How to edit a podcast in Audacity
Editing a podcast in Audacity 4 is one chain, always in the same order: import and arrange the takes, cut and tighten, clean up the noise, level the voices to a loudness target, add music and sound effects, then export an MP3. Every step has a built-in tool, and most have a plugin on MuseHub that does the same job faster.
This guide walks the whole chain with the menu path, a starting setting and the mistake to avoid at each step, a one-table summary, the loudness numbers Apple and Spotify publish, and the podcasting plugins to add once the built-in route feels slow.
Tested on Audacity 4.0.1 on macOS Sequoia; the screenshots are from that install. Shortcuts are written for Windows; macOS uses Cmd where Windows uses Ctrl. Where Audacity 3 differs, the step says so.
Audacity and podcast tools on MuseHub
What you need
- Audacity 4 — free, for Windows, macOS and Linux, from the Apps section of MuseHub or from audacityteam.org.
- Your recordings, ideally one file per voice. If you have not recorded yet, recording computer audio in Audacity covers a guest on a call.
- Headphones. Bad cuts and noise-reduction artefacts hide on laptop speakers.
- The MuseHub app, free for Windows and macOS with a free account, for the plugins and packs in steps 3 to 5.
The podcast chain at a glance
Six steps, in the order they should happen, because each one changes what the next one measures. Compression before cleanup makes the noise louder; loudness normalization before the music is in place measures the wrong thing.
| Step | Built-in tool in Audacity 4 | Where | Starting point |
|---|---|---|---|
| 1. Import and arrange | Import, clips, tracks | File ▸ Import, the tracks area | One track per voice, one for music, one for effects |
| 2. Cut and tighten | Delete and close gap, Split tool, Truncate Silence | Edit menu, S, Effect ▸ Special | Ctrl+Delete closes a gap on every track |
| 3. Clean | Noise Reduction, High-pass Filter | Effect ▸ Noise removal and repair, Effect ▸ EQ and filters | 6 dB of reduction, up to 12 dB |
| 4. Level | Compressor, Loudness Normalization, Limiter | Effect ▸ Volume and compression, the Master section | −16 LUFS, peaks at −1 dB |
| 5. Music and effects | Fade In, Fade Out, track volume, a Clip gain envelope | Effect ▸ Fading, the track panel, the toolbar | Music well under the voice |
| 6. Export | Export audio | File ▸ Export audio… (Ctrl+Shift+E) | MP3, 128 kbps mono for a talk show |
Step 1: Import and arrange the takes
Open File ▸ Import (Ctrl+Shift+I) and choose your files, or drag them onto the tracks area. Each file arrives as a clip on its own track. A track is a lane on the timeline; a clip is a block of audio inside it that you can move, trim and rename on its own. Importing adds to the open project, so bring in every take, the intro music and the sound effects in one go.
Arrange before you cut:
- One track per voice, so you can clean and level each one on its own. If your recorder put both voices in one stereo file, open the track’s ⋯ menu and choose Split stereo to center mono. Each channel becomes a mono track of its own, panned to the center.
- One stereo track for music, one for effects, added with Tracks ▸ New stereo track. Off the voice tracks, they can be moved and leveled independently.
- Name everything. Double-click a clip’s header, the strip along its top, or right-click it and choose Rename. Tracks rename the same way, with a double-click on the track’s name or Rename on its right-click menu.
- Save the project. File ▸ Save (Ctrl+S) writes an
.aup4file that keeps every track, trim and effect editable. Exporting later produces the MP3; saving is what lets you change a cut tomorrow.
Voice recordings are usually mono, one channel, and that is right for speech. A stereo voice file with identical left and right channels only costs bitrate at export. To convert it, choose Split stereo to center mono from the track’s ⋯ menu, then remove one of the two identical tracks it leaves with Delete in that track’s ⋯ menu.
