How to divide a podcast episode into segments
- Step 1Open the splitter with your episode — Go to the audio splitter and drop the finished episode (one file). It is read locally in your browser — never uploaded. The splitter processes a single episode per run.
- Step 2Choose your splitting strategy — Set Split mode to On silence to auto-detect breaks, or Manual markers to place known cut points yourself. For a conversational episode with clear pauses, silence is fastest; for a scripted show with exact ad-break times, markers are more precise.
- Step 3Tune silence detection (silence mode) — Set Threshold (dB) to your noise floor — a quiet studio sits near -50 dB, a noisy room nearer -35 dB (default -40). Set Min silence (s) to the shortest gap that should count as a break (default 0.6 s; raise to 1.5–2 s to ignore breaths and only cut between segments).
- Step 4Place markers (markers mode) — Click Add marker for each cut point and type the position in seconds (e.g. 750 for 12:30, 1680 for 28:00). Markers are sorted automatically, and any marker at 0 or beyond the episode length is ignored. The number of segments equals markers + 1.
- Step 5Run and review the segments — Start processing. The result panel lists each segment and its size. In silence mode, check that the segment count matches the number of real breaks you expected — too many means lower the threshold sensitivity (more negative dB or longer Min silence).
- Step 6Download each segment — Click Download all to save every segment as its own file, or Download per row. Files are named
<episode>-part-1,<episode>-part-2, etc. Rename them to topic titles in your file manager — the tool names by index, not by chapter title.
Silence-detection settings for common episode types
Starting points for Threshold (dB) and Min silence (s) by recording environment. Tune from here; the goal is to cut between segments without cutting mid-sentence at a breath.
| Episode type | Threshold (dB) | Min silence (s) | Why |
|---|---|---|---|
| Quiet studio, scripted | -50 to -45 | 0.8 | Low noise floor; short clean beats between segments |
| Two hosts, conversational | -40 (default) | 1.2 | Natural pauses are longer; raise Min silence to skip breaths |
| Remote guest, variable mics | -38 to -35 | 1.5 | Higher background noise; less negative threshold avoids false cuts |
| Field / noisy recording | -30 to -25 | 2.0 | Ambient noise raises the floor; only long deliberate gaps should cut |
Silence mode vs Manual markers for podcast segmenting
When to pick each mode. The splitter has no chapter-metadata mode, so these two cover all segment-by-segment workflows.
| Goal | Best mode | How |
|---|---|---|
| Auto-chapter a chatty episode | On silence | Detects pauses; cuts at silence midpoints |
| Pull ad breaks at known times | Manual markers | Add a marker (seconds) at each ad-break start/end |
| One clip per topic, exact starts | Manual markers | Mark each topic boundary in seconds |
| Split a raw take you haven't logged | On silence | Let pauses define rough segments, refine later |
| Use embedded ID3 chapter tags | Not supported | Read chapter times manually, then enter as markers |
Tier limits for episode splitting
The splitter is Pro-tier. A typical 60-minute MP3 episode (~55 MB at 128 kbps) fits the Pro ceiling; very long back-catalogue masters may need Pro-media.
| Tier | Max file size | Max duration | Splitter access |
|---|---|---|---|
| Free | 50 MB | 30 min | Blocked (Pro-only) |
| Pro | 200 MB | 120 min | Yes |
| Pro-media | 100 GB | unlimited | Yes |
| Developer | 100 GB | unlimited | Yes |
Cookbook
Podcast-specific segmenting recipes. Marker positions are in seconds (12:30 = 750 s). Segments are separate files, not a ZIP, and keep the episode's format.
Auto-chapter a conversational episode by silence
A two-host show with clear two-second beats between topics. Let silence detection find them.
Input: ep142.mp3 (58:20) Mode: On silence Threshold (dB): -40 Min silence (s): 1.2 Result: 7 segments at the 7 detected breaks ep142-part-1 ... ep142-part-7 (cuts at the midpoint of each pause)
Pull two ad breaks out by exact markers
You know the host-read ads sit at 12:30-13:15 and 34:00-34:50. Place four markers to isolate them.
Mode: Manual markers Add marker: 750 (12:30 - ad 1 start) Add marker: 795 (13:15 - ad 1 end) Add marker: 2040 (34:00 - ad 2 start) Add marker: 2090 (34:50 - ad 2 end) Segments (markers + 1) = 5: content | AD1 | content | AD2 | content (swap the AD segments per platform later)
One file per topic from a logged episode
You logged topic starts during the edit. Enter each as a marker; segment N is topic N.
Topic log -> seconds: cold open ends 95, topic A ends 1140, topic B ends 2310, topic C ends 3000 Mode: Manual markers Add markers: 95, 1140, 2310, 3000 Result: 5 segments (open + 4 topics)
Too many silence cuts? Raise Min silence
Breaths and pauses-for-emphasis were triggering extra cuts. Lengthen the minimum gap so only real segment breaks count.
First run: Min silence 0.6 -> 19 segments (too many) Second run: Min silence 1.5 -> 8 segments (breaths < 1.5 s no longer count as a break)
Tighten then segment, then publish the teaser
Strip dead air first so segments are tight, segment by markers, then merge a teaser from the best two segments.
