Create app.py

app.py

@@ -0,0 +1,37 @@
+from fastapi import FastAPI, UploadFile, File
+from fastapi.responses import JSONResponse
+import torchaudio
+import torch
+from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
+import io
+
+# Load model and processor
+processor = Wav2Vec2Processor.from_pretrained("Mustafaa4a/ASR-Somali")
+model = Wav2Vec2ForCTC.from_pretrained("Mustafaa4a/ASR-Somali")
+model.eval()
+
+# Initialize FastAPI
+app = FastAPI(
+    title="Somali Speech-to-Text API",
+    description="Upload a Somali audio file (.wav) and receive text transcription using ASR model.",
+    version="1.0",
+)
+
+@app.post("/transcribe")
+async def transcribe(audio: UploadFile = File(...)):
+    # Read audio bytes
+    audio_bytes = await audio.read()
+    waveform, sample_rate = torchaudio.load(io.BytesIO(audio_bytes))
+
+    # Ensure 16kHz sample rate
+    if sample_rate != 16000:
+        waveform = torchaudio.transforms.Resample(orig_freq=sample_rate, new_freq=16000)(waveform)
+
+    # Process input
+    inputs = processor(waveform.squeeze(), sampling_rate=16000, return_tensors="pt")
+    with torch.no_grad():
+        logits = model(**inputs).logits
+
+    predicted_ids = torch.argmax(logits, dim=-1)
+    transcription = processor.decode(predicted_ids[0])
+    return JSONResponse(content={"transcription": transcription})