Audio Wandas Analyzer
English | 日本語
Find what sounds different, then see why. Line up recordings, listen and look on one timeline, and overlay their spectra and other analyses on the same axes.
▶ Try it in your browser: press Try with samples, no files or install needed. Your audio never leaves your device. For long recordings and calibration, install the VS Code extension.

The bundled samples: the same hum three times, one with a 2.8 kHz tone and one with clicks.
How it works
- Open files or a whole folder. Every file is listed in a table; tick ☐ to show it as a track.
- Listen and look. Each track shows its waveform over its spectrogram on one timeline, with the spectrum at the cursor beside it. Play, loop, mute and nudge tracks to line them up.
- Compare. Pick a base and up to three recordings to compare (keys
0–3). Their results are drawn over each other, one colour per role: blue for the base; yellow, purple and magenta for the others.
- Analyze as a chain. The step bar is a wandas method chain, such as Input → HPF 20 Hz → Average spectrum. + adds a wandas operation anywhere in it; click a step to see its result for every recording.
- Take it with you. Export PNG, CSV, WAV or a Markdown / notebook report, or save the chain as a
.recipe.json that wandas can run on its own.


Also in the box
- Many files: the table shows a value per file (such as A-weighted Leq) to sort, filter and group, and takes an attribute CSV (anomaly scores, conditions…) to narrow down the candidates.
- Reference track: transfer function and coherence from one track to the others.
- Calibration to Pa, m/s² or V, with levels in dB SPL (VS Code).
- Bundled recipes, including loudness and roughness.
- Formats: WAV, FLAC, OGG, OGA, Opus, MP3, AIFF, AIF, AU, SND, CAF, W64, RF64 and CSV (first column = time). Playback covers WAV, FLAC, OGG, OGA, Opus and MP3.
- English and Japanese, following your VS Code or browser language. Press ? Help for the keys.
Browser or VS Code
|
Browser |
VS Code |
| Setup |
None: open the page |
Python 3.11+ |
| Files |
Short files (see Browser limits) |
Large and long files |
| Calibration |
Not available |
Available |
| Recipes |
Bundled and your own .recipe.json; steps that need Python packages the browser lacks do not run |
Anything your Python environment can run |
Get started in VS Code
- Install Python 3.11+ and this extension.
- Right-click an audio file or folder in the Explorer and choose Analyze with Audio Analyzer.
- If the Python packages are missing, choose Create .venv and Install. The extension creates a virtual environment in your workspace and installs them.
If that button does not appear (an untrusted workspace, or a .venv already there), or to use an environment of your own, install the packages into it and run Audio Analyzer: Select Python Environment:
pip install "wandas[psychoacoustic,effects]>=0.13.0,<0.14.0" "numpy>=2.0.2" "scipy>=1.13" "soundfile>=0.12"
Files and folders can also be opened from the Audio Analyzer view in the Activity Bar (or by dropping them on it), or with Audio Analyzer: Analyze File or Folder.
Reference
Settings
| Key |
Default |
Description |
audioWandasAnalyzer.pythonCommand |
python3 |
Python environment folder or executable used for the backend |
audioWandasAnalyzer.cacheMemoryMb |
1024 |
Maximum memory (MB) used to cache loaded audio |
In a workspace you have not trusted, the extension runs in limited mode: analysis works, but audioWandasAnalyzer.pythonCommand and audioWandasAnalyzer.debugFilePath are read from your user settings only, so a folder's .vscode/settings.json cannot choose the program the extension starts. Virtual workspaces (such as a GitHub repository opened in the browser) are not supported, because audio is read from the local file system.
Folder listing skips hidden folders (such as .git or .venv), node_modules and symbolic links to folders, and stops after 50,000 entries; open a subfolder to see the rest.
Troubleshooting
- Python interpreter was not found: select the environment where the packages are installed, or update
audioWandasAnalyzer.pythonCommand.
- Analysis failed: open Output: Audio Wandas Analyzer and check the Python error.
- A file does not load: check that its format is in the list above. M4A is not supported.
- Large files feel slow: the waveform is fetched only for the visible range; for spectrograms, try a smaller FFT size or a lower overlap.
Browser limits
Audio is analyzed on your device and is not uploaded. Display settings are remembered between visits; files, cursor and offsets are not, so select your files again after reloading.
| Limit |
Browser version |
| Formats |
WAV, FLAC, OGG, OGA, Opus, MP3, AIFF, AU, SND, CAF, W64, RF64, CSV |
| Per file |
Up to 80 MiB; mono or stereo |
| Sample rate |
1–96 kHz |
| Comparison |
Up to 8 tracks, 256 MiB of input files and about 17 minutes of 48 kHz stereo audio in total |
| WAV export |
Uncalibrated PCM16 audio; up to 32 MiB in total |
Use a desktop browser for the full set of mouse controls. Exported notebooks refer to the audio by file name, so place copies of the audio beside the notebook.
Links
| |