Interview Transcript Editor

What This Skill Does

Edits raw interview transcripts for publication — from minimal cleanup through full restructuring — across six distinct editing modes, each with its own standards for what to cut, what to keep, and how far to reshape the subject's words. Covers the full range of editorial work that happens between a recorded conversation and a published piece.

When To Use This Skill

• You have a raw transcript and need it publication-ready in any format (print, web, broadcast, podcast)
• You need to tighten a 60-minute conversation into a 1,500-word Q&A without losing the subject's voice
• You are editing someone else's interview and need a consistent, defensible editorial standard
• You need to combine material from two or more interviews with the same subject into one coherent piece
• You want to extract the strongest quotes for sidebars, social media, or promotional copy
• You need a fact-check pass that flags every verifiable claim in the transcript before publication
• You are producing a broadcast segment and need the transcript reshaped for spoken delivery
• Your subject has approval rights and you need transparent editorial notes explaining every change

What You Need To Provide

For any editing mode

Required: The raw transcript (or the section you want edited); the editing mode (verbatim cleanup, print Q&A, broadcast-ready, composite, fact-check pass, or pull-quote extraction); the publication format and outlet type (magazine, newspaper, radio, podcast, web, social) Optional: Target word count or airtime; the publication's house style; which sections are must-keep material; the subject's identity type (see Subject Type table below); whether the subject has approval rights; whether any portion was off-the-record; the interviewer's notes about tone, context, or intent behind specific questions

For composite edits specifically

Required: All transcripts to be combined (minimum two); a note on which interview was conducted first (chronological context matters); whether the subject knew their answers from different sessions would be combined Optional: Which topics overlap between interviews; any contradictions between sessions that need flagging

Editing Modes

Mode 1: Verbatim Cleanup

Purpose: Produce a readable transcript that stays as close to the original spoken words as possible. Used for legal proceedings, academic research, archival records, or when the subject's exact phrasing is the point.

What it does:

• Fixes transcription errors (misheard words, garbled passages, incorrect proper nouns)
• Adds punctuation and paragraph breaks for readability
• Marks inaudible passages as [inaudible] rather than guessing
• Preserves all filler words, false starts, repetition, and incomplete thoughts
• Adds speaker labels and timestamps if provided
• Notes any discrepancies between audio cues (described in brackets) and the transcription

What it does not do:

• Does not remove filler, repetition, or false starts
• Does not reorder anything
• Does not condense or paraphrase
• Does not correct the subject's grammar, factual errors, or word choices

Output: Full transcript with speaker labels, paragraph breaks, and [inaudible]/[unclear] markers. Footnotes for transcription corrections. No editorial judgment applied.

---

Mode 2: Print Q&A

Purpose: Produce a publication-ready Q&A for a magazine, newspaper, or web feature. The standard interview format — the one readers see.

What it does:

1. Reads the full transcript to identify the three to five most substantive answers — the moments where the subject says something irreducible — and treats these as structural anchors
2. Removes filler words (um, uh, like, you know, I mean, sort of, kind of), verbal repetition, and false starts
3. Condenses long answers to their essential content without changing meaning
4. Resolves pronoun ambiguity and unclear references by inserting bracketed clarifications [like this] only when necessary
5. Tightens interviewer questions to their core — removes preamble, flattery, and multi-part tangles
6. Proposes a running order that puts the strongest exchange first (unless chronological order is specified)
7. Sequences subsequent exchanges so each adds a distinct dimension — no two consecutive answers covering the same ground

What it does not do:

• Does not change the subject's vocabulary, register, or characteristic expressions
• Does not add words the subject did not say (except bracketed clarifications)
• Does not "improve" answers — only removes and compresses
• Does not merge separate answers into one (that is composite editing, Mode 4)

Output: Edited transcript in Q&A format. Interviewer questions marked Q:. Subject answers marked A:. Editorial notes section at the end flags judgment calls, full cuts, and ambiguous passages. Target word count honored (plus or minus 10%).

---

Mode 3: Broadcast-Ready Edit

Purpose: Reshape a transcript for spoken delivery — radio, podcast, or television voiceover. The edited version will be read aloud or used as the basis for an audio/video segment.

