Understanding Noise in Podcast Audio

Unwanted noise is the single biggest barrier to a professional-sounding podcast. It fatigues listeners, obscures dialogue, and can make even the most compelling content sound amateurish. Noise creeps in from many sources: your microphone’s own electronics, the room you’re recording in, electrical wiring, or even how you handle the mic. The first step to fixing noise is recognizing the type you’re dealing with:

  • Microphone self‑noise – a constant hiss produced by the mic’s internal circuitry. Cheaper mics and high‑gain condenser mics are the worst offenders.
  • Room ambiance – the low‑level rumble of HVAC, computer fans, road traffic, or echoing reflections from hard surfaces.
  • Electrical interference – a 60 Hz (or 50 Hz) hum from mains power, ground loops, or radio frequency interference from nearby electronics.
  • Handling noise – bumps, cable rustling, or vibrations transmitted through the mic stand or desk.
  • Plosives and sibilance – explosive “p” and “b” sounds and harsh “s” and “t” sounds that can mask or distort speech.

Use a spectral analyzer (built into your DAW or a free tool like iZotope’s spectrogram) to visually identify the frequency range of the noise. Once you know what you’re fighting, you can choose the right weapon.

Best Practices for Noise Reduction

Pre‑Production: Start Clean, End Clean

The most effective noise reduction happens before you hit record. Eliminating problems at the source saves hours of post‑production repair and preserves the natural tone of your voice.

  • Invest in a quality microphone and interface. Dynamic mics like the Shure SM7B, Electro‑Voice RE20, or Røde PodMic reject room noise and handling noise far better than large‑diaphragm condensers. Pair with an interface that has a low noise floor and high dynamic range, such as the Focusrite Scarlett series or Universal Audio Apollo.
  • Record in a treated space. You don’t need an expensive studio. A closet full of clothes, a whisper room, or even a homemade PVC‑and‑blanket booth can cut reverb and background rumble by 80 %. Place acoustic foam or heavy moving blankets at reflection points around the mic.
  • Set gain correctly. A too‑low input level forces you to boost the signal later, amplifying the mic’s self‑noise and any ambient hum. Aim for peaks between -12 dBFS and -6 dBFS. Use your interface’s meter, not the DAW’s after conversion.
  • Use a noise gate sparingly during tracking. Gates can silence breaths and low‑level noise between sentences, but an aggressive gate sounds choppy. Use a soft knee and set the threshold just above the noise floor (around -50 dBFS). Keep the hold time short (10‑30 ms) to avoid cutting off the end of words.

During Recording: Capture a Noise Profile

Many modern noise reduction algorithms (Audacity, iZotope RX, Adobe Audition, Waves WLM) require a “noise print” – a sample of the room tone alone. Record 5‑10 seconds of pure silence at the start of every session, in the exact same spot and with the same mic settings you’ll use for speaking.

Tips for a clean noise print:

  • Ask everyone to stay completely still – no breathing, shuffling, or rustling.
  • If the noise is cyclical (e.g., an air conditioner that cycles on and off every few minutes), record a longer sample that covers a full cycle.
  • If you change positions or microphones during the podcast, capture a new noise print for each variation.

This small habit is the single most valuable thing you can do for noise reduction. It gives the algorithm a precise map of the noise signature, allowing it to subtract only the noise and leave your voice untouched.

Post‑Production: Essential Reduction Workflow

Once you have your noise print, apply noise reduction with a careful, step‑by‑step approach. Aggressive settings will destroy vocal quality – always err on the side of subtlety.

  1. Apply noise reduction first. Use iZotope RX Voice De‑noise, Audacity’s built‑in Noise Reduction, or a dedicated plugin like Waves NS1. Start with a reduction amount of 5‑8 dB. Listen for artefacts – warble, “underwater” effect, metallic ringing. If you hear any, back off by 2‑3 dB. Less is more. You can always apply a second pass with a different noise print later.
  2. High‑pass filter. Most vocal content lives above 80 Hz. Roll off everything below 80 Hz with a steep curve (12‑24 dB/octave). This removes rumble from HVAC, traffic, and mic handling. For voices with very low fundamentals, try 60 Hz instead.
  3. Notch out hums. Use a narrow EQ notch (Q of 10‑30) at 60 Hz (or 50 Hz if you’re in a 50 Hz region) and its harmonics (120 Hz, 180 Hz, 240 Hz). Sweep through the low‑mid range (200‑500 Hz) and listen for any boxy, roomy resonances – notch those out gently too.
  4. Compress after noise reduction. Compression evens out volume differences, which makes any remaining noise less apparent between words. Use a ratio of 2:1 to 4:1, attack of 10‑30 ms, release of 40‑80 ms, and aim for no more than 6 dB of gain reduction. Over‑compression brings up the noise floor.
  5. Finish with a subtle gate. A gate with threshold around -45 dBFS, attack 1‑2 ms, release 150‑300 ms will clean up breath tails and lip smacks without truncating speech. Set the gate to close only during pauses of at least 200 ms to avoid choppy transitions.

