show more datasets

app.py

@@ -6,13 +6,44 @@ from datasets import load_dataset
 
 
 def main():
-    dataset = load_dataset("epr-labs/pijamia-midi-v1", split="train")
+
+    available_datasets = [
+        "pijamia-midi-v1",
+        "lakh-lmd-full",
+        "giant-midi-sustain-v2",
+        "maestro-sustain-v2",
+    ]
+    dataset_name = st.selectbox(
+        label="Select dataset",
+        options=available_datasets,
+    )
+    preview_dataset(dataset_name)
+
+
+def preview_dataset(dataset_name: str):
+    dataset = load_dataset(f"epr-labs/{dataset_name}", split="train[100:200]")
+
+    st.write(f"### Dataset: {dataset_name}")
+
+    st.write(f"""
+    Number of records: {len(dataset)}
+    """)
+    code = f"""
+    dataset = load_dataset("epr-labs/{dataset_name}", split="train")
+
+    record = dataset[321]
+    piece = MidiPiece.from_huggingface(record)
+
+    # Playback in streamlit
+    streamlit_pianoroll.from_fortepyan(piece)
+    """
+    st.code(code, language="python")
 
     record_idx = st.number_input(
         label="record id",
         min_value=0,
         max_value=len(dataset) - 1,
-        value=0,
+        value=50,
     )
 
     record = dataset[record_idx]