What it does:

1. Edits for the ear, not the eye — shorter sentences, simpler clause structure, no parenthetical asides longer than five words
2. Replaces written-language constructions with spoken ones (e.g., "the aforementioned" becomes "what we talked about earlier")
3. Marks natural pause points for the presenter or narrator
4. Flags any passage that requires the listener to hold more than two ideas in memory simultaneously — rewrites or splits it
5. Converts numbers to spoken form (e.g., "two and a half million" not "2,500,000")
6. Identifies the strongest 10-15 second sound bites and marks them as potential clip points
7. Notes where ambient sound, music, or a second voice could carry the narrative instead of words

What it does not do:

• Does not write new narration or links between segments (that is the presenter's job)
• Does not add dramatic framing the subject did not provide
• Does not change the subject's meaning to make it more "broadcastable"

Output: Edited transcript with broadcast markup: [CLIP] for recommended sound bites, [PAUSE] for breathing room, [BRIDGE] where a presenter link is needed, [AMBIENT] where sound could replace narration. Estimated read time noted. Editorial notes section at the end.

---

Mode 4: Composite Edit

Purpose: Combine material from two or more interviews with the same subject into a single, coherent Q&A or narrative. Common in long-form journalism where a subject is interviewed multiple times over weeks or months.

What it does:

1. Maps the thematic territory of each interview — which topics each one covers, where they overlap, where they contradict
2. Identifies the strongest version of each answer when the subject addressed the same topic in multiple sessions
3. Builds a unified Q&A or narrative sequence, drawing the best material from each interview
4. Flags any contradictions between sessions (the subject said X in interview one but Y in interview three) and notes them for the editor rather than silently resolving them
5. Ensures the combined piece reads as a single coherent conversation, not a patchwork
6. Notes the source interview for each answer (e.g., "From interview 2, March 14") in editorial notes so the editor can trace every passage back to its source

What it does not do:

• Does not combine answers from different subjects (that is a reported piece, not an edited transcript)
• Does not silently resolve contradictions — always flags them
• Does not create the illusion that the subject said things in a single sitting that were actually said weeks apart, unless the editor explicitly requests this and the publication's ethics policy permits it

Output: Unified Q&A or narrative transcript with source annotations in editorial notes. Contradiction log if applicable. A note on whether the subject was informed that material from multiple sessions would be combined.

---

Mode 5: Fact-Check Pass

Purpose: Review a transcript specifically to identify every verifiable claim — statistics, dates, proper nouns, historical references, scientific claims, financial figures — and flag them for verification before publication.

What it does:

1. Reads every answer and marks each verifiable claim with a [CHECK] tag
2. Categorizes claims by type: statistical, historical, biographical, scientific, financial, legal, geographic
3. Notes the specific assertion to verify (e.g., [CHECK: Subject claims the company was founded in 2018 — verify incorporation date])
4. Flags claims that sound precise but may be rounded, estimated, or anecdotal ("about 40%" vs. "exactly 40%")
5. Identifies claims where the subject is the only source — marks these as [SINGLE SOURCE] since they cannot be independently verified through public records
6. Flags potential defamation risks — any statement that attributes wrongdoing, incompetence, or illegal activity to a named or identifiable person or organization

What it does not do:

• Does not verify the claims itself (the journalist or fact-checker does that)
• Does not edit the transcript — leaves the text unchanged and adds tags
• Does not make legal judgments about defamation — only flags statements that a media lawyer should review

Output: Original transcript with inline [CHECK], [SINGLE SOURCE], and [DEFAMATION RISK] tags. A numbered fact-check list at the end with each claim, its category, and a suggested verification source. Total claim count and breakdown by category.

---

Mode 6: Pull-Quote Extraction

Purpose: Select and edit the strongest quotes from a transcript for use in sidebars, social media, headlines, pull-quote boxes, or promotional material.

What it does:

1. Identifies the 5-10 strongest quotes based on: specificity (concrete detail beats generality), surprise (says something the reader would not expect), emotional weight, and brevity (under 40 words preferred, never over 60)
2. Edits each quote for standalone clarity — a reader encountering the quote without the full interview must understand it
3. Adds brief context lines (one sentence) explaining what prompted each quote
4. Rates each quote for different uses: print sidebar, social media post, headline/subhead, promotional copy
5. Flags any quote that requires the full interview context to avoid misrepresentation

What it does not do:

• Does not create quotes the subject did not say
• Does not combine fragments from different answers into one quote (that is fabrication)
• Does not select quotes that misrepresent the subject's position when read in isolation

Output: Numbered list of pull quotes, each with: the edited quote, a one-line context note, recommended uses (sidebar / social / headline / promo), and a flag if the quote risks misrepresentation out of context.

---

Subject Type and Editing Latitude

How aggressively you can edit depends on who the subject is and what they are talking about. More power means less latitude.

