You have finished a one-hour podcast episode. You know there are five moments in there that would make great TikTok clips — a strong opinion, a funny aside, a quotable line. The problem is finding them. Scrubbing a timeline at 1x speed takes longer than the episode itself.
The fix is on-device transcription. Your iPhone can turn the episode into a searchable transcript in minutes, and from there you can pick the moments worth clipping without watching the whole thing.
Why on-device matters for podcasts
Most podcasters do not upload raw audio to a cloud service, full stop. Conversations with sources, unreleased material, sensitive topics — these all stay local. If your highlight tool uploads the audio, it is the wrong tool.
ReelClip runs the entire workflow on-device via Apple's SFSpeechRecognizer. Your audio is copied into the app's private sandbox, transcribed locally, and never sent to any server. The transcript itself is on-device.
The workflow
1. Import the episode audio
ReelClip can import any video or audio file from the Files app or Photos. If you have the audio as a separate file (mp3, wav, m4a), import that — it is faster and skips the video decoding step. If you only have video, that works too.
2. Choose Transcript mode
ReelClip's Transcript mode does three things:
- Transcribes the audio on-device.
- Finds pauses longer than 0.75 seconds and flags them as cuts.
- Identifies filler words ("um", "uh", "you know") and lets you remove them individually.
By the end of the import step you have a transcript with timecodes, a list of pause markers, and a list of filler-word markers. This is the part that used to take a human assistant hours.
3. Read the transcript, not the audio
This is the unlock. Open the transcript side panel and skim it. Strong opinions, jokes, and quotable lines are obvious in text form — you can read a one-hour transcript in 15 minutes and pick the moments you would have scrubbed for.
Tap any line in the transcript to jump the playhead to that moment. You are not scrubbing a waveform; you are reading a document.
4. Mark the highlight ranges
When you find a moment worth clipping, mark a range around it. ReelClip gives you a one-tap "make this a clip" button plus fine-grained drag handles for the exact start and end. A good highlight clip is usually 30–90 seconds — long enough to give context, short enough to keep attention.
5. Stack up 4–8 candidates before exporting
Plan more than you will use. From a one-hour podcast you can usually find 6–10 highlight candidates. Pick the strongest 4–5 for the first round — the algorithm rewards consistency over a single viral hit, so a steady drip of clips outperforms one big post.
6. Export with captions
Each clip exports with the matching transcript slice attached. From there you can:
- Burn in captions using one of four styles (Pop, Karaoke, Clean, One word) — useful when the clip will live on Reels or TikTok, where most viewers watch on mute.
- Export the SRT or VTT file separately and bring it into CapCut or your editor of choice.
- Export the plain-text transcript for show notes, a Substack post, or an X thread.
7. Post to all three short-form platforms
One clip, three platforms. The caption files export the same on each platform — you do not need to re-cut for vertical versus square if your source audio was landscape video. The same clip works on Reels (vertical 9:16), TikTok (vertical 9:16), and Shorts (vertical 9:16).
What this looks like in practice
A one-hour podcast episode, end-to-end:
- 10–15 minutes to import and transcribe (depends on device; Apple Silicon does this fast).
- 15–20 minutes to read the transcript and mark highlight ranges.
- 5 minutes to review each clip and trim the start/end.
- 5 minutes to export and post across platforms.
Total: roughly 40 minutes from episode to five posted clips, compared to several hours of timeline scrubbing. And because the audio never left the device, you can do this on a sensitive recording without sending it to a third party.
What to do next
See how to cut a long video for TikTok for the broader short-form workflow, or remove silence from video on iPhone for a faster path when you do not need captions.