Initial commit

transcription-api/main.py

@@ -0,0 +1,43 @@
+from dotenv import load_dotenv
+from fastapi import FastAPI, UploadFile, File
+from models import ModelSize, Languages, DeviceType
+from transcribe import transcribe_file, transcribe_from_filename
+import uvicorn
+import os
+from enum import Enum
+from typing import Annotated
+from backends.fasterwhisper import FasterWhisperBackend
+
+app = FastAPI()
+    
+@app.post("/transcribe/")
+async def transcribe_endpoint(file: UploadFile = File(None),
+                              filename: str = None,
+                              model_size: ModelSize = ModelSize.small, 
+                              language: Languages = Languages.auto,
+                              device: DeviceType = DeviceType.cpu):
+    
+    print(f"Transcribing with model {model_size.value} on device {device.value}...")
+    if file is not None:
+        # if a file is uploaded, use it
+        return await transcribe_file(file, model_size.value, language.value, device.value)
+    elif filename is not None:
+        # if a filename is provided, use it
+        return await transcribe_from_filename(filename, model_size.value, language.value, device.value)
+    else:
+        return {"detail": "No file uploaded and no filename provided"}
+
+@app.get("/healthcheck/")
+async def healthcheck():
+    return {"status": "healthy"}
+
+if __name__ == "__main__":
+    load_dotenv()
+
+    # Get model list (comma separated) from environment variable
+    model_list = os.environ.get("WHISPER_MODELS", "tiny,base,small")
+    model_list = model_list.split(",")
+    for model in model_list:
+        m = FasterWhisperBackend(model_size=model)
+        m.get_model()
+    uvicorn.run(app, host="0.0.0.0", port=8000)