How to tighten podcast edits by automatically removing gaps
- Step 1Export the voice stem from your DAW — Render the edited voice track (or per-guest stems) to WAV, then drop it onto the silence-stripper tool. Stripping a clean voice stem — not a music-and-voice mix — gives the most predictable result. Processing is local; nothing uploads.
- Step 2Set the threshold for the stem's noise floor — On a clean rendered stem the default
-40 dBusually works; on a noisier raw track raise it to-35. Threshold (dB) decides what's a gap. Keep it consistent across episodes of the same show for a uniform house style. Range-80…-10, step 1. - Step 3Dial the tightness with Min silence (s) — Min silence (s) is the air left at each cut — your tightness dial.
0.3 sfor a punchy, fast show;0.5–0.8 sfor conversational. Pick one value per show and stick to it. Range0.1…5, step 0.1. - Step 4Run the strip — FFmpeg applies
silenceremovewithdetection=peak: leading silence trimmed once (start_periods=1), every internal and trailing gap tightened (stop_periods=-1). Only time changes — your levels and EQ from the DAW are untouched. - Step 5QC against your house style — A/B the tightened stem against a reference episode. If transitions feel rushed, raise Min silence by
0.1; if it's still baggy, lower it. Once dialled, those two numbers are your repeatable spec for the show. - Step 6Drop it back on the timeline — Re-import
stem-tightened.wavto your DAW for music beds, ads, and final QC. Then run loudness on the full mix with loudness-normalizer, or hand the whole finishing chain to podcast-master.
The two real controls
These are the only options the silence stripper exposes, verified against the tool's client component and the FFmpeg filter built in the processor.
| Control | FFmpeg parameter | Default | Range / step | Editor's meaning |
|---|---|---|---|---|
| Threshold (dB) | start_threshold + stop_threshold (detection=peak) | -40 dB | -80 to -10, step 1 | What counts as a gap. Set per the stem's noise floor and keep it constant across the show for a uniform style |
| Min silence (s) | start_silence + stop_silence | 0.5 s | 0.1 to 5, step 0.1 | Your tightness dial — the air left between phrases after each cut. Lower = punchier, higher = more natural |
House-style presets (set these yourself)
Suggested two-number specs to codify a consistent feel across a show. These are recommendations, not built-in presets — the tool has no preset menu.
| House style | Threshold | Min silence | Feel |
|---|---|---|---|
| Punchy / fast-paced | -42 dB | 0.3 s | Snappy, almost no air between phrases |
| Standard conversational | -40 dB | 0.5 s | Tight but natural — the default zone |
| Relaxed / narrative | -45 dB | 0.8 s | Breathing room preserved; only true dead air cut |
| Documentary / dramatic | -48 dB | 1.2 s | Long intentional pauses survive; minimal tightening |
Tier limits for editors
The tool requires the Pro tier or higher. Size and duration are separate caps. Back-catalogue and long-form work fits Pro-media.
| Tier | Max file size | Max duration | Files per batch |
|---|---|---|---|
| Pro (minimum) | 200 MB | 120 min | 10 |
| Pro-media | 100 GB | Unlimited | 100 |
| Developer | 100 GB | Unlimited | Unlimited |
Cookbook
Editor recipes that treat the silence stripper as the first step of a larger edit. Strip the voice stem, then build the show around the tightened track. Remember it removes only time — your DAW levels/EQ are preserved.
Lock a house style across a season
Standardise the feel of every episode by using the same two numbers for the whole season's voice stems.
House spec: Threshold -40 dB · Min silence 0.5 s
For each episode stem:
ep01-voice.wav -> ep01-voice-tightened.wav
ep02-voice.wav -> ep02-voice-tightened.wav
...
FFmpeg applies (identical every time):
silenceremove=start_periods=1:start_silence=0.5:start_threshold=-40dB:
stop_periods=-1:stop_silence=0.5:stop_threshold=-40dB:detection=peak
-> consistent tightness, no per-episode guessworkPunchy promo cut from a normal episode
Pull a tight social/promo clip out of an existing episode by stripping more aggressively than the show's house style.
Input: ep14-segment.wav (3 min 20 s) Settings: Threshold -42 dB · Min silence 0.3 s Output: ep14-segment-tightened.wav (2 min 48 s) -> snappier than the house style, ideal for a promo
Per-guest stems on a panel show
On a four-person panel, strip each isolated guest stem so cross-talk isn't an issue, then mix the tightened stems.
host.wav -> host-tightened.wav (-40 dB, 0.5 s) guestA.wav -> guestA-tightened.wav (-38 dB, 0.5 s) // noisier room guestB.wav -> guestB-tightened.wav (-40 dB, 0.5 s) guestC.wav -> guestC-tightened.wav (-40 dB, 0.5 s) -> mix the four tightened stems back in the DAW; each cut on its own track avoids cross-talk artefacts
Strip first so loudness reads true
Long dead-air stretches skew an integrated-loudness measurement. Strip before normalising so the LUFS reading reflects actual speech.
Step 1 - silence-stripper: mix.wav -> mix-tightened.wav (-40 dB, 0.5 s) Step 2 - loudness-normalizer (/audio-tools/loudness-normalizer): mix-tightened.wav -> -16 LUFS, true-peak -1 dBTP (measurement no longer dragged down by removed silence)
Documentary style — minimal tightening
A narrative documentary episode where pauses carry meaning. Use a strict threshold and a large cushion so only true dead air goes.
