Init

.gitignore

@@ -0,0 +1,3 @@
+/build/
+/dist/
+__pycache__/

MakePackage.ps1

@@ -0,0 +1,129 @@
+function MakeDir {
+    param($Dir)
+    if (-not (Test-Path $Dir)) {
+        mkdir -Path $Dir | Out-Null
+    }
+}
+
+function Registry {
+    param($Path,$Name,$Value,$Type)
+    $Json = (ConvertTo-Json $Value)
+    $command = ""
+    if (-not (Test-Path $Path)) {
+        $command = "New-Item `"$Path`" -Force | New-ItemProperty -Name `"$Name`" -PropertyType $Type -Force -Value "
+    } elseif (-not ((ConvertTo-Json (Get-ItemProperty $Path | Select-Object -ExpandProperty $Name -ErrorAction Ignore)) -eq $Json)) {
+        $command = "Set-ItemProperty `"$Path`" -Name `"$Name`" -Type $Type -Force -Value "
+    }
+    if (-not ($command -eq "")) {
+        Write-Host "Registry: $Path!$Name -> $Value"
+        if ($Type -eq "String") {
+            $command += "`"$Value`""
+        } else {
+            $command += "(ConvertFrom-Json `"$Json`")"
+        }
+        RunAsAdmin $command
+    }
+}
+
+function RunAsAdmin {
+    param($Command)
+    $bytes = [System.Text.Encoding]::Unicode.GetBytes($Command)
+    $encodedCommand = [Convert]::ToBase64String($bytes)
+    Start-Process powershell -Verb runAs -ArgumentList "-EncodedCommand $encodedCommand"
+}
+
+function UnpackUrl {
+    param($Url,$File,$UnpackDir,$TestPath)
+    if (-not $File) {
+        $File = $Url.Substring($Url.LastIndexOf("/") + 1)
+    }
+    $Output = "dist\downloads\$File"
+    if (-not $TestPath) {
+        $TestPath = $UnpackDir
+    }
+    if (-not (Test-Path "$TestPath")) {
+        Write-Host "UnpackUrl: $Url -> $UnpackDir"
+        if (-not (Test-Path $Output)) {
+            Import-Module BitsTransfer
+            Start-BitsTransfer -Description "Downloading $File from $Url" -Source $Url -Destination $Output
+        }
+        switch ((Get-Item $Output).Extension) {
+            '.zip' {
+                $shell = New-Object -com shell.application
+                $shell.Namespace([IO.Path]::Combine($pwd, $UnpackDir)).CopyHere($shell.Namespace([IO.Path]::Combine($pwd, $Output)).Items())
+            }
+            '.exe' {
+                Start-Process $output -Wait -ArgumentList "-y -o$UnpackDir"
+            }
+        }
+    }
+}
+
+# disable bits branchcache https://powershell.org/forums/topic/bits-transfer-with-github/
+Registry -Path HKLM:\SOFTWARE\Policies\Microsoft\Windows\BITS -Name DisableBranchCache -Value 1 -Type DWord
+
+MakeDir build\
+MakeDir dist\downloads\
+
+UnpackUrl -Url https://github.com/winpython/winpython/releases/download/2.3.20200530/Winpython64-3.7.7.1dot.exe `