Subject Type	Editing Latitude	Rationale
Elected official or executive	Low — preserve exact phrasing, even when awkward	Their words are accountable. A cleaned-up quote can shield them from scrutiny they deserve.
Expert or academic	Medium — clean up delivery but preserve technical precision	Their expertise is the reason for the interview. Simplifying jargon is acceptable; changing their scientific claims is not.
Celebrity or public figure	Medium — clean up delivery, preserve personality	Voice and personality are the point. Over-editing makes them sound generic.
Ordinary person affected by events	High — clean up freely, preserve meaning and dignity	They did not choose public life. Rough delivery can make them sound less credible than they are. Clean editing serves them.
Whistleblower or confidential source	Minimal — preserve exact phrasing, add nothing	Legal exposure is real. Every word choice may matter in court. Consult a media lawyer before publishing.
Minor (under 18)	High — clean up freely, protect identity if needed	Extra care required. Remove any detail that could identify them if anonymity was promised. Parental consent issues may apply.

---

Editing Ethics and Standards

The Core Rule

An edited transcript must be a shorter, clearer version of what the subject actually said. It must never become a version of what the editor wishes they had said.

What You Can Do

• Remove filler words, false starts, and verbal repetition
• Condense a long, circling answer to its essential point
• Reorder exchanges for narrative flow (with disclosure if your publication requires it)
• Fix obvious mis-speakings where the intended meaning is clear and the subject would agree (e.g., they said "2019" but clearly meant "2020" based on context) — always note this in editorial notes
• Insert bracketed clarifications for pronoun ambiguity: "He [the defense attorney] said..."
• Tighten interviewer questions (the interviewer's words are not sacrosanct the way the subject's are)

What You Cannot Do

• Add words the subject did not say, even if they "would have said it that way"
• Merge separate answers into one answer without disclosure (composite editing requires transparency)
• Change the subject's vocabulary to sound more articulate, more folksy, more dramatic, or more moderate
• Remove qualifiers that weaken a quote but change its meaning ("I think this might be a problem" cannot become "This is a problem")
• Rearrange clauses within a sentence to change emphasis without disclosure
• Present paraphrased material in quotation marks

The Bracket Test

If you need to add a word to make a quote work, put it in brackets. If you need more than three bracketed words in a single quote, the quote does not work — paraphrase instead and attribute with "she said" or "he explained."

---

Style Guide: Handling Common Transcript Problems

Filler Words

Rule: Remove in all modes except Verbatim Cleanup. Includes: um, uh, er, like (when not a comparison), you know, I mean, sort of, kind of, basically, literally (when not literal), right?, so (sentence-initial filler), actually (when adding nothing), honestly (when not contrasting with dishonesty) Exception: Keep a filler word when it signals hesitation that is editorially meaningful — when the subject paused before a difficult admission, "um" tells the reader something. Note in editorial notes why you kept it.

False Starts

Rule: Remove in all modes except Verbatim Cleanup. Example: "The problem is — well, what I would say is — the real issue here is funding" becomes "The real issue here is funding." Exception: Keep a false start when the abandoned thought reveals something the completed thought conceals. If someone starts to say "We knew about the contamination" and then corrects to "We learned about the situation later," both versions matter. Flag in editorial notes.

Repetition

Rule: Keep the strongest version of a repeated point. Cut the others. How to choose: The version with the most specific detail wins. If equal in specificity, the version with the most natural phrasing wins. If the subject repeated a point three times, they consider it central — make sure the surviving version is prominently placed.

Off-the-Record Material

Rule: Remove entirely. Do not paraphrase. Do not hint at it. Do not leave a visible gap that implies something was removed. Process: If the subject said "this is off the record" and the interviewer agreed, every word from that point until the subject or interviewer explicitly returned to on-the-record status is off the record. There is no partial off-the-record. The agreement governs. Dispute: If there is ambiguity about whether off-the-record was properly invoked (e.g., the interviewer did not agree, or the subject said it after the fact), do not resolve this in editing. Flag it for the editor-in-chief or media lawyer. The edited transcript should note: [OFF-THE-RECORD PASSAGE REMOVED — DISPUTED, SEE EDITOR'S NOTE].

Profanity

Rule: Follow the publication's house style. The three standard approaches:

Approach	Example	Typical Use
Print in full	"That's bullshit and everyone knows it."	Long-form magazines, literary journalism, most digital publications
First letter and dash	"That's b------- and everyone knows it."	Legacy newspapers, broadcast transcripts
Remove or paraphrase	He dismissed the claim forcefully.	Family publications, corporate media, broadcast standards

If no house style is specified: Print profanity in full. Censoring an adult's speech without editorial reason is a form of editing their voice. Note in editorial notes that profanity was preserved and flag it for the editor.

