Initial commit

.gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*.tfevents* filter=lfs diff=lfs merge=lfs -text
+*.db* filter=lfs diff=lfs merge=lfs -text
+*.ark* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*data* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.meta filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.index filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.AppImage filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,5 @@
+example/*
+*__pycache__*
+test.py
+rename.sh
+flagged/*

README.md

@@ -0,0 +1,12 @@
+---
+title: Piano Transcriptor
+emoji: 🎹
+colorFrom: yellow
+colorTo: green
+sdk: gradio
+sdk_version: 5.22.0
+app_file: app.py
+pinned: true
+license: mit
+short_description: Piano transcription tool
+---

app.py

@@ -0,0 +1,102 @@
+import os
+import torch
+import shutil
+import gradio as gr
+from piano_transcription_inference import PianoTranscription, load_audio, sample_rate
+from convert import midi2xml, xml2abc, xml2mxl, xml2jpg
+
+EN_US = os.getenv("LANG") != "zh_CN.UTF-8"
+
+if EN_US:
+    import huggingface_hub
+
+    MODEL_PATH = huggingface_hub.snapshot_download(
+        "Genius-Society/piano_trans",
+        cache_dir="./__pycache__",
+    )
+
+else:
+    import modelscope
+
+    MODEL_PATH = modelscope.snapshot_download(
+        "Genius-Society/piano_trans",
+        cache_dir="./__pycache__",
+    )
+
+
+ZH2EN = {
+    "五线谱": "Staff",
+    "状态栏": "Status",
+    "下载 MXL": "Download MXL",
+    "ABC 记谱": "ABC notation",
+    "上传音频": "Upload an audio",
+    "下载 MIDI": "Download MIDI",
+    "下载 PDF 乐谱": "Download PDF score",
+    "下载 MusicXML": "Download MusicXML",
+    "钢琴转谱工具": "Piano Transcription Tool",
+    "请上传音频 100% 后再点提交": "Please make sure the audio is completely uploaded before clicking Submit",
+}
+
+
+def _L(zh_txt: str):
+    return ZH2EN[zh_txt] if EN_US else zh_txt
+
+
+def clean_cache(cache_dir):
+    if os.path.exists(cache_dir):
+        shutil.rmtree(cache_dir)
+
+    os.mkdir(cache_dir)
+
+
+def audio2midi(audio_path: str, cache_dir: str):
+    audio, _ = load_audio(audio_path, sr=sample_rate, mono=True)
+    transcriptor = PianoTranscription(
+        device="cuda" if torch.cuda.is_available() else "cpu",
+        checkpoint_path=f"{MODEL_PATH}/CRNN_note_F1=0.9677_pedal_F1=0.9186.pth",
+    )
+    midi_path = f"{cache_dir}/output.mid"
+    transcriptor.transcribe(audio, midi_path)
+    return midi_path, os.path.splitext(os.path.basename(audio_path))[0].capitalize()
+
+
+def upl_infer(audio_path: str, cache_dir="./__pycache__/cache"):
+    status = "Success"
+    midi = pdf = xml = mxl = abc = jpg = None
+    try:
+        clean_cache(cache_dir)
+        midi, title = audio2midi(audio_path, cache_dir)
+        xml = midi2xml(midi, title)
+        abc = xml2abc(xml)
+        mxl = xml2mxl(xml)
+        pdf, jpg = xml2jpg(xml)
+
+    except Exception as e:
+        status = f"{e}"
+
+    return status, midi, pdf, xml, mxl, abc, jpg
+
+
+if __name__ == "__main__":
+    gr.Interface(
+        fn=upl_infer,
+        inputs=gr.Audio(label=_L("上传音频"), type="filepath"),
+        outputs=[
+            gr.Textbox(label=_L("状态栏"), show_copy_button=True),
+            gr.File(label=_L("下载 MIDI")),
+            gr.File(label=_L("下载 PDF 乐谱")),
+            gr.File(label=_L("下载 MusicXML")),
+            gr.File(label=_L("下载 MXL")),
+            gr.Textbox(label=_L("ABC 记谱"), show_copy_button=True),
+            gr.Image(label=_L("五线谱"), type="filepath", show_share_button=False),
+        ],
+        title=_L("钢琴转谱工具"),
+        description=_L("请上传音频 100% 后再点提交"),
+        flagging_mode="never",
+        cache_examples=False,
+        examples=[
+            f"{MODEL_PATH}/examples/1945798894.mp3",
+            f"{MODEL_PATH}/examples/1945798973.mp3",
+            f"{MODEL_PATH}/examples/1946098771.mp3",
+        ],
+    ).launch()

