Update README.md

README.md

@@ -10,40 +10,15 @@ pinned: false
 license: apache-2.0
 ---
 
-# Samaali — Whisper ASR Post-Processing (Arabic)
+## Samaali — Whisper ASR Post-Processing (Arabic)
 
-- Transcribes audio with **faster-whisper** (word timestamps + probabilities)
-- Aligns with the original text and distinguishes **ASR errors** vs **memorization errors**
-- Restores ASR errors to the ground-truth and computes:
-  - **Literal score** (Levenshtein + word-overlap + BLEU-1)
-  - **Semantic score** (SBERT + MARBERT-CLS)
+- Word-level timestamps & probabilities (faster-whisper)
+- Alignment to GT + ASR-vs-Memorization classification
+- Confidence gating + Numbers handling
+- Literal & Semantic scores
 
-## Usage
-1. Upload/record audio and paste the **Original Text**.
-2. Pick Whisper size (`large-v3` on GPU, `small/medium` on CPU).
-3. Click **Transcribe & Evaluate**.
+### API
+- `/run/evaluate` (UI outputs)
+- `/run/predict` (JSON-only)
 
-Outputs:
-- **Corrected Transcript** (ASR-only corrections applied)
-- **Raw ASR Transcript**
-- **JSON Report** (scores & thresholds)
-- **Token-level decisions table**
-
-## API (Spaces Inference)
-Two endpoints are exposed:
-
-### 1) `/run/evaluate` (UI-equivalent)
-**Python**
-```python
-from gradio_client import Client, file
-client = Client("<username>/<space_name>")
-corrected, asr_out, report, table = client.predict(
-    audio=file("audio.wav"),
-    original_text="النص الأصلي...",
-    whisper_size="small",
-    compute_type="int8",
-    vad=True,
-    use_marbert=False,   # True if GPU
-    api_name="/evaluate"
-)
-print(report)  # JSON
+**Note (CPU Spaces):** the app enforces `whisper=small`, `compute=int8`, and disables MARBERT by default to avoid OOM.