upd attr

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.mp4 filter=lfs diff=lfs merge=lfs -text

README.md

@@ -8,6 +8,4 @@ sdk_version: 5.22.0
 app_file: app.py
 pinned: false
 license: lgpl-3.0
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+---

app.py

@@ -398,18 +398,26 @@ if __name__ == "__main__":
         shutil.rmtree("./flagged")
 
     with gr.Blocks() as demo:
-        gr.Markdown(
-            "## The current CPU-based version on HuggingFace has slow inference, you can access the GPU-based mirror on [ModelScope](https://www.modelscope.cn/studios/monetjoe/EMusicGen)"
-        )
         with gr.Row():
             with gr.Column():
+                gr.Video(
+                    "./tutorial.mp4",
+                    label="Tutorial",
+                    show_download_button=False,
+                    show_share_button=False,
+                )
                 dataset_option = gr.Dropdown(
                     ["VGMIDI", "EMOPIA", "Rough4Q"],
                     label="Dataset",
                     value="Rough4Q",
                 )
-                gr.Markdown(
-                    "# Generate by emotion condition<br><img width='100%' src='https://www.modelscope.cn/studio/monetjoe/EMusicGen/resolve/master/src/4q.jpg'>"
+                gr.Markdown("# Generate by emotion condition")
+                gr.Image(
+                    "https://www.modelscope.cn/studio/monetjoe/EMelodyGen/resolve/master/src/4q.jpg",
+                    show_label=False,
+                    show_download_button=False,
+                    show_fullscreen_button=False,
+                    show_share_button=False,
                 )
                 valence_radio = gr.Radio(
                     ["Low", "High"],
@@ -427,16 +435,8 @@ if __name__ == "__main__":
                 )
                 gen_btn = gr.Button("Generate")
                 gr.Markdown("# Generate by feature control")
-                std_option = gr.Radio(
-                    ["Low", "High"],
-                    label="Pitch SD",
-                    value="High",
-                )
-                mode_option = gr.Radio(
-                    ["Minor", "Major"],
-                    label="Mode",
-                    value="Major",
-                )
+                std_option = gr.Radio(["Low", "High"], label="Pitch SD", value="High")
+                mode_option = gr.Radio(["Minor", "Major"], label="Mode", value="Major")
                 tempo_option = gr.Slider(
                     minimum=40,
                     maximum=228,
@@ -468,51 +468,41 @@ if __name__ == "__main__":
                     label="The emotion to which the current template belongs",
                 )
                 save_btn = gr.Button("Save template")
-                gr.Markdown(
-                    """
-## Cite
-```bibtex
-@article{Zhou2024EMusicGen,
-title     = {EMusicGen: Emotion-Conditioned Melody Generation in ABC Notation},
-author    = {Monan Zhou, Xiaobing Li, Feng Yu and Wei Li},
-month     = {Sep},
-year      = {2024},
-publisher = {GitHub},
-version   = {0.1},
-url       = {https://github.com/monetjoe/EMusicGen}
-}
-```
-"""
-                )
+                # gr.Markdown(
+                #     """
+                #     ## Cite
+                #     ```bibtex
+                #     @article{Zhou2024EMelodyGen,
+                #     title     = {EMelodyGen: Emotion-Conditioned Melody Generation in ABC Notation},
+                #     author    = {Monan Zhou, Xiaobing Li, Feng Yu and Wei Li},
+                #     month     = {Sep},
+                #     year      = {2024},
+                #     publisher = {GitHub},
+                #     version   = {0.1},
+                #     url       = {https://github.com/monetjoe/EMelodyGen}
+                #     }
+                #     ```
+                #     """
+                # )
 
             with gr.Column():
-                gr.Video(
-                    "https://www.modelscope.cn/studio/monetjoe/EMusicGen/resolve/master/src/tutorial.mp4",
-                    label="Tutorial",
-                    show_download_button=False,
-                    show_share_button=False,
-                )
                 wav_audio = gr.Audio(label="Audio", type="filepath")
                 midi_file = gr.File(label="Download MIDI")
                 pdf_file = gr.File(label="Download PDF score")
                 xml_file = gr.File(label="Download MusicXML")
                 mxl_file = gr.File(label="Download MXL")
-                abc_textbox = gr.Textbox(
-                    label="ABC notation",
-                    show_copy_button=True,
-                )
+                abc_textbox = gr.Textbox(label="ABC notation", show_copy_button=True)
                 staff_img = gr.Image(label="Staff", type="filepath")
 
-        with gr.Row():
-            gr.Interface(
-                fn=feedback,
-                inputs=gr.Radio(
-                    ["Q1", "Q2", "Q3", "Q4"],
-                    label="Feedback: the emotion you believe the generated result should belong to",
-                ),
-                outputs=gr.Textbox(show_copy_button=False, show_label=False),
-                allow_flagging="never",
-            )
+                gr.Interface(
+                    fn=feedback,
+                    inputs=gr.Radio(
+                        ["Q1", "Q2", "Q3", "Q4"],
+                        label="Feedback: the emotion you believe the generated result should belong to",
+                    ),
+                    outputs=gr.Textbox(show_copy_button=False, show_label=False),
+                    allow_flagging="never",
+                )
 
         gen_btn.click(
             fn=inference,

utils.py

@@ -5,10 +5,10 @@ import torch
 import requests
 import subprocess
 from tqdm import tqdm
-from modelscope import snapshot_download
+from huggingface_hub import snapshot_download
 
 TEMP_DIR = "./flagged"
-WEIGHTS_DIR = snapshot_download("monetjoe/EMusicGen", cache_dir="./__pycache__")
+WEIGHTS_DIR = snapshot_download("monetjoe/EMelodyGen", cache_dir="./__pycache__")
 DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 PATCH_LENGTH = 128  # Patch Length
 PATCH_SIZE = 32  # Patch Size
@@ -47,7 +47,7 @@ if sys.platform.startswith("linux"):
     extra_dir = "squashfs-root"
     download(
         filename=apkname,
-        url="https://www.modelscope.cn/studio/Genius-Society/piano_transcription/resolve/master/MuseScore.AppImage",
+        url="https://master.dl.sourceforge.net/project/musescore-linux-mirror/MuseScore.AppImage?viasf=1",
     )
     if not os.path.exists(extra_dir):
         subprocess.run(["chmod", "+x", f"./{apkname}"])