High-pass filter

It's way easier to use for audio stuff than FFmpeg is, and then he lists what he actually does with it.

High pass, normalize, automatically trim audio files, add fade in or fade out, down mix to mono, then resample and dither.

That's a whole mastering chain in one sentence.

And the real-world pattern shows up in that police scanner transcription project, where dockerized FFmpeg processes, slice the streamed audio into 30-second WAV files, and then SoEx removes the silence.

Equalizer, which is an RBJ peaking EQ biquad.

Bass and Treble, which are RBJ shelving biquads.

High Pass and Low Pass.

Band Pass and Band Reject.

Pass means what gets through.

A high pass filter, or HPF, lets highs pass and reduces lows.

It is the remove rumble or mud move.

A low pass filter, LPF, lets lows pass and reduces highs.

In voice work, too much low content can muddy harmonics and cause bleed, especially on small monitors.

So we often use an HPF before even applying any EQ or compression.

Low-pass filtering trims the high side of a vocal spectrum, normally around eight, ten kilohertz for a telephone vibe.

The human voice naturally extends far beyond that, so by cutting everything above the cutoff, we lose sibilance and clarity but gain that soft, muffled feel typical of handset audio.

Setting an envelope that gradually lowers the high cut at the start of a bridge and then ramps up again gives a natural growl before the drop, adding a subtle movement that feels organic rather than mechanical.

When working on radio simulation, we should also consider bandwidth constraints of different transmitters.

For example, AM radio can only pass about three kilohertz, so an HPF at three hundred hertz and LPF at three kilohertz produces a voice that sounds as if it's being broadcast live over the air, great for tracks that reference oldies or for cinematic audio cues.

If you need source material that already feels band limited, Soundstock provides curated packs of telephone-style vocals and voice loops.