Podcast Audio Cleanup: A Practical Checklist
Most rough podcast recordings share the same three problems: a low-frequency rumble from handling noise or AC hum, a steady hiss or hum sitting under the voice, and loudness that jumps around between takes or speakers. The fixes are simple individually, but the order you apply them in actually matters — doing it out of order makes each later step work harder and perform worse.
Step 1: Cut low-frequency rumble first
Rumble — the low, muddy energy from handling noise, desk vibration, HVAC hum, or a mic stand picking up footsteps — sits well below the range of the human voice. Cutting it is a high-pass filter, not a denoiser: it removes a whole frequency band rather than trying to distinguish noise from signal. Doing this first matters because rumble can otherwise confuse a denoising step in the next stage — some of that low-end energy can get mistaken for part of the noise profile, making the denoiser work less precisely than it would on a recording that's already had the rumble cut clean.
Step 2: Apply speech-tuned noise reduction
With rumble already removed, a denoiser tuned specifically for speech can focus entirely on the noise that's actually left — hiss, hum, background static — without needing to also account for low-end rumble it would otherwise have to work around. A speech-specific preset can be more targeted than a general denoiser precisely because it only has to protect one type of signal: the frequency range of a human voice, not the much wider range music or field recordings can occupy.
Step 3: Normalize loudness last
Loudness normalization should come after cleanup, not before. Normalizing first means you're setting levels based on a recording that still has rumble and noise contributing to its overall level — the normalization ends up calibrated against content you're about to remove. Normalizing after cleanup sets the final loudness based on what will actually ship, giving a more accurate and consistent result, especially across multiple takes or speakers recorded at different levels.
Why order matters more than it seems
Each of these three steps changes the input the next step sees. Skip the rumble cut and the denoiser has to work around low-end energy it wasn't designed to prioritize. Normalize before cleanup and you're calibrating loudness against noise you're about to strip out. The fixed order — rumble, then denoise, then normalize — exists because each stage performs better when it's only solving the problem it's actually built for, not compensating for a step that should have happened earlier.
Doing it in one pass
Running these three steps manually with separate tools means getting the order right yourself every time. Our Voice Cleaner runs this exact chain — rumble cut, speech-tuned denoise, loudness normalization — in the correct order automatically, with nothing to configure. It's built specifically for speech content: podcasts, interviews, phone recordings, and voice memos. If you're working with music or need to control noise reduction strength directly instead, the Noise Remover is the better fit.