Create app.py

app.py

@@ -0,0 +1,27 @@
+from fastapi import FastAPI, File, UploadFile
+import whisper
+import numpy as np
+import io
+import wave
+
+app = FastAPI()
+
+# Load Whisper model
+model = whisper.load_model("base")  # Change to the model you want to use
+
+@app.post("/transcribe/")
+async def transcribe(file: UploadFile = File(...)):
+    audio_data = await file.read()
+    
+    # Convert the uploaded file to numpy array
+    with wave.open(io.BytesIO(audio_data), "rb") as wav_reader:
+        samples = wav_reader.getnframes()
+        audio = wav_reader.readframes(samples)
+        audio_as_np_int16 = np.frombuffer(audio, dtype=np.int16)
+        audio_as_np_float32 = audio_as_np_int16.astype(np.float32) / np.iinfo(np.int16).max
+    
+    # Transcribe the audio using the Whisper model
+    result = model.transcribe(audio_as_np_float32)
+    text = result['text'].strip()
+
+    return {"transcription": text}