Professional-grade, high-performance audio visualization suite built for modern workflows.
PyDSPMeters is a modular, always-on-top audio monitoring suite for engineers, producers, and audio nerds who want critical signal feedback without giving up screen real estate. It targets Windows 10/11 and should also run on common Linux desktops, though Linux builds are less tested.
PyDSPMeters is built around compact audio blocks, vectorized DSP, and Qt painting that stays on the main thread where it belongs.
- Vectorized DSP: NumPy/SciPy handle FFTs, filters, loudness metering, and signal analysis.
- Rust native acceleration: The optional
dsp_accelmodule speeds up hot paths such as bulk array operations, circular buffers, waveform reduction, spectrogram column generation, and correlation helpers. - Worker-thread audio processing: Heavy module processing can run through a latest-only DSP worker pool. The Qt main thread stays focused on painting and window management while stale audio jobs are dropped instead of queued into lag.
- Runtime profiling tools:
profile_app.pyrecords stack samples, paint cadence, audio queue drain timing, module worker timing, and memory snapshots underprofile_logs/. - Circular buffer rendering: Modules such as Spectrogram use cached image buffers for smooth scrolling.
- Low-latency capture: SoundDevice/PortAudio feeds compact blocks into a Qt-side queue bridge.
- Adaptive UI: Dynamic text scaling and layout distribution keep meters readable from tiny strips to full-height panels.
PyDSPMeters adapts to your workflow.
- Stackable architecture: Add, remove, and reorder modules on the fly.
- Omni-layout: Toggle between vertical stacking and horizontal strips.
- Ghost mode: Transparent and glass presets combine with an auto-hiding title bar for floating meters.
- Edge snapping: Window management that locks to screen edges.
- AppBar functionality: Windows-only docking mode that reserves desktop space for PyDSPMeters.
| Module | Description | Key Features |
|---|---|---|
| Loudness | EBU R128 style metering | Integrated LUFS, RMS, true peak, worker-thread processing. |
| Spectrum | High-resolution FFT | Mel/log/linear scales, tilt, smoothing, and peak tracking. |
| Stereo | Phase analysis | Lissajous/vectorscope rendering and multi-band correlation. |
| Spectrogram | Frequency history | Circular heatmap buffer with customizable palettes. |
| Waveform | Amplitude history | Mirrored intensity-based waveform with real-time scaling. |
| VU Meter | Analog classic | Precision ballistics with themed peak and clip indicators. |
| Oscilloscope | Signal detail | Responsive waveform plotting at high zoom levels. |
Built-in presets include Midnight, Modern Light, Abyss, Transparent Ghost, Aurora, Crimson, Solar, and more. Custom themes can be created by editing the generated themes.json in the app data directory, or by using app/theme.py as a base.
- Python 3.9+
- An audio input device, Stereo Mix, loopback device, or virtual cable
git clone https://github.com/ThatOneFBIAgent/PyDSPMeters.git
cd PyDSPMeters
pip install -r requirements.txt
pip install --find-links ./wheels dsp_accel
python main.pywTo launch without a console on Windows, create a shortcut to pythonw:
pythonw "path/to/main.pyw"The build helper prefers Nuitka when available and falls back to PyInstaller:
python build_dist.py --yesFor a direct PyInstaller build that includes the executable icon, runtime icon file, and Rust accelerator:
python -m PyInstaller --noconfirm --onefile --windowed --name=PyDSPMeters --distpath=dist --workpath=build --clean --hidden-import=app --hidden-import=dsp_accel --hidden-import=dsp_accel.dsp_accel --collect-submodules=dsp_accel --icon=icon.ico --add-data "icon.ico;." main.pywThe output will be written to:
dist/PyDSPMeters.exe
The application logs whether the Rust accelerator is active at startup:
DSP accelerator active: True
PyDSPMeters includes a custom Rust crate, dsp_accel, built with PyO3 and maturin. If it is unavailable, the app falls back to Python/NumPy implementations so it can still run, just with less headroom.
Precompiled wheels are included under wheels/:
pip install --find-links ./wheels dsp_accelTo build the native module yourself:
cd native
maturin build --release
pip install target/wheels/dsp_accel-*.whl --force-reinstallOn Windows, install the Visual Studio "Desktop development with C++" workload if maturin cannot find a compiler.
Run a full app session with lightweight profiling:
python profile_app.py --interval 0.1 --top 80Close the app normally. A report folder will be written under profile_logs/, including:
profile_report.txt: human-readable timing report.profile.json: structured metrics.stacks.folded: sampled stacks for flamegraph tooling.profile.prof: generated only when--cprofileis used.
Useful report sections include paint cadence, audio queue drain timing, module audio time, worker-thread time, and ghost-titlebar update time.
- Right-click: Open module-specific settings.
- Gear icon: Device selection, themes, input gain, and label scaling.
- Plus icon: Add a module.
- Layout icon: Switch between vertical and horizontal layouts.
- Move mode: Rearrange modules without fighting splitter handles.
- Portability: Settings are stored in the app data directory, with source-run
settings.jsonignored by git.
python -m pytest --basetemp profile_logs/pytest_tmpThe explicit --basetemp keeps pytest's temporary files inside the workspace on locked-down Windows environments.
- Python 3.14 plus SoundDevice/CFFI can soft-crash after extended silence on some systems. PyDSPMeters includes a keep-alive workaround, but Python 3.13 or 3.12 may be more stable.
- Onefile executables built with PyInstaller/Nuitka can trigger antivirus false positives. Build from source if you want to inspect the full chain yourself.
- Linux support is community-tested. PRs and issue reports are welcome.
Licensed under the MIT License. Build, modify, and share.