Roll pyharp and tidy the card description

app.py

@@ -252,15 +252,13 @@ def process_fn(input_audio_path, checkpoint_choice, gap_length_s, gap_position_s
 model_card = ModelCard(
     name="Audio Inpainting Diffusion",
     description=(
-        (
-            "Fills a gap in an audio clip using a diffusion model trained on either "
-            "solo piano (MAESTRO) or mixed classical instrumentation (MusicNet). The "
-            "input is resampled to the chosen model's rate and center-cropped or "
-            "zero-padded to its fixed segment length (about 8.4s for MAESTRO, 4.2s "
-            "for MusicNet) before the gap is filled. Repository: "
-            "https://github.com/eloimoliner/audio-inpainting-diffusion Paper: "
-            "https://arxiv.org/abs/2305.15266"
-        )
+        "Fills a gap in an audio clip using a diffusion model trained on either "
+        "solo piano (MAESTRO) or mixed classical instrumentation (MusicNet). "
+        "The input is resampled to the chosen model's rate and center-cropped "
+        "or zero-padded to its fixed segment length (about 8.4s for MAESTRO, "
+        "4.2s for MusicNet) before the gap is filled. Repository: "
+        "https://github.com/eloimoliner/audio-inpainting-diffusion Paper: "
+        "https://arxiv.org/abs/2305.15266"
     ),
     author="Eloi Moliner, Vesa Välimäki",
     tags=[Subcategory.INFILLING],