Sume TTS Router: Eleven rows have no timestamps or avatar voice

On Sume's TTS Router, Eleven rows refuse timestamps, segmentation, transcript_source and avatar selectors. Sonic rows accept them. Pick by the fields you need.

5 min readSume
All posts

Which TTS Router model you pick decides which fields you can send. In the model catalog on main, every ElevenLabs row carries the constraint that voice.id is an ElevenLabs voice id from GET /v1/tts-router/voices?family=eleven, and that avatar selectors, transcript_source, timestamps and segmentation are not supported. Cartesia Sonic rows have no such constraint. If you need word timings straight from the TTS job or an avatar's own voice, use a Sonic row.

Source: packages/api-contract/src/tts-router-models.ts on origin/main b28f3b0 (read 2026-10-13). The older docs/api/tts-router.md still says Cartesia only, so read the live catalog.

Field support by TTS Router family (read 2026-10-13)
FieldSonic rowsEleven rows
voice.idCartesia voice (library or voi_* clone)ElevenLabs voice id from the eleven family list
avatar_id / avatar_handleSupportedNot supported
transcript_sourceSupportedNot supported
timestamps, segmentationSupportedNot supported
Max characters per job20,000v3: 5,000; v4, v4 Turbo, Multilingual v2: 10,000; Turbo v2.5: 20,000

What if I picked Eleven and still need timings?

Run speech-to-text on the finished audio. POST /v1/video-inspect with transcribe: true returns word timings and sentence segments for a hosted clip, and POST /v1/video-captions burns captions onto a public video URL. Either way it costs a second job, so choose Sonic if timings are the goal.

How do I see the rules for one model?

Ask the catalog. GET /v1/tts-router/models/{model_id} returns the capabilities, pricing and constraints of one row, and a 404 for an unknown id.

curl https://api.sume.com/v1/tts-router/models/eleven-v4-turbo \
  -H "Authorization: Bearer $SUME_API_KEY"

Sources

Related posts

More in Developers

All Developers posts

Written by Sume