A podcast clip should make clear whose words the viewer is hearing and whether they are asking, answering or quoting someone else. Keep essential speaker handoffs, use a frame that does not imply the wrong person is speaking and verify names manually. CastCuts captions are not a promise of speaker diarization, automatic name labels or active-speaker tracking.
Identify the roles in the exchange before choosing a crop
Listen for a question, response, interruption and quoted example. Record the speaker for each meaningful turn in your own notes. A host may repeat a guest's claim to challenge it, and a guest may imitate a customer without endorsing that customer's view. If the excerpt removes those signals, the viewer can attach the statement to the wrong person. Preserve a short introduction or choose a clearer passage when necessary.
Match visible composition to the conversation
For a two-person recording, inspect whether a fixed crop shows only a listener while the other person answers. Full-frame composition may better preserve the exchange even though each face appears smaller. A listener's nod, pause or smile can look like agreement when separated from its context, so review the complete audiovisual impression. CastCuts provides fixed crop and full-frame options for this workflow. Do not describe them as automatic active-speaker switching or infer that any separate limited framing preview identifies voices and names.
Verify identities and distinguish credit from captions
Check the guest's preferred name and role against information you have confirmed with them or the episode's approved materials. Put necessary identification in the post description or concise contextual text supported by the available editor. Timed speech captions should reflect the recording rather than silently presenting invented labels as spoken words. CastCuts does not provide a dedicated automatic speaker-name labeling workflow here. When the guest credits a colleague or quotes a source, keep that attribution in the excerpt or make the relationship explicit in the accompanying description.
Perform a who-said-what review on the final clip
Watch the downloaded MP4 without the full episode open. Write one sentence identifying the person making the main point and the person, if any, being quoted. Compare that sentence with the original exchange. Check the frame during interruptions and the captions during short responses such as 'right' or 'no'. If two voices overlap, use the original recording to establish the words rather than accepting a plausible transcript.
A fictional quote that could be assigned to the wrong person
Illustrative example. The people, recordings and results described here are fictional.
Guest Nia says, 'A client asked, Why do we need a rehearsal? Then the first test failed.' A rough cut beginning at 'Why do we need a rehearsal?' makes the question sound like Nia's own objection. The editor includes 'A client asked', uses full-frame composition for the host's follow-up and identifies Nia accurately in the post. They keep the quote's role clear without claiming an automatic speaker label or changing the question into Nia's recommendation.
Attribution includes the relationship between speakers and quoted people, not just a name beside a face.
Check before you finish
- The questioner, main speaker and any quoted person are identified correctly.
- The crop does not imply that a visible listener is delivering the answer.
- Names, roles and credit are manually verified.
- Overlapping speech and short interruptions are checked against the source.
Practical questions
Does CastCuts identify podcast speakers automatically?
Do not assume speaker diarization or automatic name labels. Review who says each line yourself and use the available composition and accompanying post text to keep the exchange understandable.
Should I crop to the guest throughout an interview clip?
Only when that fixed view represents the passage clearly. An exchange with important host reactions or questions may be easier to understand in full-frame composition.
Can I remove the words introducing a quotation?
Keep them when removing them would assign the quote to the wrong person or change endorsement into disagreement. A shorter excerpt is not useful when it misstates who holds the view.
Use this workflow in CastCuts
Keep interview questions and guest answers clear
CastCuts selects continuous 1–90 second excerpts and exports 9:16, 720 × 1280 H.264/AAC MP4 files. Automatic transcription and transcript-based suggestions support English and require review. Free exports include CastCuts branding; Pro removes it from new renders for $9 USD/month, with the same processing and storage limits. Episode media is retained for 24 hours. Keep your originals and download finished clips before expiry. Publishing is manual.