first update

app.py

@@ -1,7 +1,23 @@
 import gradio as gr
+import whisperx
+import whisper
 
-def greet(name):
-    return "Hello " + name + "!!"
+def transcribe(audio_file):
+    device = "cuda" if torch.cuda.is_available() else "cpu"
 
-demo = gr.Interface(fn=greet, inputs="text", outputs="text")
-demo.launch()
+    # Transcribe with original Whisper
+    model = whisper.load_model("large", device)
+    result = model.transcribe(audio_file)
+
+    # Load alignment model and metadata
+    model_a, metadata = whisperx.load_align_model(language_code=result["language"], device=device)
+
+    # Align Whisper output
+    result_aligned = whisperx.align(result["segments"], model_a, metadata, audio_file, device)
+
+    return {"aligned": result_aligned["segments"], "word_segments": result_aligned["word_segments"]}
+
+inputs = gr.inputs.Audio(source="upload", type="filepath")
+outputs = gr.outputs.JSON()
+
+gr.Interface(fn=transcribe, inputs=inputs, outputs=outputs).launch()

requirements.txt

@@ -0,0 +1,2 @@
+gradio
+git+https://github.com/m-bain/whisperx.git