Update app.py

app.py

@@ -12,7 +12,7 @@ app = FastAPI()
 
 import torch
 
-model_id = "whitefox123/whisper-small-ar2"
+model_id = "WajeehAzeemX/whisper-smal-ar-testing-kale-5000"
 model = WhisperForConditionalGeneration.from_pretrained(
     model_id
 )
@@ -50,38 +50,3 @@ async def transcribe_audio(request: Request):
         raise HTTPException(status_code=500, detail=str(e))
 
 
-model_id_2 = "openai/whisper-medium"
-model_2 = WhisperForConditionalGeneration.from_pretrained(
-    model_id_2
-)
-processor_2 = WhisperProcessor.from_pretrained(model_id_2, attn_implementation="sdpa")
-model_2.config.forced_decoder_ids = None
-forced_decoder_ids_2 = processor_2.get_decoder_prompt_ids(language="Arabic", task="transcribe")
-model_2.generation_config.cache_implementation = "static"
-pipe = pipeline(
-    "automatic-speech-recognition",
-    model=model_2,
-    tokenizer=processor_2.tokenizer,
-    feature_extractor=processor_2.feature_extractor,
-)
-@app.post("/transcribe-large/")
-async def transcribe_audio_2(request: Request):
-    try:
-        # Read binary data from the request
-        audio_data = await request.body()
-        # Convert binary data to a file-like object
-        audio_file = io.BytesIO(audio_data)
-        # Load the audio file using pydub
-        audio_array, sampling_rate = librosa.load(audio_file, sr=16000)
-        # Process the audio array
-        input_features = processor_2(audio_array, sampling_rate=sampling_rate, return_tensors="pt").input_features
-        # Generate token ids
-        predicted_ids = model_2.generate(input_features, forced_decoder_ids=forced_decoder_ids_2)
-        # Decode token ids to text
-        transcription = processor_2.batch_decode(predicted_ids, skip_special_tokens=True)
-        # Print the transcription
-        print(transcription[0])  # Display the transcriptiontry:
-        return {"transcription": transcription[0]}
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-