Six interviews, one paper edit
A corporate film has six interviews, each around forty minutes, with two cameras. The director wants a paper edit before the editor starts. The assistant editor sends the audio to a transcription service, gets back six Word documents with no timecode, and the director marks up printed pages with a highlighter. The editor then scrubs through footage to find each highlighted line, in both camera angles.
The client wants to see the paper edit too. It goes as a PDF of scanned pages. The client's notes refer to page numbers that do not match anything in the edit.
Why selects take so long
- Transcripts arrive without timecode, or with timecode that does not match the clips.
- Paper edits are done on printed pages or in Word, far from the footage.
- Finding a highlighted line in the footage means scrubbing.
- Multi-camera interviews double the scrubbing.
- Client input on selects comes back in a format the editor cannot use directly.
Editing software now includes transcription, such as the text-based editing in Premiere Pro and DaVinci Resolve, and for many companies that is enough. The gap tends to be the director and client working on selects outside the edit suite.
What slow selects cost
Editor days spent scrubbing are expensive and dull. Directors spend evenings with highlighters. Clients who cannot easily take part in selecting are more likely to reject the first cut. And good lines are missed because nobody could search for a word or phrase across hours of interviews. On films with several interviewees, the strongest story often comes from connecting lines said by different people, which is exactly what is hardest to spot on paper.
How we build transcription and selects
- After offload, interview clips are sent to a speech-to-text service that returns words with timings.
- Transcripts are linked to the clips by timecode, so any word can be played from that point.
- Multi-camera clips are synced by timecode or audio, so one transcript covers every angle.
- The director and client can open a browser view of the transcripts, search across all interviews, play any line, and mark selects with a colour and a note.
- Selects are ordered into a paper edit by dragging, still linked to timecode.
- The paper edit is exported to the edit as markers or a rough sequence, in a format your editing software can import.
| Step | Before | After |
|---|---|---|
| Transcription | Word files without timecode | Words linked to timecode |
| Marking selects | Highlighter on paper | Marked in the browser, playable |
| Client input | Scanned pages | Selects and notes in the same view |
| Into the edit | Scrubbing | Markers or a rough sequence |
Transcripts are good but not perfect, especially with accents, jargon and crosstalk. On-screen text or subtitles taken from them are always checked by a person.
The next interview-led film
Transcripts are ready the day after the shoot. The director searches all six interviews for the word sustainability, finds eight lines, plays each, and marks four. The client does the same from their office. The director orders the selects into a paper edit and exports it. The editor imports the sequence and starts cutting the story rather than finding the lines.
Later, when the client asks for a social cutdown about hiring, the producer searches the same transcripts for recruitment and finds the lines in minutes.
The transcripts stay with the project in the footage catalogue, so an interview filmed for one film can be searched when a client wants something new from the same material months later.
Is this your selects process?
- Paper edits are done on printed transcripts.
- Transcripts do not include timecode.
- Editors scrub to find highlighted lines.
- Clients give selects feedback as page numbers.
- Nobody can search across all interviews for a phrase.