VoiceAI API
Voice AI brings speech and language understanding to African languages. Convert speech to text, generate natural speech from text, and extract named entities, all over a simple REST API.
Speech-to-Text
Transcribe audio in Kinyarwanda, Kiswahili, and Luganda.
Text-to-Speech
Generate natural-sounding Kinyarwanda speech from text.
Named Entities
Detect people, places, organizations, and more in text.
API reference
Every Voice AI endpoint in one place.
Base URL
All Voice AI endpoints are served under the /ai path on the Pindo API host.
Public vs. authenticated access
Every model is reachable two ways. Append /public to call it without a token,
ideal for prototyping, subject to per-IP rate limits. For production, drop the
/public suffix and send a bearer token.
| Mode | Path | Auth |
|---|---|---|
| Public | /ai/stt/rw/public | None · rate-limited per IP |
| Authenticated | /ai/stt/rw | Bearer token |
Authenticated requests require an Authorization header. Get a token from the
Messaging & CLI → Authentication guide,
or read the source on pindo-ai-docs.
Supported languages
| Model | Languages | Codes |
|---|---|---|
| Speech-to-Text | Kinyarwanda, Kiswahili, Luganda | rw · sw · lg |
| Text-to-Speech | Kinyarwanda | rw |
| Named Entity Recognition | Kinyarwanda, English | rw · en |