Exception: Slurs (racial, ethnic, sexual orientation, gender identity) follow different rules. Most publications do not print slurs in full even in direct quotes. Consult the publication's policy. When no policy exists, describe the slur rather than printing it: "He used a racial slur directed at [group]."

Dialect, Accent, and Non-Standard Grammar

Rule: Do not attempt to render dialect, accent, or non-standard English phonetically. It reads as mockery on the page regardless of intent.

Specifically:

• Do not write "gonna," "wanna," "gotta" — use "going to," "want to," "got to" (everyone says these contractions; spelling them out for some speakers and not others is discriminatory)
• Do not drop g's ("runnin'," "talkin'") unless the subject specifically requested this representation of their speech
• Do not correct grammar that is characteristic of the subject's dialect (e.g., "We was" in African American Vernacular English is grammatical in that dialect — correcting it to "We were" erases the subject's voice)
• Do not exoticize speech patterns. If a non-native English speaker uses an unusual construction, clean it up to standard English unless the construction is the point of the quote

The test: Would you apply the same spelling and grammar conventions to a university professor? If not, you are applying a double standard.

Incomplete Thoughts and Trailing Off

Rule: If the subject trailed off and the meaning is clear from context, complete the thought in brackets or end with an em dash.

• Clear meaning: "And after that we just—" [with context making it obvious they stopped going to the clinic] becomes "And after that we just stopped going."
• Unclear meaning: "And after that we just—" [no clear context] becomes "And after that we just—" (preserve the trailing off, note in editorial notes that the meaning was unclear)

Crosstalk and Interruptions

Rule: In most editing modes, resolve crosstalk by giving each speaker their complete thought in sequence. Note in editorial notes where crosstalk occurred. Exception in Verbatim Cleanup: Mark overlapping speech with [CROSSTALK] and preserve both speakers' fragments.

Laughter, Sighs, and Non-Verbal Sounds

Rule: Include [laughs], [sighs], [long pause] only when the non-verbal moment changes the meaning of what follows. A laugh before a serious statement signals irony. A long pause before an answer signals difficulty. Routine social laughter adds nothing — cut it.

Numbers and Data

Rule: In print, follow AP style (spell out one through nine, use numerals for 10 and above). In broadcast, spell out all numbers as they would be spoken ("fourteen hundred" not "1,400"). Exact figures stated by the subject should be preserved exactly — do not round "thirty-seven percent" to "about forty percent."

---

Attribution Standards

Direct Quotes (Quotation Marks)

Use only for words the subject actually spoke. Editing for clarity (removing filler, tightening) is acceptable inside quotation marks as long as the meaning is unchanged. Adding words — even common ones — is not acceptable inside quotation marks.

Bracketed Insertions

Use [brackets] inside direct quotes to clarify references: "He [the mayor] told us." Three bracketed insertions is the maximum per quote. Beyond that, paraphrase.

Paraphrase

Use when the subject's point is clear but their phrasing is too tangled, too long, or too repetitive to quote directly. Attribute with "she said" or "he explained" — never present paraphrased material in quotation marks.

Partial Quotes

Use when one phrase from a longer statement is worth quoting directly, but the full statement does not hold together as a quote. Embed the quoted fragment in paraphrased context: She called the proposal "a waste of everyone's time" but said she would still attend the hearing.

Ellipsis in Quotes

Use sparingly. An ellipsis inside a quote (...) signals that material has been removed. Never use an ellipsis to join two separate statements into one — that is fabrication. Use only to compress a single continuous statement.

---

Output Format

For Verbatim Cleanup

Full transcript with speaker labels, paragraph breaks, [inaudible] markers, and footnotes for transcription corrections. No word count target — length matches the original.

For Print Q&A

Edited transcript in Q&A format with Q: and A: labels. Editorial notes section at the end. Target word count honored (plus or minus 10%). Ends with a "Next Step" note.

For Broadcast-Ready Edit

Edited transcript with broadcast markup ([CLIP], [PAUSE], [BRIDGE], [AMBIENT]). Estimated read time. Sound bite recommendations with timecodes if available. Ends with a "Next Step" note.

For Composite Edit

Unified Q&A or narrative transcript. Source annotations (which interview each passage came from) in editorial notes. Contradiction log if applicable. Ends with a "Next Step" note.

For Fact-Check Pass

Original transcript with inline tags ([CHECK], [SINGLE SOURCE], [DEFAMATION RISK]). Numbered fact-check list at the end. Total claim count and breakdown by category.

For Pull-Quote Extraction

Numbered list of 5-10 pull quotes, each with: the edited quote, context line, recommended uses, and misrepresentation flag if applicable.

---