Update app.py

app.py

@@ -1,46 +1,33 @@
-from transformers import pipeline
-import gradio as gr
-from unsloth import FastModel
-from transformers import WhisperForConditionalGeneration
 import torch
+from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
+from datasets import load_dataset
 
 
-model, tokenizer = FastModel.from_pretrained(
-    model_name = "jsbeaudry/creole-speech-to-text",
-    dtype = None, # Leave as None for auto detection
-    load_in_4bit = False, # Set to True to do 4bit quantization which reduces memory
-    auto_model = WhisperForConditionalGeneration,
-    whisper_language = "Haitian",
-    whisper_task = "transcribe",
-    # token = "hf_...", # use one if using gated models like meta-llama/Llama-2-7b-hf
+device = "cuda:0" if torch.cuda.is_available() else "cpu"
+torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
+
+model_id = "openai/whisper-large-v3"
+
+model = AutoModelForSpeechSeq2Seq.from_pretrained(
+    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True
 )
-# Reuse the previously created pipeline object
-# pipe = pipeline(model) # This line caused the error
+model.to(device)
+
+processor = AutoProcessor.from_pretrained(model_id)
 
-# Initialize the pipeline correctly
 pipe = pipeline(
     "automatic-speech-recognition",
     model=model,
-    tokenizer=tokenizer.tokenizer,
-    feature_extractor=tokenizer.feature_extractor,
-    processor=tokenizer,
-    return_language=True,
-    torch_dtype=torch.float16
+    tokenizer=processor.tokenizer,
+    feature_extractor=processor.feature_extractor,
+    chunk_length_s=30,
+    batch_size=16,  # batch size for inference - set based on your device
+    torch_dtype=torch_dtype,
+    device=device,
 )
 
+dataset = load_dataset("distil-whisper/librispeech_long", "clean", split="validation")
+sample = dataset[0]["audio"]
 
-def transcribe(audio):
-    # Use the 'pipe' pipeline
-    text = pipe(audio)["text"]
-    return text
-
-
-iface = gr.Interface(
-    fn=transcribe,
-    inputs=gr.Audio(type="filepath"),
-    outputs="text",
-    title="Whisper medium Creole",
-    description="Realtime demo for Haitian Creole speech recognition using a fine-tuned medium small model.",
-)
-
-iface.launch()
+result = pipe(sample)
+print(result["text"])