plus 32 hex | From Assets, Voices |
| avatar_id | Ready avatar id | Uses that avatar's TTS voice |
| avatar_handle | 2 to 31 characters | Must match avatar_id if both are sent |
| Voice name | alloy, korean_female | Rejected with invalid_voice_id |
Common ways it fails
Re-typing an id by eye is the usual cause. The schema's own advice is to copy the id verbatim and re-read it character by character. A second cause is a name copied from another vendor's voice list; those never synthesized on Sume.
A voice also carries a language tag. If you send a request language that differs from it, Sume warns you before creating a job, as the post on the language mismatch warning explains.
Sources
Related posts
More in Developers
- Find dead air in a Reel: sentence segments and silence split
Sume video inspect splits a transcript into sentence segments at pauses of 0.2 to 3 s. Use the gaps between segments to find dead air before you trim.
- Find every Sora call left in your codebase: a retired-id scan
OpenAI's Videos API shut down 2026-09-24. A 20-line Python scan lists every file still naming a retired sora-2 model id so you can swap them.
- Find micro-drama premises: TikTok trending search for $0.10 a call
One trending-videos search returns ranked TikTok metadata for a keyword, from 1 to 50 results, for $0.10. How to read it for premises, not to copy clips.
- Five Sume video submit errors and which ones to retry
insufficient_credits, idempotency_conflict, queue_full, rate_limited and provider_capacity_exceeded each need a different response. What to do, in a table.
Written by Sume