andiekobbietks/abena-ai-api

Abena AI API — African-language Text-to-Speech and Speech Recognition over HTTP. Free tier, no sign-up. # Abena AI API African‑language **Text‑to‑Speech** (speech from text) and **Speech Recognition** (text from speech), over plain HTTP. Use it from a website, a phone app, a server, or just your terminal — **no special library required**. > 🟢 **Start free:** no account or API key is required. Speech Recognition includes 25 free transcriptions; other free API endpoints include up to 30 requests. Sign in for 50 free Speech Recognition transcriptions. - 🌍 Live playground: - 🎙️ Speech Recognition playground: - 📖 Full docs: --- ## Base URL ``` abena.mobobi.com ``` > ℹ️ **First request to a voice may take ~10–15 seconds.** The first time a voice is used, its model is loaded and prepared on our server (offline). After that the same voice responds in about a second. --- ## Text‑to‑Speech Send text + a voice, get audio back. The audio comes as a **Base64‑encoded WAV** inside JSON, which is easy to use from any language and from the browser. **Endpoint** ``` POST /tts/synthesize/ ``` **Request body (JSON)** | Field | Required | Description | |---------|----------|-------------| | `text` | yes | Text to speak (up to 500 characters). | | `voice` | yes | A voice ID, e.g. `akua_eng`, `abena_twi_high`, `kobby_gpe` — see Voices. | | `speed` | no | `1.0` normal, `0.5` slowest, `2.0` fastest. Default `1.0`. | **Response (JSON)** ```json { "status": "success", "voice": "abena_twi_high", "audio_base64": "UklGRiQ...", "duration_seconds": 2.91, "mime_type": "audio/wav", "quality": "high" } ``` `abena_twi_high` always uses the High voice. If the voice service is temporarily at capacity, retry after the interval returned by the API. **Example (curl)** — saves `speech.wav`, no API key needed: ```bash curl -s -X POST abena.mobobi.com \ -H "Content-Type: application/json" \ -d '{"text": "Akwaaba, wo ho te sen?", "voice": "abena_twi_high", "speed": 1.0}' \ | python3 -c "import sys,json,base …