Home Apps & Tools Tutorials & Guides
Tutorials Audio

Clean Audio for Podcasts

Beginner · ~60 min

Overview

Good podcast audio isn't about expensive gear — it's about controlling your environment and knowing two or three techniques in post. This guide covers both.

What You Need

  • Any USB microphone (even a basic one works)
  • Audacity (free) or GarageBand (free on Mac)
  • A reasonably quiet room

Steps

1

Treat your room first

Sit in a corner — walls reflect less than open space. Add soft furnishings: a duvet behind you, a bookshelf in front, a rug below. Hard flat surfaces are the enemy. You're building a recording booth with what you already own.

2

Set your input level correctly

Your mic level should peak between −12dB and −6dB when speaking normally. Too quiet and you amplify noise when you boost in post. Too loud and you clip (which is unrecoverable). Set this before recording a single word.

3

Record 10 seconds of room tone

Sit silently and record the ambient noise before every session. This captures your room's unique background hum, which you'll use as a profile for noise reduction.

4

Apply noise reduction

In Audacity: select your room tone clip → Effect → Noise Reduction → Get Noise Profile. Then select your whole recording → Effect → Noise Reduction → OK. Start with 12dB reduction, sensitivity 6.

5

Compress your audio

Compression evens out the difference between quiet and loud moments. In Audacity: Effect → Compressor. Use Threshold: −18dB, Ratio: 3:1, Attack: 0.2s. This makes your voice sit more consistently in the mix.

6

Normalise to −16 LUFS

This is the target loudness for most podcast platforms (Spotify, Apple Podcasts). In Audacity: Effect → Loudness Normalization → −16 LUFS. Do this last, after all other processing.

Pro Tips

  • Record in mono, not stereo, for voice-only podcasts. Halves the file size with no quality loss.
  • Edit breaths and long pauses before you compress — you'll get cleaner, more consistent results.
  • Always listen back on headphones before publishing.