#Edit & Refine Transcripts

Whisper is highly accurate, but no transcription is perfect. This guide shows how to refine your transcripts by fixing errors, correcting speakers, and reorganizing segments.

#Inline Editing

Fix transcription errors by editing text directly in the transcript:

#Edit a Segment

1. Open a meeting (click to view transcript)
2. Click on any segment (a chunk of text)
3. The text becomes editable (cursor appears, background highlights)
4. Type your corrections
5. Click away or press Enter to save
   • A green checkmark appears briefly, confirming the save

[SCREENSHOT: Transcript segment being edited with cursor visible]

Examples:

• Change "octobet" to "October"
• Fix "a eye" → "AI"
• Add missing punctuation: "hello everyone" → "Hello everyone."

#Triple-Click to Select Whole Segment

1. Triple-click a segment to select the entire text at once
2. Type replacement text (replaces everything)
3. Press Enter to save

Useful for completely rewriting garbled segments.

#Auto-Save

Changes are automatically saved:

• 2 seconds after you stop typing
• Green checkmark confirms the save
• Changes persist if you close and reopen the meeting
• No manual save button needed

#Fix Speaker Labels

Correct speaker names (from diarization or manual labeling):

#Rename a Single Speaker Instance

1. Click on a segment you want to fix
2. Click the speaker label (e.g., "Speaker 1", "Speaker 2")
3. A renaming dialog appears or inline edit becomes active
4. Type the person's name: "Alice", "Bob", "Sarah Chen"
5. Press Enter to save
6. That segment's speaker is updated to the new name

[SCREENSHOT: Speaker label being edited, showing "Speaker 1" → "Alice"]

#Rename All Instances of a Speaker

After you rename one instance, Clearminutes learns the mapping and automatically updates other segments by the same speaker:

1. In the first segment by "Speaker 1", change the label to "Alice"
2. Scroll through the transcript
3. Other segments from "Speaker 1" now show "Alice" automatically
4. If any instance is still wrong, fix it individually

#Bulk Rename Speakers (Pro)

To rename a speaker across the entire transcript at once:

1. Look for the Speaker List panel (right side of transcript, Pro feature)
2. Find the speaker you want to rename (e.g., "Speaker 1")
3. Right-click or click the edit icon next to their name
4. Type the new name: "Alice"
5. Click Apply to All or press Enter
6. All segments from that speaker are renamed instantly

[SCREENSHOT: Speaker list panel with rename option]

#Merge Segments

Whisper sometimes splits one person's continuous speech into multiple segments (due to short pauses, audio artifacts, etc.). Merge them to fix this:

#Merge Two Adjacent Segments

1. Click the first segment you want to merge
2. Hold Shift and click the second segment (to select both)
3. Right-click on one of the selected segments
4. Click Merge or Merge Segments
5. Confirm: "Merge 2 segments?"
6. The segments combine into one with a single timestamp

[SCREENSHOT: Two segments selected (highlighted), merge context menu]

Example: Before merge:

[00:12:30] Speaker 1
"Let me tell you about our Q4 roadmap."

[00:12:45] Speaker 1
"We have three key initiatives."

After merge:

[00:12:30] Speaker 1
"Let me tell you about our Q4 roadmap. We have three key initiatives."

#Undo a Merge

If you merged the wrong segments:

1. Press ⌘Z (macOS) or Ctrl+Z (Windows)
2. The merge is undone and segments are separate again

#Split Segments

If a segment contains speech from two different people or two unrelated thoughts, split it:

#Split a Segment

1. Click the segment you want to split
2. Position your cursor between the words where you want to split
   • Click right before the word that starts the second part
3. Right-click and choose Split Here or Split Segment
4. Confirm: "Split segment at this point?"
5. The segment splits into two at that point
6. The second segment appears below with a new timestamp
7. Assign the correct speaker label to the second segment if needed

[SCREENSHOT: Segment with split point indicated, split context menu]

Example: Original:

[00:45:30] Speaker 1
"Okay, so that's the plan. Alex, what do you think about the timeline?"

After splitting before "Alex":

[00:45:30] Speaker 1
"Okay, so that's the plan."

[00:45:35] Speaker 1
"Alex, what do you think about the timeline?"