Step 2: Cut and tighten
The everyday edit is a deletion. Drag across the waveform to select the false start, then press Delete. The first time, Audacity asks what to do with the space. Leave gap replaces the audio with silence so everything after it stays put. Close gap (ripple) slides the later audio back to fill the hole. Ripple editing is what you want for speech; it is why the show gets shorter when you cut.
Two shortcuts ripple without the question. Shift+Delete closes the gap on the selected track only. Ctrl+Delete closes it on every track at once, and that is the one for a podcast. Ripple the host track alone and the guest and the music stay where they were, drifting further out of sync with every cut. Audacity 3 used the Sync-Lock option for this; Audacity 4 replaced it with ripple editing.
To rearrange rather than remove, press S for the split tool and click where you want the cut; a guideline follows the pointer so you can land it in a pause. Then drag the new clip by its header to where it belongs. The handles on a selected clip trim its ends without deleting anything, so a trim is undone by dragging the edge back out. Ctrl+L replaces a selection with silence at the same length, the right fix for a cough that sits over a music bed.
For the pauses, let an effect do the counting. Truncate Silence finds every stretch quieter than a threshold and shortens it. Select all the voice tracks and open Effect ▸ Special ▸ Truncate silence. The defaults treat anything below −20 dB that lasts 0.5 s as silence and cut it to 0.5 s. If it clips the ends of quiet words, lower Threshold toward the noise floor; if the pacing feels rushed, raise Truncate to. Leave Truncate tracks independently off so every track shortens together and the guest stays in sync with the host. Every control is on the Truncate Silence manual page.
For a slower walk through each of these edits, with a screenshot per move, see how to trim and cut audio in Audacity.
Step 3: Clean up the noise
Clean before you compress. Compression raises the quiet parts of a recording, and the noise floor is the quietest part, so any hiss you leave now gets louder in step 4.
Noise Reduction learns what the noise sounds like and subtracts it. Select a second or two of room tone with nobody talking and open Effect ▸ Noise removal and repair ▸ Noise reduction; click Get noise profile and the dialog closes. Now select the whole voice track (Ctrl+A) and open the effect again. Start at the default 6 dB of Noise reduction and raise it toward 12 dB only if the hiss still bothers you. Past that the voice goes hollow and watery, the most common sound of an over-edited podcast. Preview, apply, and repeat for each voice track with its own profile. What each slider does, and the noises the effect cannot fix, are in Audacity noise reduction.
Two smaller cleanups take a minute. A High-pass Filter (Effect ▸ EQ and filters ▸ High-pass filter) rolls off the rumble below the voice, the desk thumps and traffic a listener feels more than hears. It opens with Frequency (Hz) at 1000, which would thin the voice itself, so lower it to around 80 Hz, the usual starting point for speech, before you apply it. Click Removal (Effect ▸ Noise removal and repair ▸ Click removal) catches the mouth clicks a noise profile never will. It opens at Threshold 200 and Max spike width 20; lower the threshold, or raise the spike width, if clicks get through.
The AI route. OpenVINO AI Tools, a free plugin from Muse Group and Intel, adds a Noise Suppression effect trained on speech that needs no noise profile, the better pick when the recording never goes quiet. It runs in Audacity 3 only for now, so it means installing Audacity 3 beside Audacity 4 and cleaning the file there before you edit.
Step 4: Level the voices
Three effects do three different jobs here, and they get mixed up more than any other part of the chain. A Compressor narrows the gap between the loud words and the quiet ones. Loudness Normalization sets the average perceived loudness of a track to a target in LUFS (Loudness Units relative to Full Scale, the measure podcast platforms use). A Limiter stops the peaks from passing a ceiling. Normalize, in the same menu, only moves the highest peak; a track can be peak-normalized to −1 dB and still be far too quiet on average, so it is the wrong leveling tool for speech.
Do them in this order.