Always listen on multiple playback systems – studio monitors, earbuds, laptop speakers, and a car stereo. What sounds clean in a treated room may still have audible hiss on consumer gear.

Advanced Techniques for Stubborn Noise

Some noise problems resist standard reduction. If you’ve got persistent clicks, crackles, or momentary spikes (like a dog bark or a passing siren), turn to these advanced methods:

  • De‑click / De‑crackle. Mouth clicks and plosive remnants are impulsive – they occupy a brief time window. Dedicated plugins like iZotope RX De‑click, Accusonus ERA De‑clicker, or Steinberg Clean can remove them with minimal impact on the voice. Use them sparingly – too much de‑clicking creates a “waxy” sound.
  • Spectral editing. In the spectrogram view of iZotope RX, Adobe Audition, or Audacity, you can literally paint over noise artefacts. This is the only way to remove a one‑time sound like a cough, a car horn, or a dropped book. Zoom in, select the offending pixels, and silence or subtract them.
  • Multiband compression. Instead of compressing the whole signal, compress only the frequency band where the noise lives. For hiss, that’s usually 4‑8 kHz. Use a multiband compressor (e.g., FabFilter Pro‑MB, Waves C4) and set the threshold so that only the hiss band gets reduced. This keeps the clarity and presence of the voice intact.
  • Phase cancellation. If you recorded with two microphones (e.g., one on each speaker) or a dual‑capsule mic, you can sometimes invert the phase of one track to cancel out common noise. This is very advanced and works only if the mics are identical and positioned the same distance from the noise source. It’s rarely worth the trouble for standard podcasts.

Common Mistakes to Avoid During Noise Reduction

Even experienced podcasters make these errors. Save yourself the headache by avoiding them:

  • Over‑relying on plug‑ins. No amount of software can fix a terrible recording. If the room is too live, the mic is too hot, or the gain is too low, re‑record. Aggressive noise reduction always sounds unnatural.
  • Using one setting for the whole track. Background noise changes – an air conditioner cycles, a truck passes, the speaker moves closer to the mic. Use automation or adaptive plugins that adjust in real time (like iZotope RX’s Adaptive mode). Or split the audio into sections and apply different noise prints to each.
  • Ignoring plugin chain order. Noise reduction should come before EQ and compression. If you compress first, you raise the noise floor. If you EQ first, you may shape frequencies that the noise reduction algorithm relies on. Stick to: noise reduction → high‑pass → de‑es · de‑click → EQ → compression → gate.
  • Applying noise reduction before cleanup. Always silence coughs, long breaths, lip smacks, and other non‑speech events first. Why? Because noise reduction algorithms try to process those sounds, often amplifying them or leaving artefacts. Strip silence or manually delete them before you run any noise reduction.
  • Not referencing on multiple systems. What sounds clean on your studio monitors may have audible hiss on a smartphone speaker. Listen to your final master on earbuds, a laptop speaker, and a Bluetooth speaker. If the noise is still there, go back to step one.

Building a Noise‑Free Workflow

Integrate noise reduction into your regular mastering chain. Consistency is key – if you capture a noise print in every session and apply the same gentle reduction routine, your listeners will never notice the noise they aren’t hearing.

  • Create a mastering template with your favourite noise reduction plugin, high‑pass filter, and compressor already set up with sensible defaults.
  • Use a loudness meter to ensure your final track hits -16 dB LUFS (or -19 LUFS for stereo) – this is the standard for most podcast platforms. Over‑loud masters expose noise.
  • Export a 30‑second sample and listen on a pair of cheap earbuds before committing to the full export.

For additional guidance, check out Sound On Sound’s guide to noise reduction tools, Music Production Nerds’ podcast audio cleaning checklist, and Podcast Engineer’s walkthrough of RX Voice De‑noise.

Conclusion

Mastering a podcast is about preserving the natural quality of a voice while eliminating distractions. Noise reduction is a critical piece of that puzzle, but it must be handled with care. Start with a quiet room, a proper microphone, and good gain structure. During post‑production, use noise profiles, gentle EQ, and thoughtful compression. For persistent problems, explore spectral editing and de‑click tools. By following these best practices, you’ll produce episodes that sound professional, engaging, and fatigue‑free – no matter where your listeners press play.

Your ears are your best tool. Listen critically, compare often, and always let the voice be the star.