convert.py

@@ -0,0 +1,124 @@
+import os
+import sys
+import fitz
+import requests
+import subprocess
+from PIL import Image
+from music21 import converter
+
+
+def download(url: str, directory: str, filename: str):
+    if directory != "" and not os.path.exists(directory):
+        os.makedirs(directory)
+    # Create the full path for the file to be saved
+    file_path = os.path.join(directory, filename)
+    # Send a GET request to the URL
+    response = requests.get(url, stream=True)
+    # Check if the request was successful
+    if response.status_code == 200:
+        # Open the file in write-binary mode
+        with open(file_path, "wb") as file:
+            # Write the contents of the response to the file
+            for chunk in response.iter_content(chunk_size=1024):
+                if chunk:  # Filter out keep-alive new chunks
+                    file.write(chunk)
+
+        print(f"The file has been downloaded and saved to {file_path}")
+
+    else:
+        print(f"Failed to download the file. Status code: {response.status_code}")
+
+    return os.path.join(directory, filename)
+
+
+if sys.platform.startswith("linux"):
+    apkname = "MuseScore.AppImage"
+    extra_dir = "squashfs-root"
+    if not os.path.exists(apkname):
+        download(url=os.getenv("mscore"), directory="./", filename=apkname)
+
+    if not os.path.exists(extra_dir):
+        subprocess.run(["chmod", "+x", f"./{apkname}"])
+        subprocess.run([f"./{apkname}", "--appimage-extract"])
+
+    MSCORE = f"./{extra_dir}/AppRun"
+    os.environ["QT_QPA_PLATFORM"] = "offscreen"
+
+else:
+    MSCORE = os.getenv("mscore")
+    if not MSCORE:
+        raise EnvironmentError("Please add musescore environment var!")
+
+
+def add_title_to_xml(xml_path: str, title: str):
+    midi_data = converter.parse(xml_path)
+    midi_data.metadata.movementName = title
+    midi_data.metadata.composer = "Transcripted by AI"
+    midi_data.write("musicxml", fp=xml_path)
+
+
+def xml2abc(xml_path: str):
+    result = subprocess.run(
+        ["python", "xml2abc.py", xml_path], stdout=subprocess.PIPE, text=True
+    )
+    if result.returncode == 0:
+        return result.stdout
+
+    return ""
+
+
+def xml2mxl(xml_path: str):
+    mxl_file = xml_path.replace(".musicxml", ".mxl")
+    command = [MSCORE, "-o", mxl_file, xml_path]
+    result = subprocess.run(command)
+    print(result)
+    return mxl_file
+
+
+def midi2xml(mid_file: str, title: str):
+    xml_file = mid_file.replace(".mid", ".musicxml")
+    command = [MSCORE, "-o", xml_file, mid_file]
+    result = subprocess.run(command)
+    add_title_to_xml(xml_file, title)
+    print(result)
+    return xml_file
+
+
+def xml2midi(xml_file: str):
+    midi_file = xml_file.replace(".musicxml", ".mid")
+    command = [MSCORE, "-o", midi_file, xml_file]
+    result = subprocess.run(command)
+    print(result)
+    return midi_file
+
+
+def pdf2img(pdf_path: str):
+    output_path = pdf_path.replace(".pdf", ".jpg")
+    doc = fitz.open(pdf_path)
+    images = []
+    for page_number in range(doc.page_count):
+        page = doc[page_number]
+        image = page.get_pixmap()
+        images.append(
+            Image.frombytes("RGB", [image.width, image.height], image.samples)
+        )
+
+    merged_image = Image.new(
+        "RGB", (images[0].width, sum(image.height for image in images))
+    )
+    y_offset = 0
+    for image in images:
+        merged_image.paste(image, (0, y_offset))
+        y_offset += image.height
+
+    merged_image.save(output_path, "JPEG")
+    doc.close()
+    return output_path
+
+
+def xml2jpg(xml_file: str):
+    pdf_score = xml_file.replace(".musicxml", ".pdf")
+    command = [MSCORE, "-o", pdf_score, xml_file]
+    result = subprocess.run(command)
+    print(result)
+    return pdf_score, pdf2img(pdf_score)

requirements.txt

@@ -0,0 +1,6 @@
+-f https://download.pytorch.org/whl/torch
+librosa==0.9.2
+music21
+piano_transcription_inference
+pymupdf
+torch==2.6.0+cu118

test.sh

@@ -0,0 +1,10 @@
+#!/bin/sh
+
+MOD_BRANCH="main"
+
+git checkout --orphan latest_branch
+git add -A
+git commit -m "Initial commit"
+git branch -D "$MOD_BRANCH"
+git branch -m latest_branch "$MOD_BRANCH"
+git push -u origin "$MOD_BRANCH" -f