matlab-process-streaming-audio

Use Audio Toolbox streaming objects such as crossoverFilter, compressor, and multibandParametricEQ inside a loop driven by dsp.AudioFileReader and isDone. These objects maintain internal state across frames, and you call release on all objects after the loop completes.

Use audioLevelMeter for sample-peak or true-peak levels in dBFS/dBTP, loudnessMeter for EBU R128 loudness in LUFS, splMeter for per-band sound pressure levels, and octaveSpectrumEstimator for weighted octave-band spectra. All except splMeter support the visualize method for built-in real-time displays.

Call parameterTuner(obj) on any Audio Toolbox streaming object to open a slider UI for all tunable properties, and visualize(obj) for a live response plot. Include drawnow limitrate inside the processing loop so UI events are flushed and tuning changes take effect.

Use dsp.STFT with dsp.ISTFT when you need per-bin spectral manipulation such as gating or subtraction, since they handle windowing, overlap, and perfect reconstruction. Use dsp.FrequencyDomainFIRFilter only for convolving with a fixed long impulse response, such as room IRs or convolution reverb.

MATLAB does not process UI events inside a tight processing loop, so slider changes are ignored. Add drawnow limitrate inside the loop, and set PlayCount=Inf on dsp.AudioFileReader so the file does not end before you finish tuning.

Avoid this approach for audio device I/O setup, audio plugin generation with createAudioPluginClass, deep learning audio inference, and offline batch processing of entire files. Those workflows use different tools such as audiostreamer or one-shot functions like audioresample.