+    -UnpackDir build\ -TestPath build\python\
+if (-not (Test-Path build\python\)) {
+    mv build\WPy64-3771\ build\python\
+}
+
+$Python="build\python\scripts\python.bat"
+$ScriptsDir="build\python\python-3.7.7.amd64\Scripts"
+$LibsDir="build\python\python-3.7.7.amd64\Lib\site-packages"
+
+if (-not (Test-Path $LibsDir\torch)) {
+    & $Python -m pip install torch==1.4.0 -f https://download.pytorch.org/whl/torch_stable.html
+}
+
+if (-not (Test-Path $LibsDir\piano_transcription_inference)) {
+    & $Python -m pip install piano_transcription_inference
+}
+
+if (-not (Test-Path $ScriptsDir\pyinstaller.exe)) {
+    & $Python -m pip install pyinstaller
+}
+
+& $Python -m pip freeze | Out-File -encoding UTF8 pip.txt
+
+if (-not (Test-Path build\dist\PianoTrans\)) {
+    cp PianoTrans.py, PianoTrans.spec build\
+    & $Python $ScriptsDir\pyinstaller.exe `
+    --noconfirm `
+    --distpath build\dist\ `
+    --workpath build\build\ `
+    --specpath build\ `
+    build\PianoTrans.spec
+}
+
+MakeDir build\dist\PianoTrans\piano_transcription_inference_data\
+UnpackUrl -Url 'https://zenodo.org/record/4034264/files/CRNN_note_F1%3D0.9677_pedal_F1%3D0.9186.pth?download=1' `
+    -File 'note_F1=0.9677_pedal_F1=0.9186.pth' -TestPath 'dist\downloads\note_F1=0.9677_pedal_F1=0.9186.pth'
+if (-not (Test-Path 'build\dist\PianoTrans\piano_transcription_inference_data\note_F1=0.9677_pedal_F1=0.9186.pth')) {
+    cp dist\downloads\note_F1=0.9677_pedal_F1=0.9186.pth 'build\dist\PianoTrans\piano_transcription_inference_data\note_F1=0.9677_pedal_F1=0.9186.pth'
+}
+
+$ffmpeg_version='ffmpeg-n4.3.1-30-g666d2fc6e2-win64-gpl-4.3'
+MakeDir build\dist\PianoTrans\ffmpeg\
+UnpackUrl -Url https://github.com/BtbN/FFmpeg-Builds/releases/download/autobuild-2021-02-01-12-44/$ffmpeg_version.zip `
+    -UnpackDir build\ -TestPath build\dist\PianoTrans\ffmpeg\ffmpeg.exe
+if (Test-Path build\$ffmpeg_version\) {
+    mv build\$ffmpeg_version\bin\ffmpeg.exe build\dist\PianoTrans\ffmpeg\
+    rm -r build\$ffmpeg_version\
+}
+
+MakeDir build\dist\PianoTrans\reg\
+cp README.md build\dist\PianoTrans\README.txt
+cp RightClickMenuRegister.bat, RightClickMenuUnregister.bat build\dist\PianoTrans\
+cp RightClickMenuRegister.reg.in, RightClickMenuUnregister.reg build\dist\PianoTrans\reg\
+
+if (-not (Test-Path dist\PianoTrans.zip)) {
+    Add-Type -assembly "system.io.compression.filesystem"
+    [IO.Compression.ZipFile]::CreateFromDirectory([IO.Path]::Combine($pwd, "build\dist\"), [IO.Path]::Combine($pwd, "dist\PianoTrans.zip"))
+}
+
+Write-Host
+Read-Host "Done, press enter to exit"

