How to cut a shareable clip from a podcast episode
- Step 1Drop the episode file — MP3, M4A, WAV, FLAC, OGG, Opus, or even the video version of the episode. FFmpeg 8.1 loads locally; the episode is read in the tab, not uploaded.
- Step 2Find the timestamp of the moment — Note the start and end of the line you want in seconds. If your podcast host shows mm:ss, convert (e.g. 12:30 → 750). End (s) is auto-filled with the full duration so you only change what you need.
- Step 3Set Start (s) to the in-point — Type the second the quote begins. FFmpeg seeks there with
-ss <start>. Give yourself a fraction of a second of lead so the first word isn't clipped. - Step 4Set End (s) to the out-point — Type the second the quote ends. JAD keeps
end − startseconds via-t. Add a beat after the last word so it doesn't feel cut off. - Step 5Choose Lossless for MP3 episodes — If the episode is MP3 and you want speed with no quality change, turn Lossless (stream-copy) on — JAD runs
-c copyand the cut snaps to the nearest packet (~30 ms), which you won't hear in speech. Leave it off if you want sample-accurate edges or the format isn't stream-copy eligible. - Step 6Run and grab the clip — The clip downloads with a
-trimmedsuffix in the episode's format. Caption it, build an audiogram, or post it — and because nothing was uploaded, the rest of the episode stays private.
Clip-cutting settings for podcast work
The trimmer exposes exactly these three controls. Recommended settings for typical podcast clipping are noted.
| Control | What it controls | Recommended for clips |
|---|---|---|
| Start (s) | In-point in seconds (-ss) | Set ~0.3 s before the first word so the attack isn't clipped |
| End (s) | Out-point in seconds (kept length = end − start) | Set ~0.5 s after the last word for a natural tail |
| Lossless (stream-copy) | -c copy instead of re-encode, same-format only | On for MP3 episodes (instant, no quality loss); off for sample-accurate edges |
Format behaviour for common episode types
Output format always matches the episode (no format dropdown on the trimmer). Stream-copy eligibility decides whether Lossless can apply.
| Episode format | Re-encode encoder | Lossless (stream-copy) works? | Clip output |
|---|---|---|---|
| MP3 | libmp3lame | Yes | MP3 |
| M4A / AAC | aac | Yes | M4A |
| WAV (raw recording) | pcm_s16le | Yes | WAV |
| FLAC (archive master) | flac | Yes | FLAC |
| OGG / Opus | libvorbis / libopus | Yes | OGG / Opus |
| Video version | encoder for the audio track | Only if audio container is eligible | Audio file |
How long an episode you can clip per tier
Per-file duration is capped separately from size. Full-length interview shows usually need Pro or higher.
| Tier | Max size | Max episode length | Files at once |
|---|---|---|---|
| Free | 50 MB | 30 min | 1 |
| Pro | 200 MB | 120 min | 10 |
| Pro-media | 100 GB | Unlimited | 100 |
| Developer | 100 GB | Unlimited | Unlimited |
Cookbook
Real clipping scenarios from podcast production, with the exact Start/End and Lossless choices.
Pull a 20-second quote for social
A guest's best line lands at 18:42–19:02 in a 55-minute MP3. Lossless on for an instant, lossless clip.
Input: ep142.mp3 (3300.0 s) Start (s): 1122 (18:42) End (s): 1142 (19:02) Lossless: on (MP3, stream-copy eligible) FFmpeg: -ss 1122 -i ep142.mp3 -t 20 -c copy -avoid_negative_ts make_zero out.mp3 Output: ep142-trimmed.mp3 (~20 s, no re-encode)
Cold-open hook with sample-accurate edges
You want the first 12 seconds as a teaser and the edge must be exact (no ~30 ms packet drift), so Lossless off.
Input: ep143.mp3 (2880.0 s) Start (s): 0 End (s): 12.0 Lossless: off (re-encode for exact boundary) FFmpeg: -ss 0 -i ep143.mp3 -t 12 -c:a libmp3lame out.mp3 Output: ep143-trimmed.mp3 (exactly ~12 s)
Clip from a WAV master before mastering
Editing a raw WAV recording and want a quote for the promo before the episode is mastered. Same format, Lossless on = no PCM loss.
Input: interview-raw.wav (4100.0 s) Start (s): 305.5 End (s): 333.0 Lossless: on (WAV stream-copy) FFmpeg: -ss 305.5 -i interview-raw.wav -t 27.5 -c copy -avoid_negative_ts make_zero out.wav Output: interview-raw-trimmed.wav (byte-identical PCM)
Clip the audio from a video episode
Your show also publishes on YouTube; pull an audio soundbite straight from the video file.
Input: ep144.mp4 (3600.0 s, AAC audio) Start (s): 600 End (s): 615 Lossless: off FFmpeg: -ss 600 -i ep144.mp4 -t 15 -c:a aac out.m4a Output: ep144-trimmed.m4a (15 s audio clip)
Three clips from one episode (run three times)
The trimmer keeps one range per run, so make three passes with different in/out points. Each download is independent.