- Compress each voice track. Select the track and open Effect ▸ Volume and compression ▸ Compressor. The defaults, Threshold −12 dB, Ratio 4, Attack 3 ms, Release 100 ms and Make-up gain 9 dB, are a usable speech setting; lower the threshold if only the loudest words move the meter. Applying it from the Effect menu changes the waveform, which is what you want here, because the next effect measures the waveform. How to use a compressor explains each control.
- Normalize loudness to the platform target. With the voice track selected, open Effect ▸ Volume and compression ▸ Loudness Normalization, choose Perceived loudness and type the target. The default is −23 LUFS, the broadcast figure; podcasts run louder. Leave Treat mono as dual mono (recommended) on, so a mono voice track is measured the way it plays back, on both ears.
- Put a Limiter on the Master. Press E to open the realtime effects panel, scroll to the Master section at the bottom and add Limiter. Its default Output of −1 dB keeps every peak in the finished mix under the ceiling Apple sets, and because it sits on the master it also catches the moments where a voice, the music and an effect stack up.
Which target? These are the figures the platforms publish; other numbers you will read online are estimates.
| Platform | Loudness | Peak ceiling | Source |
|---|---|---|---|
| Apple Podcasts | −16 LKFS, ±1 dB | −1 dB FS true peak | Apple Podcasts audio requirements |
| Spotify | No loudness figure in its podcast delivery specification; the Spotify player normalizes music to −14 LUFS | None for podcasts; the music page recommends −1 dB true peak | Spotify Podcast Delivery Specification v1.10; Spotify loudness normalization page |
LKFS and LUFS are the same unit under two names. Set −16 LUFS in Audacity and the file meets Apple’s spec; Spotify’s specification asks only for a format and a bitrate, so the same file passes there too.
One thing the numbers hide: normalizing each voice to −16 LUFS and then adding music makes the mix a little louder than −16. A quiet bed, as step 5 describes, keeps the difference inside Apple’s ±1 dB. For an exact result, render the finished mix through a file, since Audacity 4.0.1 has no command that mixes tracks down inside the project. Save the project, export the mix as WAV with File ▸ Export audio… and Type Full project audio, then bring the WAV back with File ▸ Import; it lands on a track of its own. Mute every other track, run Loudness Normalization once on the new one, and export the MP3 from there with the Master limiter still in place.
How to make audio louder in Audacity compares Amplify, Normalize, Loudness Normalization, Compressor and Limiter side by side, and what is a limiter covers the ceiling.
Step 5: Add music and sound effects
Music and effects live on their own tracks, never pasted into a voice. Drag the intro clip by its header until it sits where the show starts, and slide the outro under the last sentence. Sound effects work the same way, one clip each, on the effects track.
- Fade it. Select the last few seconds of the intro and apply Effect ▸ Fading ▸ Fade Out; it applies at once, with no dialog. Fade In sits beside it. A hard start on a music bed is the most audible sign of an unfinished edit.
- Keep it under the voice. The volume slider on the left of the music track sets its level for the whole episode, and a bed almost always wants to be quieter than you first set it. For a bed that dips under speech and rises in the gaps, click Clip gain in the toolbar. Every clip shows a volume line; click the line to add a point and drag a point to move it. Audacity 3’s Auto Duck effect, which did this automatically, is not in the Audacity 4 Effect menu.
The five types of podcast sound effect, where each goes and what “royalty-free” covers are in podcast sound effects. MuseHub sells paid, royalty-free packs for this job; you buy once and use the sounds in every episode. Two made for podcasts:
Podcast Creator SFX by Audio Hero is a pack of sound effects for spoken-word shows, the stingers and transitions that mark a segment change without competing with the voice.
Audacity Podcast Jingles by Tha Loops is a set of short musical jingles, the intro or stinger a show keeps as its signature from episode to episode.
Step 6: Export the MP3
Open File ▸ Export audio… (Ctrl+Shift+E); if Audacity first asks where to save, click Export to computer. The dialog asks for a file name and folder, then the settings that matter:
- Type: Full project audio, which mixes every track through its volume, its effects and the Master limiter, exactly as playback does.
