Partner guide
Use this if your product has known prompts or transcripts and you want to decide quickly whether Prosody fits.
Best fit
Strong fit
- Scripted read-aloud, assessment, and pronunciation practice
- Workflows with known prompts, transcripts, or reference text
- Teams that want lower-friction APIs and structured outputs
Weak fit
- Open-ended conversation scoring without reference text
- Generic speech-to-text replacement
- Broad voice-agent platforms where assessment is not central
Integration path
Decide
Measure output quality, response shape, latency, and fit with your product logic.
Suggested evaluation checklist
| Question | What to look for |
|---|---|
| Does your workflow have reference text? | Prosody is strongest when the expected utterance is known. |
| Do you need word and phoneme detail? | Check whether your UI or QA workflow benefits from structured timing and per-word signals. |
| Do you need batch, live, or both? | Batch is the public default today. Streaming is selective beta after batch fit is proven. |
| Do you need a generic speech vendor? | If you mainly need transcription, Prosody is probably the wrong first layer. |
| Can one good prompt prove the workflow? | Start with a narrow read-aloud or known-transcript task before testing broader product scope. |
Fastest useful evaluation packet
These three things are enough to tell whether Prosody is worth integrating. You can send them over or just try the playground with your own prompts first.
Known prompts
Send 3-5 reference utterances that actually matter to your workflow.
Representative audio
Include a few recordings that show strong, weak, and edge-case pronunciations.
Decision lens
Tell us whether you are testing output quality, UI fit, latency, or pricing first.
What partners usually check first
The shortest path is usually: try the playground → read the docs → check pricing → ask us anything.