updated app.py

app.py

@@ -1,18 +1,36 @@
-import gradio as gr
-from transformers import pipeline
-import spaces
+import torch
+from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
+# from datasets import load_dataset
 
-# Load the Whisper model from Hugging Face
-model = pipeline("automatic-speech-recognition", model="ylacombe/whisper-large-v3-turbo", chunk_length_s=30, device=0)
-# model = pipeline("automatic-speech-recognition", model="distil-whisper/distil-large-v3", chunk_length_s=30, device=0)
+
+device = "cuda:0" if torch.cuda.is_available() else "cpu"
+torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
+
+model_id = "distil-whisper/distil-large-v3"
+
+model = AutoModelForSpeechSeq2Seq.from_pretrained(
+    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
+)
+model.to(device)
+
+processor = AutoProcessor.from_pretrained(model_id)
+
+pipe = pipeline(
+    "automatic-speech-recognition",
+    model=model,
+    tokenizer=processor.tokenizer,
+    feature_extractor=processor.feature_extractor,
+    max_new_tokens=128,
+    torch_dtype=torch_dtype,
+    device=device,
+)
 
 # Function to process audio input and transcribe it
 @spaces.GPU
 def transcribe(audio):
     # Load and preprocess the audio
-    transcription = model(audio,batch_size=1000, generate_kwargs={"task": "transcribe"}, return_timestamps=True)["text"]
-    return transcription
-
+    result = pipe(sample)
+    return result["text"]
 
 # Gradio interface
 interface = gr.Interface(