- Format: MP3 (listed as MP3 Files). Every podcast host takes it, and Apple and Spotify both list it first.
- Channels: Mono for a voice-only show, Stereo when the music matters to the listener. Mono halves the data the encoder has to spend, so a mono file sounds better than a stereo one at the same bitrate.
- Sample rate: 44100 Hz, the default. Apple accepts 44.1 or 48 kHz.
- Bit rate: set Bit Rate Mode to Constant, then set Quality to 128 kbps mono for a talk show or 192 kbps stereo when there is music. The default, Preset at Standard quality (170–210 kbps), varies the rate and runs above Apple’s range for mono. Apple recommends 96–128 kbps for mono and 128–256 kbps for stereo; Spotify asks for 128 kbps or more. 320 kbps mostly buys a bigger file.
Export once as MP3 for upload and once as WAV for your archive. An MP3 made from an MP3 loses quality a second time, so a re-edit starts from the WAV or the saved project. Title and author fields are behind the dialog’s Edit metadata button.
Common mistakes
- Too much noise reduction. Above 12 dB the voice turns watery. A little hiss under a natural voice beats silence under a damaged one.
- Compressing before cleaning. The compressor raises the noise floor with everything else.
- Peak-normalizing and calling it done. Normalize moves one peak; Loudness Normalization moves the average, which is what a listener hears and what Apple measures.
- Rippling one track. Shift+Delete leaves the other tracks where they were, and every later cut pulls the guest further out of sync. Use Ctrl+Delete.
- Music set on speakers. Check the mix in earbuds and on a phone; the bed will be too loud.
- Skipping the project file. An exported MP3 cannot be re-edited. Save the
.aup4project too, and archive a WAV.
Habits that make the next episode faster
- Record five seconds of room tone at the start of every session. That is your noise profile, guaranteed clean.
- Reuse a finished episode’s project as the template for the next one: same tracks, same Master limiter, new clips.
- Preview before you apply, and A/B against the untouched audio. Then listen through once at the end, in headphones, without touching anything.
Starting points by show type
- Solo narration. One mono track. Noise Reduction at 6 dB, Compressor defaults, Loudness Normalization to −16 LUFS, Limiter on the Master, export 128 kbps mono.
- Two-voice interview. Two mono tracks, each with its own noise profile, compressor and −16 LUFS pass so the voices match. Cut with Ctrl+Delete only. Export 128 kbps mono.
- Interview with a music bed. The same, plus a stereo music track with fades and an envelope dip under speech. Export 192 kbps stereo.
Podcasting plugins on MuseHub
The built-in chain gets an episode published. Plugins add speed and repeatability: one window instead of four dialogs, and settings you set once and reuse every week. They run as realtime effects, so you hear the change while you adjust and nothing is written to the waveform until export. Press E, click Add effect and choose the plugin. How to use effects and plugins in Audacity covers installing them through MuseHub and the formats Audacity 4 loads.
Viator Voice Denoiser by Viator DSP is free and is step 3 as a plugin: one-knob noise reduction for voice, with no noise profile to capture, that stays on the track and is adjusted by ear.
Viator Voice Leveler, also free, is step 4 as a single control: one-knob compression and leveling that evens out a voice without the threshold and ratio setup.
Soap Voice Cleaner by Soap Audio is paid and rolls steps 3 and 4 into one pass aimed at spoken voice, the pick for a weekly show where the same room and microphone need the same fix every time.
The free MuseFX collection from Muse Group covers the general mix jobs, level and tone included. The ranked view of all of these, free and paid, is the best podcasting plugins. For the whole process on video, MuseHub carries the course Podcasting With Audacity.
FAQ
Next steps
The two ends of this chain have their own guides: recording computer audio covers the capture, and how to mix your podcast goes deeper on compression and EQ once the edit is done. Browse the plugins on MuseHub for the voice tools above, and get Audacity and the free ones in one place — download MuseHub free.