PianoTrans.py

@@ -0,0 +1,70 @@
+#!/usr/bin/python
+
+import argparse
+import os
+import sys
+import time
+import torch
+
+
+class Args:
+    def __init__(self, **entries):
+        self.__dict__.update(entries)
+
+def inference(args):
+    """Inference template.
+
+    Args:
+      model_type: str
+      audio_path: str
+      cuda: bool
+    """
+
+    from piano_transcription_inference import PianoTranscription, sample_rate, load_audio
+
+    # Arugments & parameters
+    audio_path = args.audio_path
+    output_midi_path = args.output_midi_path
+    checkpoint_path = args.checkpoint_path
+    device = 'cuda' if args.cuda and torch.cuda.is_available() else 'cpu'
+
+    # Load audio
+    (audio, _) = load_audio(audio_path, sr=sample_rate, mono=True)
+
+    # Transcriptor
+    transcriptor = PianoTranscription(device=device, checkpoint_path=checkpoint_path)
+    """device: 'cuda' | 'cpu'
+    checkpoint_path: None for default path, or str for downloaded checkpoint path.
+    """
+
+    # Transcribe and write out to MIDI file
+    transcribe_time = time.time()
+    transcribed_dict = transcriptor.transcribe(audio, output_midi_path)
+    print('Transcribe time: {:.3f} s'.format(time.time() - transcribe_time))
+
+if __name__ == '__main__':
+    files = tuple(sys.argv)[1:]
+
+    if len(files) == 0:
+        import tkinter as tk
+        from tkinter import filedialog
+        root = tk.Tk()
+        root.withdraw()
+        files = filedialog.askopenfilenames(filetypes = [('audio files', '*')])
+        files = root.tk.splitlist(files)
+
+    script_dir = os.path.dirname(sys.argv[0])
+    os.environ['PATH'] += os.pathsep + os.path.abspath(os.path.join(script_dir, 'ffmpeg'))
+    checkpoint_path = os.path.abspath(os.path.join(script_dir, 'piano_transcription_inference_data', 'note_F1=0.9677_pedal_F1=0.9186.pth'))
+
+    for file in files:
+        args = Args(**{
+            'audio_path': file,
+            'output_midi_path': '{}.mid'.format(file),
+            'cuda': True,
+            'checkpoint_path': checkpoint_path
+        })
+        print('Transcribe {}, please wait...'.format(file))
+        inference(args)
+
+    input("\nPress Enter to exit...")

PianoTrans.spec

@@ -0,0 +1,40 @@
+# -*- mode: python ; coding: utf-8 -*-
+
+block_cipher = None
+
+from PyInstaller.utils.hooks import collect_data_files, copy_metadata
+datas = collect_data_files('librosa')
+hiddenimports = ['sklearn.utils._weight_vector']
+
+a = Analysis(['PianoTrans.py'],
+             pathex=['build\\'],
+             binaries=[],
+             datas=datas,
+             hiddenimports=hiddenimports,
+             hookspath=[],
+             runtime_hooks=[],
+             excludes=[],
+             win_no_prefer_redirects=False,
+             win_private_assemblies=False,
+             cipher=block_cipher,
+             noarchive=False)
+pyz = PYZ(a.pure, a.zipped_data,
+             cipher=block_cipher)
+exe = EXE(pyz,
+          a.scripts,
+          [],
+          exclude_binaries=True,
+          name='PianoTrans',
+          debug=False,
+          bootloader_ignore_signals=False,
+          strip=False,
+          upx=True,
+          console=True )
+coll = COLLECT(exe,
+               a.binaries,
+               a.zipfiles,
+               a.datas,
+               strip=False,
+               upx=True,
+               upx_exclude=[],
+               name='PianoTrans')

README.md

@@ -0,0 +1,49 @@
+## Simple GUI for ByteDance's Piano Transcription with Pedals
+
+[ByteDance's Piano Transcription][1] is the PyTorch implementation of the
+piano transcription system, "High-resolution Piano Transcription with Pedals
+by Regressing Onsets and Offsets Times `[1]`".
+
+Using this, we can transcribe piano recordings into MIDI files with pedals.
+
+This is a simple GUI and packaging for Windows.
+
+![screenshot](screenshot.png)
+
+### Requirement
+
+* OS: Windows 7 or later (64-bit)
+* Memory: at least 8G
+
+Only Windows 10 is tested.
+
+It works on Linux, mac OS, Windows, but only Windows package is provided
+here.
+
+If you use other OS, follow those [install and usage][2] guides instead.
+
+### How to use
+
+1. Download [Microsoft Visual C++ Redistributable for Visual Studio 2015, 2017 and 2019][3] `vc_redist_x64.exe` and install
+2. Download and unpack [PianoTrans.zip][4]
+3. Close other apps to free memory, need at least 4G free memory
+4. Run `PianoTrans.exe` in `PianoTrans` directory
+5. Choose audio files, multiple selection is supported
+6. Result MIDI files are in the same directory as the audio files
+
+If you want right click menu for audio files, run `RightClickMenuRegister.bat`,
+then you can right click an audio file, and choose `Piano Transcribe`.
+
+[1]: https://github.com/bytedance/piano_transcription
+[2]: https://github.com/qiuqiangkong/piano_transcription_inference
+[3]: https://support.microsoft.com/en-us/help/2977003/the-latest-supported-visual-c-downloads
+[4]: https://github.com/azuwis/PianoTrans/releases/download/v0.1/PianoTrans-v0.1.zip
+
+### Changelog
+
+#### [0.1] - 2021-02-02
+
+* Initial release.
+
+## Cite
+`[1]` Qiuqiang Kong, Bochen Li, Xuchen Song, Yuan Wan, and Yuxuan Wang. "High-resolution Piano Transcription with Pedals by Regressing Onsets and Offsets Times." arXiv preprint arXiv:2010.01815 (2020). [[pdf]](https://arxiv.org/pdf/2010.01815.pdf)