Input: doc-ep.wav (38 min) Settings: Threshold -48 dB · Min silence 1.2 s Output: doc-ep-tightened.wav (37 min 10 s) -> ~2% removed; dramatic pauses fully intact, only between-take dead air trimmed
Edge cases and what actually happens
Strip a music-and-voice mix and the bed gets cut
ExpectedThe stripper can't separate a quiet music bed from dead air. Run it on the voice stem before the bed is added, not on the finished mix. For the finished mix, use ripple edits or audio-trimmer for manual control over the musical sections.
Cross-talk on a mixed multi-guest track stays untouched
By designWhere two voices overlap there's no silence to remove, so cuts only land in single-speaker gaps. On a mixed panel track this makes tightening uneven. Strip per-guest stems instead so each voice is cut consistently before you mix.
Transitions feel rushed after stripping
By designA low Min silence removes most of the inter-phrase air, so the edit can feel breathless. This is your tightness dial overshooting — raise Min silence by 0.1–0.2 s until the cadence matches your house style.
First word of a segment is shaved
ExpectedLeading silence is trimmed once per file. A segment that begins with a soft word can have its onset clipped. Lower the threshold magnitude or raise Min silence, or leave a beat of room tone at the head of the stem before stripping.
Different episodes need different thresholds
ExpectedIf episodes were recorded in different rooms, one threshold won't suit all. Strip per-episode with a threshold matched to each stem's noise floor, but keep Min silence constant to preserve a uniform tightness across the show.
Stem over the Pro 200 MB / 120 min cap
RejectedLong uncompressed stems can exceed Pro's 200 MB size or 120 min duration. Split with audio-splitter, strip the parts, rejoin in the DAW — or use Pro-media (100 GB, unlimited duration) for long-form.
WAV stem stays WAV — no quality loss
PreservedBecause output matches input format, a WAV stem returns as WAV at the same sample rate with no lossy re-encode — exactly what you want for a stem going back on the timeline. (An MP3 in would be re-encoded once; always strip on WAV stems in an editing workflow.)
Heavily-gated raw track yields few cuts
By designIf a noise gate was already applied that fills gaps with low-level noise above the threshold, fewer regions read as silent. Raise the threshold, or strip the pre-gate stem if you have it.
Frequently asked questions
How do I get a consistent tightness across every episode?
Codify two numbers as your house spec — a Threshold (dB) matched to your stems' noise floor and a Min silence (s) as the tightness dial — and use them for every episode. The tool applies the same silenceremove filter every time, so identical settings give identical feel. There's no preset menu, so you set and reuse your own numbers.
Which setting controls how tight the edit feels?
Min silence (s). It's the amount of air left between phrases after each cut. 0.3 s gives a punchy, fast show; 0.5–0.8 s feels conversational; 1.0 s+ is relaxed/narrative. The threshold decides *what* is a gap; min silence decides *how much* survives.
Should I strip the voice stem or the full mix?
Strip the voice stem before music and ads are added. The stripper can't distinguish a quiet music bed from dead air, so running it on a finished mix risks cutting intentional musical sections. Export the voice track to WAV, strip it, then build the show around the tightened track.
Does it preserve my DAW levels and EQ?
Yes. The stripper changes only the time axis — it never re-levels or re-EQs. For a WAV stem the audio is bit-for-bit identical except for the removed silence, so your DAW processing is preserved when you drop it back on the timeline.
Is client audio uploaded anywhere?
No. FFmpeg 8.1 runs in WebAssembly inside your browser tab, so an unreleased client episode is processed locally and never uploaded — suitable for NDA work. The tool requires the Pro tier (200 MB / 120 min per file).
How do I handle a multi-guest panel edit?
Strip each guest's isolated stem separately so cross-talk isn't a problem and each voice is tightened consistently, then mix the tightened stems in your DAW. You can vary the threshold per guest's room while keeping Min silence constant for a uniform feel.
Can I make a punchy promo clip from a normal episode?
Yes — strip the segment more aggressively than your house style, e.g. Min silence 0.3 s with threshold around -42 dB. The result is snappier than the full episode, which suits social and promo cuts. Trim the exact in/out with audio-trimmer.
Why does my first word keep getting clipped?
Leading silence is trimmed once per file (start_periods=1), and a soft opening word can fall below the threshold. Lower the threshold magnitude (-40 → -45), raise Min silence, or leave a beat of room tone at the head of the stem before stripping.
Can I batch a whole season at the same settings?
The Pro tier allows up to 10 files per batch (Pro-media 100, Developer unlimited), all processed locally with identical settings — ideal for applying a house style across episodes recorded in the same room. For mixed rooms, run them individually with per-stem thresholds.
What about loudness — does this set my LUFS target?
No, the stripper only removes time. Strip first so the loudness measurement isn't dragged down by long silences, then run loudness-normalizer for the -16 LUFS podcast target (and a true-peak ceiling).
My long-form episode is over the limit — what now?
Pro caps at 200 MB / 120 min. For long-form, either split with audio-splitter, strip each part, and rejoin in your DAW, or move to Pro-media (100 GB / unlimited duration), which is built for long and back-catalogue work.
Is there a one-pass finishing option for editors?
Yes — podcast-master runs silence handling, noise reduction, leveling, and loudness in a single pass for a quick turnaround. Use the standalone silence stripper when you want exact, repeatable control over the tightness of the edit.
Privacy first
Every JAD Audio tool runs entirely in your browser via FFmpeg (WebAssembly) and RNNoise. Your audio files never leave your device — verified by zero outbound network requests during processing.