Loading page
About this tool
Reduce hiss, hum and room echo in a podcast recording.
The Podcast De-noiser & Echo Reducer cleans a spoken-word recording in your browser by running an FFmpeg chain of a 70 Hz high-pass, a 14 kHz low-pass and the afftdn FFT noise reducer at one of three noise-floor settings (-22, -28 or -35 dB). Load an audio file, pick light, medium or strong, and a cleaned WAV is written back out without the file ever leaving the machine. It is for podcasters and interviewers who recorded in an untreated room and need hum, rumble and tape-style hiss pulled down before editing.
Open Podcast De-noiser & Echo Reducer on AltFTool — it loads instantly in your browser.
Pick your recording in the Source file field, which accepts audio/* files.
Set Strength to light, medium or strong — afftdn noise floors of -22, -28 and -35 dB — and press Process locally.
The cleaned altftool-podcast-de-noiser-echo-reducer.wav is written out and downloaded from the page.
Light, medium and strong map to concrete afftdn noise floors of -22, -28 and -35 dB rather than a vague slider, so you can retry one notch at a time when speech starts sounding processed.
FFmpeg is compiled to WebAssembly and loaded only when you press Process; the audio is written to its in-memory filesystem, never to a server.
Output is uncompressed WAV, so the cleaned track goes into your editor without a second generation of lossy encoding.
Only the FFT denoiser's noise-floor figure: light uses -22 dB, medium -28 dB and strong -35 dB. The band limiting is identical at every setting — a high-pass at 70 Hz to drop rumble and mains hum, and a low-pass at 14 kHz to cut hiss above the speech band.
Not really — it reduces broadband noise, not reverb tails. Filtering and spectral subtraction thin out hiss and hum and can make a room sound slightly drier, but true de-reverberation needs a different class of processing; the reliable fix is recording closer to the mic in a softer room.
No. Processing runs entirely in the browser tab through FFmpeg WebAssembly; only the FFmpeg core binary itself is fetched, and that happens after you click Process, not when the page loads.
Because at a -35 dB noise floor the FFT denoiser starts subtracting energy that belongs to speech, not just to noise. Step back to medium or light, and remember the 14 kHz low-pass also removes air and sibilance detail at every setting.