Pass 1: Start 130 End 152 → ep-trimmed.mp3 (hook) Pass 2: Start 1810 End 1839 → ep-trimmed.mp3 (quote) Pass 3: Start 3140 End 3168 → ep-trimmed.mp3 (CTA) Rename each download as you go (the suffix is always -trimmed). For many regular cuts, [audio-splitter](/audio-tools/audio-splitter) takes multiple ranges in one pass.
Edge cases and what actually happens
First word of the clip gets cut off
AvoidableSet Start (s) ~0.3 s earlier than the exact onset so the word's attack is included. With Lossless on, also remember the cut snaps to the nearest packet (~30 ms), which can shave the very start — give yourself a little lead.
Lossless clip starts a frame late
By designStream-copy cuts on packet boundaries, so a copy clip can drift up to ~30 ms. In speech this is inaudible; if you need an exact edge (e.g. cutting mid-word for a meme), turn Lossless off for a re-encoded, sample-accurate cut.
Full episode is over 30 minutes on Free
Tier limitA typical interview episode exceeds the 30-minute Free duration cap, so the episode won't load. Upgrade to Pro (120 min) or Pro-media (unlimited), or clip from a shorter exported segment.
Episode file is larger than 50 MB
Tier limitFree caps at 50 MB. A long, high-bitrate MP3 episode can exceed that. Pro (200 MB) handles most full shows; Pro-media (100 GB) handles WAV masters.
Stream-copy returns nothing and JAD re-encodes
RecoveredSome containers (partial MP4, certain Opus-in-WebM) can't be copied after a seek and return a 0-byte file or throw. JAD automatically falls back to a re-encode, so you still get a valid clip — just check the metrics line to see which engine ran.
Clip clicks at the cut point
ExpectedA hard trim mid-waveform can click. For a polished social clip, run the trimmed file through fade-in-out to add a short fade at both edges.
Clip is quieter than the rest of the feed
Wrong toolTrimming never changes level. If a clip needs to match podcast loudness (e.g. -16 LUFS), run it through loudness-normalizer after cutting — the trimmer only cuts.
You wanted to remove an ad from the middle
Wrong toolThe trimmer keeps one contiguous range. To delete a mid-roll ad and rejoin the show, see remove-intro-outro-from-audio for the split-and-merge workflow, or use audio-splitter + audio-merger.
Output format isn't the one your CMS wants
By designThe clip stays in the episode's format. If your audiogram tool needs WAV but the episode is MP3, trim first, then convert with mp3-to-wav.
Frequently asked questions
Can I clip an episode that isn't published yet?
Yes — that's a key reason to use this tool. FFmpeg 8.1 runs in your browser, so the unreleased master never uploads. You can prepare promo clips before the episode goes live without exposing the file.
Should I turn Lossless on for podcast clips?
For MP3 episodes, usually yes: it's instant and there's no quality loss, and the ~30 ms packet snap is inaudible in speech. Turn it off only when you need a sample-exact edge or the source isn't a stream-copy-eligible format.
How do I enter a timestamp like 18:42?
Convert to seconds: 18:42 = (18 × 60) + 42 = 1122. The Start (s) and End (s) fields are in seconds with a 0.1-second step.
Will the clip be in the same format as my episode?
Yes. The trimmer has no format selector, so an MP3 episode yields an MP3 clip. Need a different format? Trim, then use a converter like mp3-to-wav.
Can I make several clips from one episode?
Yes — run the trimmer once per clip with different Start/End points, renaming each download. For many evenly spaced cuts, audio-splitter accepts multiple ranges in a single pass.
Does it add a watermark or intro sting?
No. The output is purely your selected range with no added audio. Anything like an intro sting you'd add separately.
My episode is 90 minutes — can I clip it?
On Free, no: the duration cap is 30 minutes. Pro raises it to 120 minutes, and Pro-media/Developer remove the cap. Alternatively, export a shorter chapter from your host and clip that.
Why does my lossless clip sometimes start slightly early or late?
Stream-copy can only cut at packet boundaries, so it snaps to the nearest one (~30 ms). It's the trade-off for an instant, no-re-encode cut. Use Lossless off for exact boundaries.
How do I make the clip loud enough for social?
Trim first, then normalize loudness with loudness-normalizer (EBU R128, two-pass) to a target like -16 LUFS. The trimmer itself doesn't touch level.
Can I remove a sponsor read from the middle and keep the rest?
Not in one trim — the trimmer keeps one continuous range. Split around the ad with audio-splitter and rejoin the parts with audio-merger; the remove intro/outro guide walks the same pattern.
Is there a file-size limit for episodes?
Free: 50 MB. Pro: 200 MB. Pro-media and Developer: 100 GB. The size cap is checked before processing, alongside the per-file duration cap.
What if the clip has a click at the end?
Hard cuts can click when they don't land on a zero crossing. Apply a short fade with fade-in-out after trimming to smooth both edges.
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.