Update app.py

app.py

@@ -15,6 +15,10 @@ import whisper
 
 YT_AUDIO_FORMAT = "bestaudio[ext=m4a]"
 
+MODEL_SIZES = ["tiny", "base", "small", "medium", "large", "turbo"]
+for size in MODEL_SIZES:
+    whisper.load_model(size, device="cpu")
+
 
 def download_youtube(url: str, tmp_dir: Path) -> Path:
     """Download the audio track from a YouTube video and return the local path."""
@@ -91,7 +95,7 @@ def transcribe_audio(
                 "segments": segments,
             }
         )
-    df_results = make_results_table(results, return_timestamps)
+    df_results = make_results_table(results)
     return df_results
 
 
@@ -107,7 +111,7 @@ def build_demo() -> gr.Blocks:
         with gr.Row():
             model_choices = gr.Dropdown(
                 label="Model size(s)",
-                choices=["tiny", "base", "small", "medium", "large", "turbo"],
+                choices=MODEL_SIZES,
                 value=["turbo"],
                 multiselect=True,
                 allow_custom_value=False,