1) silence-stripper -> remove long pauses (/audio-tools/silence-stripper) 2) splitter, Manual markers at topic starts 3) audio-merger: join segment 1 + segment 4 -> teaser.mp3 (/audio-tools/audio-merger)
Edge cases and what actually happens
Episode has embedded ID3 chapter tags
Not readThe splitter does not parse ID3 chapter tags or Podlove Simple Chapters. It only cuts by fixed time, by detected silence, or by manual markers. To use existing chapter marks, read the chapter start times from your editor and enter them as manual markers (in seconds).
Silence detection finds too many cuts
Tune settingsBreaths, sighs, and dramatic pauses can each register as silence. Raise Min silence (s) (try 1.5–2.0) so only genuine inter-segment gaps count, and/or make Threshold (dB) more negative so quiet room tone is not treated as silence.
Silence detection finds no cuts
Tune settingsIf background noise keeps the level above the threshold during pauses, no silence is detected and you get a single segment (the whole episode). Make Threshold (dB) less negative (e.g. -30) to match the noisier floor, or shorten Min silence (s).
A segment cuts mid-word
Expected behaviourSilence mode cuts at the midpoint of a detected pause, which is robust for clean gaps but can clip a trailing word when speech runs right up to the pause. For word-perfect boundaries, switch to manual markers and place the cut a fraction of a second into the gap.
Markers entered out of order
Auto-sortedYou can add markers in any order; the splitter sorts them ascending before cutting. Markers at 0 or at/after the episode duration are ignored, so a stray out-of-range value will not create an empty segment.
Two markers very close together
Tiny segment dropped if < 0.05 sMarkers a few seconds apart produce a short segment, which is fine. Only a segment under 0.05 seconds (effectively two identical markers) is dropped. Use this to your advantage for isolating a brief stinger between two close markers.
Expecting segments named by chapter title
Indexed names onlyOutput files are named <episode>-part-1, -part-2, etc. — by index, not by topic or chapter title (the tool has no title input). Rename them in your file manager after download to match your show-notes titles.
Free tier producer tries to split
Tier blockedSplitting requires Pro. On Free it is blocked. A standard 60-minute MP3 episode fits comfortably in Pro's 200 MB / 120 minute limits; only multi-hour masters need Pro-media (unlimited duration).
MP3 segments lose a hair of quality at the cuts
Re-encodedSegments are re-encoded (not stream-copied), so MP3 output incurs one generation of lossy re-encode. For listener-facing podcast audio this is inaudible at typical 96–192 kbps. If you need byte-exact lossless single cuts, use the audio trimmer's lossless option.
Frequently asked questions
Can the splitter use my episode's chapter markers?
Not automatically — it does not read ID3 chapter tags or Podlove chapters. Open your episode's chapter list in your editor, note each chapter's start time in seconds, and enter those as manual markers. The splitter then cuts exactly at those points.
How do I split a podcast at natural pauses?
Set Split mode to On silence. FFmpeg's silencedetect finds the pauses and the tool cuts at the midpoint of each one. Tune Threshold (dB) to your noise floor and Min silence (s) so only real gaps (not breaths) trigger a cut.
What threshold should I use for silence detection?
Start at the -40 dB default. A quiet studio can go to -50/-45; a noisy or remote recording should be less negative (-35/-30) so room tone is not mistaken for silence. Combine with Min silence (s) of 1.2–2.0 to ignore breaths.
How do I isolate an ad break?
Use Manual markers. Add a marker at the ad's start and another at its end (in seconds — 12:30 is 750). The ad becomes its own segment that you can swap or replace per platform. Number of segments = markers + 1.
Do my segments come out as a ZIP?
No — each segment is a separate file. Download all saves them individually in one click; per-row Download links grab a single segment. There is no ZIP option.
Are the segments the same format as my episode?
Yes. The splitter has no format selector, so an MP3 episode produces MP3 segments, a WAV produces WAV segments, and so on. Convert beforehand if you need a different format.
Why did silence mode give me one giant segment?
Because no silence crossed the threshold — usually the room tone sat above it. Make Threshold (dB) less negative (closer to 0, e.g. -30) so the quieter pauses register, or lower Min silence (s).
Why did silence mode over-split into dozens of clips?
Breaths and short pauses each counted as a break. Raise Min silence (s) to 1.5–2.0 so only the longer inter-segment gaps qualify, and/or make the threshold more negative.
Can I name segments by topic?
Not in the tool — segments are named <episode>-part-N by index. Rename them after download. The splitter has no title field, so chapter-title naming has to happen in your file manager.
Is the episode uploaded for processing?
No. FFmpeg 8.1 runs in your browser via WebAssembly; the episode stays on your device. This keeps embargoed or unreleased interviews private.
How long an episode can I split?
On Pro, up to 120 minutes and 200 MB per file. Pro-media and Developer remove the duration cap (100 GB file ceiling), which covers multi-hour specials and live-show recordings.
Can I recombine segments after editing?
Yes. Drop the segments you want into the audio merger in order; it concatenates them end-to-end. This is how you build a teaser or a clean cut from chosen segments.
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.