README.md

---
title: Tahkik Basic Warsh
emoji: 📖
colorFrom: green
colorTo: blue
sdk: docker
app_port: 7860
---

# Tahkik Inference API

FastAPI inference server for the `benhadjermed/tahkik-basic-warsh` Whisper model.
Accepts Arabic Quranic audio and returns a transcription with a confidence score.

## Endpoints

| Method | Path        | Description                  |
|--------|-------------|------------------------------|
| GET    | `/health`   | Liveness check               |
| POST   | `/evaluate` | Transcribe an audio file     |

## POST /evaluate

**Request** — `multipart/form-data`

| Field   | Type | Required | Notes                                      |
|---------|------|----------|--------------------------------------------|
| `audio` | file | yes      | `.wav`, `.mp3`, `.m4a`, `.flac`, or `.ogg` |

**Response** — `application/json`

```json
{
  "transcription": "الحمد لله رب العالمين",
  "confidence_score": 0.9423,
  "processing_time_ms": 1350
}
```

**Error** — non-200 status

```json
{
  "detail": "unsupported audio format: .xyz"
}
```

## Environment / Secrets

| Name       | Where to set      | Purpose                                        |
|------------|-------------------|------------------------------------------------|
| `HF_TOKEN` | Space secret      | Required if `tahkik-basic-warsh` is private    |