Cut words, not waveforms: transcript-first editing

When the content is mostly speech, the edit decision is usually about meaning. Keep the answer. Remove the false start. Cut the part where the guest repeats the point. A waveform shows volume, but it does not tell you which sentence is the useful one.

Podcast Studio’s Editor starts with the transcript instead. Edit the words and the audio follows. For interviews, panels, and commentary, that puts the decision where the producer is already making it: in the language.

How a word-level cut works

Open an episode from Episodes with the scissors icon. If a transcript is not ready, choose Transcribe Audio and wait for the word-level map. Select a phrase, press Delete or Backspace, and the words are marked for removal. They appear struck through rather than disappearing, so the edit stays visible.

Hover a cut word to Restore it, or use Restore All when a pass went too far. Undo and Redo are available during the session. The source recording stays recoverable while you work; the cut list becomes the version listeners hear when you publish or export.

Why text is faster for talk

On a waveform, “um,” a breath, and a key point can look like similar bumps. On a transcript, you can read the digression and select it directly. If you remember that a guest mentioned regulatory capture near the second coffee story, you can scan for the phrase instead of searching by minute and hoping the quote is nearby.

Word-level cuts also match how people hear conversation. Listeners do not think “remove 400 milliseconds at 12:04.” They hear a false start, a repeated clause, or a point that arrives after too much setup. Cutting the words makes the edit easier to explain to a collaborator and easier to review yourself.

Speaker labels make the document usable

Transcript editing gets harder when both speakers look like one wall of text. Click a speaker name to Rename Speaker or Assign Color. If two voices were grouped incorrectly, Shift+click a word to split a new speaker from that point.

That is not cosmetic. Clean labels let you find the guest’s answer, check who is being cut, and hand off a specific revision without sending someone on a full re-listen.

Where a DAW still makes sense

Transcript-first editing is a strong fit for speech, not a replacement for every audio workflow. Music-heavy narrative, intricate sound design, and stem-level mixing still benefit from a dedicated DAW. Pro Audio exists for that route: the add-on provides DAW masters and stems for Hobby and Creator, and is included on Scale and Enterprise.

The transcript is also not a script to rewrite fictionally. Cutting a word removes that part of the recording. It does not invent a cleaner sentence. If transcription is wrong because of overlap, noise, or a soft microphone, listen before deleting a whole passage.

A useful first pass

1. Remove pre-roll logistics and the “can you hear me?” material.
2. Cut structural repetition, keeping the clearest version of the answer.
3. Leave the pauses that show thought or let a joke land.
4. Only then decide whether filler-word or silence tools are needed.

This order matters. A Magic cleanup cannot rescue a two-minute tangent, and a perfect waveform splice cannot tell you whether the tangent was the point.

If your episode is a conversation, edit the document of what people said. Let the waveform support the decision instead of making it for you.

Transcript-first editing will not turn a weak interview into a strong one. It can turn a good recording into a finished episode on a weeknight, which is often the production problem that matters.