MAI-Voice-2.1 emotion control vs Sonic's six emotions on Sume
Microsoft lists granular emotion control for MAI-Voice-2.1. On Sume, Sonic 3 rows take six named emotions in generation_config.emotion; Eleven rows have none.
Sume's Sonic 3 rows take one of six emotions: neutral, content, excited, sad, angry or scared, set in generation_config.emotion, alongside speed (0.6 to 1.5) and volume (0.5 to 2). Microsoft says both MAI-Voice-2.1 variants have granular emotion control, and its page does not publish a fixed emotion list. Sume does not list MAI-Voice-2.1, so on Sume the six-emotion list is what you get.
Microsoft facts from its model page, read 2026-10-13; Sume settings from packages/api-contract/src/tts-router-models.ts on origin/main b28f3b0.
| Control | MAI-Voice-2.1 (Microsoft page) | Sume Sonic 3 rows | Sume Eleven rows |
|---|---|---|---|
| Emotion | Granular emotion control | Six named emotions | No emotion field in the catalog |
| Speed | Not stated on the page | 0.6 to 1.5, step 0.05 | Multilingual v2 and Turbo v2.5: 0.7 to 1.2 |
| Volume | Not stated on the page | 0.5 to 2, step 0.05 | Not in the catalog |
| Voice from a short clip | Instant voice matching, no fine-tuning | Voice from the Sume Voices library (voice_id) | Eleven voice ids |
How do I use an emotion on Sume?
Pick a Sonic 3 row, send the emotion in generation_config, and keep the same voice and script across takes so you hear only the emotion change. The catalog field names are the source of truth; ask GET /v1/tts-router/models/sonic-3.6 for the exact settings object before you build a form around it.
What if the line needs a feeling the six do not cover?
Write it into the text and test a few takes. Delivery cues in the script are model-dependent, so run the same sentence on two rows and compare. Keep the take whose captions still match the words: script_text on video-captions burns your approved spelling, whatever the voice did.
Sources
Related posts
More in Models
- Sume video model ids in one table (October 2026)
All Sume video ids with resolution and length limits: Seedance, Kling 3, Wan 3.0, H3, H3 Max, Omni, Grok, plus Google Cloud and Tencent rows. Read 2026-10-13.
- Can Meta Muse Wonder generate images? It returns text; use /v1/images
Muse Wonder is listed on OpenRouter as a chat-completions model with text responses. It is not in Sume's image catalog; pick an image row instead.
- Decision-1 GlobalStandard or DataZoneStandard for a Sume gate?
Microsoft lists both deployment types for Decision-1. DataZoneStandard keeps processing in a data zone; GlobalStandard may vary more in latency. Benchmark.
- Mistral Large 4 function calling: validate arguments against Sume
Mistral Large 4 (Public Preview) lists function calling and structured outputs. Validate its arguments with Sume's tools_schema before any paid call.
Written by Sume