Native macOS utility for finding dual-mono multitrack audio and converting it to mono. The app and CLI are universal2 (Intel + Apple Silicon), with a bundled runtime. Mixers do not install Python or FFmpeg.
下载 Mononizer 1.5.3(Intel + Apple Silicon 通用版) · 全部版本 / Releases
下载 ZIP 后解压,打开其中的 Mononizer.app,无需安装 Python 或 FFmpeg。请下载 Release 中的应用压缩包,而不是 GitHub 自动生成的 Source code。
此版本尚未进行 Apple Developer ID 签名和公证,macOS 首次打开可能提示无法验证开发者。请先用素材副本验证工作流;程序保留原文件,不覆盖已有输出。
主要功能:双单声道识别与转换、+3.0103 dB 补偿、超限自动输出 32-bit float、独立 iXML 清理、实时进度与安全导出。点击左上角“大包小明”logo 可打开漫画彩蛋,点漫画即可关闭。
Version 1.5.1 adds the supplied portrait as the macOS app icon, with an ivory tile and blue-gray rounded border. Version 1.5.2 adds a faint MONONIZER wordmark to the icon's backplate, keeping the approved portrait and border. The in-app 大包小明 sketch branding and audio behavior are unchanged. build_icon.sh generates the ICNS size set from Design/Brand/mononizer-app-icon-v2.png; build_app.sh embeds it before signing and archiving. The source artwork and built-in image editing prompts are recorded in Design/Brand/app-icon-v1-notes.md and Design/Brand/app-icon-v2-notes.md.
Version 1.5.3 adds an offline comic Easter egg: click the 大包小明 logo to pop up the four-panel comic, with the original dog replaced by 大包. It is image-only: no title bar, toolbar or visible controls. Click anywhere on the comic to dismiss (Escape also works). The image fits the screen, never opens automatically and does not pause audio processing. Clicking the logo again dismisses an existing popup. Artwork and the editing prompt are in Design/EasterEgg/.
Download the app ZIP from Releases, extract it and open Mononizer. Local builds also produce dist/Mononizer-macOS-universal-app.zip.
- Choose or drop a folder. Subfolders are included by default.
- Select the tasks: dual-mono conversion and/or iXML cleanup.
- Click 分析 to inspect detection, source peak, and intended output format.
- Click 导出设置…. Choose a save location and a new result folder name, then export either all supported audio or changed files only. The default name is
<source>_整理后; subsequent exports suggest_02,_03, etc. Cancel retains completed files and removes the interrupted temporary conversion.
The interface includes search, result filters, counts, selected-file details, a processing log, and a button to reveal the output. Settings are locked while running. Source/scan/tolerance changes invalidate analysis; changing the destination, copying unchanged files, or choosing iXML removal does not require rereading audio. The GUI exports only the analyzed file inventory and verifies each source fingerprint before writing. Added files are not silently included. Search, filters and row selection do not restrict the export scope.
Version 1.5 separates analysis from export. A preflight rejects existing result folders, conflicting output names, changed sources and insufficient disk space before exporting. Analysis and export display throttled per-block progress; large files continue updating rather than waiting until the entire file is done. Directory scanning and preflight use an indeterminate progress indicator; header-only inspection is file-weighted, audio analysis/export is source-byte-weighted. Final completion is shown only after writing the task report.
Each GUI export includes Mononizer Report.json, listing written/skipped/failed files and excluded file types or links. Partial or cancelled exports also retain Mononizer — 未完成.txt. They are not automatically opened as successful results. Successful exports can open automatically. Retrying creates a fresh result folder, never merges with a prior run. The legacy CLI's default names and behavior remain unchanged.
Version 1.3 adds a compact, file-first layout with the 大包小明讲音频 transparent sketch logo. Choose 外观 → 跟随系统 / 浅色 / 深色 in the sidebar. Both themes use the same template image tinted to match the interface; there is no image background or branding slogan. The appearance preference is remembered. Audio processing is unchanged.
Version 1.4 highlights the next action (choose folder → analyze → process → open output). Clicking 处理… opens output options: export only changed audio, or include unchanged supported audio for a complete folder. The latter is the initial default. You can also choose to open the output folder automatically after processing. Both choices are remembered. With iXML cleanup enabled, “only changed” also includes cleaned stereo/mono WAV files, not just converted mono files.
Source files are kept. Existing output names are errors, never silently overwritten. Use a new output folder for a fresh run. The optional copy setting copies supported audio only, not DAW projects, sidecars, or arbitrary other files.