RightClickMenuRegister.bat

@@ -0,0 +1,16 @@
+@echo off
+echo Register right click menu for audio files.
+set dir=%~dp0
+for /f "delims=" %%i in ('type "reg\RightClickMenuRegister.reg.in" ^& break ^> "reg\RightClickMenuRegister.reg" ') do (
+    set "line=%%i"
+    setlocal enabledelayedexpansion
+    >>"reg\RightClickMenuRegister.reg" echo(!line:@REPLACE@=%dir:\=\\%!
+    endlocal
+)
+regedit /S "%~dp0reg\RightClickMenuRegister.reg"
+echo Done.
+IF %0 == "%~0" (
+echo Press any key to exit...
+pause >nul
+)
+

RightClickMenuRegister.reg.in

@@ -0,0 +1,8 @@
+Windows Registry Editor Version 5.00
+
+[HKEY_CLASSES_ROOT\*\shell\PianoTrans]
+@="Piano Transcribe"
+"AppliesTo"=".flac OR .m4a OR .mp3 OR .mp4 OR .wav"
+
+[HKEY_CLASSES_ROOT\*\shell\PianoTrans\command]
+@="\"@[email protected]\" \"%1\""

RightClickMenuUnregister.bat

@@ -0,0 +1,9 @@
+@echo off
+echo Unregister right click menu for audio files.
+regedit /S "%~dp0reg\RightClickMenuUnregister.reg"
+echo Done.
+IF %0 == "%~0" (
+echo Press any key to exit...
+pause >nul
+)
+

RightClickMenuUnregister.reg

@@ -0,0 +1,5 @@
+Windows Registry Editor Version 5.00
+
+[-HKEY_CLASSES_ROOT\*\shell\PianoTrans\command]
+
+[-HKEY_CLASSES_ROOT\*\shell\PianoTrans]

pip.txt

@@ -0,0 +1,44 @@
+altgraph==0.17
+appdirs==1.4.4
+audioread==2.1.9
+certifi==2020.12.5
+cffi==1.14.4
+chardet==4.0.0
+cycler==0.10.0
+decorator==4.4.2
+future==0.18.2
+idna==2.10
+importlib-metadata==3.4.0
+joblib==1.0.0
+kiwisolver==1.3.1
+librosa==0.8.0
+llvmlite==0.35.0
+matplotlib==3.3.4
+mido==1.2.9
+numba==0.52.0
+numpy==1.20.0
+packaging==20.9
+pefile==2019.4.18
+piano-transcription-inference==0.0.4
+Pillow==8.1.0
+pooch==1.3.0
+pycparser==2.20
+pyinstaller==4.2
+pyinstaller-hooks-contrib==2020.11
+pyparsing==2.4.7
+python-dateutil==2.8.1
+pywin32-ctypes==0.2.0
+requests==2.25.1
+resampy==0.2.2
+scikit-learn==0.24.1
+scipy==1.6.0
+six==1.15.0
+SoundFile==0.10.3.post1
+sqlite-bro==0.9.1
+threadpoolctl==2.1.0
+torch==1.4.0+cu92
+torchlibrosa==0.0.7
+typing-extensions==3.7.4.3
+urllib3==1.26.3
+winpython==2.4.20200425
+zipp==3.4.0