For interleaved stereo WAV files that Pro Tools imports as separate A1/A2 mono files, enable 清理 iXML(Pro Tools) and disable 双单声道转 Mono. Analyze, then choose export settings. Both tasks can also be enabled together.
This mode removes every top-level iXML chunk from RIFF WAV. It does not convert dual mono, boost gain, change channel layout, or decode/re-encode samples. The fmt, data, bext, and all other chunks (including their padding) are copied byte for byte. Stereo remains interleaved stereo. The entire iXML block is removed, including any scene/take and track-name information it contains. Originals are retained.
Files without iXML are skipped, or copied unchanged if 复制其余音频 is enabled. AIFF is not cleaned by this mode. RF64 is not supported. Metadata-only mode does not validate or repair the underlying audio codec.
With 双单声道转 Mono enabled, enable 清理 iXML(Pro Tools) to also clean WAVs that remain stereo or mono. Converted dual-mono files already drop iXML as part of the existing conversion behavior. The result table reports the iXML action separately from audio conversion.
Sample/chunk preservation is tested; Pro Tools import with the user's actual Cubase exports has not yet been verified. Other channel-layout metadata can also affect DAW interpretation, and this feature does not alter those fields. Avid documents support for BWF with or without iXML.
- Exact L/R comparison by default. Integer tolerance is in source-format sample units (LSBs). Float comparisons are always exact.
- Mono is the left channel multiplied by
sqrt(2): +3.0102999566 dB, preserving the established project behavior. This compensates a matching constant-power center pan law; it is not a universal requirement for every DAW pan configuration, nor exactly +3.0000 dB. - Integer output retains source depth unless the boosted samples would exceed the representable integer range. Positive full scale is one LSB below +1.0, so the float decision checks the actual signed limits.
- Over-range output is 32-bit IEEE float WAV, preserving values above 0 dBFS without limiting. Float inputs stay float. Hot AIFF becomes
track.aif.wav. - Sample rate, sample count, relative paths, and filesystem timestamps are retained. Integer gain is rounded symmetrically without added dither.
- During mono conversion, WAV BWF
bexttiming and identity are retained. Previous BWF v2 loudness fields are marked unknown after changing gain. Other custom metadata (iXML, cue/loop chunks, AIFF markers) is not copied. Metadata-only cleanup instead preserves all non-iXML chunks unchanged. - Silent stereo is identified and can be converted; zero-frame files are left unchanged.
Supported: RIFF WAV PCM 8/16/24/32-bit, IEEE float 32/64-bit; WAV extensible PCM/float with full-width valid bits; AIFF PCM and AIFF-C NONE/sowt/fl32/fl64. Unsupported formats, malformed or truncated data, and NaN/Inf samples produce per-file errors. RF64, compressed audio, packed valid-bit WAV, and outputs exceeding the 4 GB RIFF/AIFF container limit are not supported.
./Mononizer "/path/to/multitrack" --dry-run
./Mononizer "/path/to/multitrack" "/path/to/output"
./Mononizer "/path/to/multitrack" "/path/to/output" --copy-others
# Keep stereo/mono audio exactly as-is and strip iXML only:
./Mononizer "/path/to/multitrack" --metadata-only --dry-run
./Mononizer "/path/to/multitrack" --metadata-only
# Convert dual mono AND strip iXML from the other WAVs:
./Mononizer "/path/to/multitrack" --strip-ixmlOptions: --tolerance 1, --no-recursive, --json (streaming events). Exit status: 0 success, 1 file errors, 2 invalid invocation or folder setup, 130 cancelled.
The Python sources need Python 3.9+ and no audio packages. Packaging requires PyInstaller and a universal2 Python installation. Building the UI requires Apple's Swift/AppKit/SwiftUI SDK. GUI deployment target: macOS 12+. Actual oldest-OS compatibility also depends on the build runtime; test the release on the intended older OS.
/path/to/universal2/python3 -m venv .venv
.venv/bin/python -m pip install 'pyinstaller==6.22.3'
BUILD_PYTHON="$PWD/.venv/bin/python" ./build_app.sh
/usr/bin/python3 -m unittest -v
/usr/bin/python3 test_packaged.pybuild_app.sh rebuilds the engine before embedding it, avoiding stale converter binaries. --reuse-cli is available for UI-only development. Override BUILD_PYTHON to select another universal Python. build_macos.sh produces the standalone universal CLI archive.
The build is ad-hoc signed, not Developer ID signed or notarized. Both architectures can be verified with lipo; Intel execution on this build machine uses Rosetta, not a physical Intel Mac.