Then change the second segment's speaker to "Speaker 2" (Alex) if needed.

#Find & Replace

Replace text across the entire transcript at once (useful for repeated errors):

#Open Find & Replace

1. Press ⌘H (macOS) or Ctrl+H (Windows)
2. A Find & Replace dialog appears
3. Or: Right-click in transcript → Find & Replace

[SCREENSHOT: Find & Replace dialog with two input fields]

#Replace Text

1. Find field: Type the text to search for (e.g., "octobet")
2. Replace field: Type the replacement (e.g., "October")
3. Choose an option:
   • Replace All — Replace every instance (no confirmation)
   • Replace Next — Find and replace one at a time (with confirmation)
   • Find Next — Jump to next match without replacing
4. Close when done (press Escape or click Close)

Examples:

• "teh" → "the"
• "AI" → "AI"
• "um" → "" (empty, to remove filler words)
• "2024" → "2025"

#Case Sensitivity

Some Find & Replace tools offer case sensitivity (optional):

• Match case: "The" matches only "The", not "the"
• If not available, replacements work on all case variations

#Undo & Redo

Reverse your changes with keyboard shortcuts:

Undo depth: You can undo ~20 edits (depends on app configuration)

Important: Undo history is reset when you:

• Close and reopen the meeting
• Switch to a different meeting
• Refresh the app

Keep that in mind if you make major edits — save often by exporting.

#Regenerate Summary After Editing

When you edit transcripts, the summary may become outdated. Regenerate it:

1. Make all your transcript edits
2. Go to the Summary panel (right side)
3. Click Regenerate or Re-summarise
4. Choose:
   • Same model (faster, updates summary with edited text)
   • Different model (try Claude for higher quality)
   • Custom prompt (focus on specific areas)
5. Click Generate
6. Processing begins — results appear in 30 seconds to 2 minutes

[SCREENSHOT: Regenerate Summary button in summary panel]

#Export Edited Transcript

Save your refined transcript to a file:

1. When you're done editing, click Export (top-right of meeting detail)
2. Choose format:
   • Markdown — Full transcript with formatting, timestamps, speaker labels
   • Plain Text — Simple text, no formatting
3. Choose destination:
   • Copy to Clipboard — Paste into email or note app
   • Save to File — Choose location and filename
4. Your edits are included in the export

See Export Notes for detailed export guide.

#Fixing Common Transcription Errors

#Typos & Misspellings

Error: "octobet", "feburary", "Febuary"

Fix:

1. Find & Replace: "octobet" → "October"
2. Or edit individually

#Filler Words

Error: "Um", "uh", "like", "you know" appearing too often

Fix:

1. Find & Replace: "um " → "" (removes all "um" instances)
2. Or manually delete sentences with too many fillers

#Wrong Homophone

Error: Whisper transcribed "to" but meant "too" or "two"

Fix:

1. Click the segment
2. Change "to" → "too" (in context: "too large" vs "too the store")
3. Common mistakes:
   • "to/too/two" (preposition, also, number)
   • "their/there/they're" (possessive, place, contraction)
   • "its/it's" (possessive, contraction)

#Wrong Proper Noun

Error: "John" transcribed as "Jon" or vice versa

Fix:

1. Find & Replace: "Jon" → "John"
2. Or fix the speaker label if it's a systematic error for one person

#Overlapping Speech / Garbled

Error: Two people talking at once gets combined into gibberish

Fix:

1. Listen to the original audio (click timestamp)
2. Manually rewrite what each person said
3. Split the segment if needed to separate speakers
4. Assign correct speaker labels

Prevention: VAD or noise suppression may help reduce overlapping speech captures. See Recording Quality.

#Missing Words

Error: "The meeting is scheduled for Tuesday" becomes "The meeting is for Tuesday"

Fix:

1. Listen to the audio (click the timestamp)
2. Type the missing word back in
3. Prevention: Lower VAD sensitivity or increase mic gain for next recording. See Recording Quality.

#Wrong Language Detected

Error: English meeting transcribed with incorrect language model

Fix:

1. Check Settings → Transcription → Language
2. Manually set to "English" (or your meeting language)
3. The transcript won't change, but future recordings will use the correct language

#Transcript Editing Best Practices

1. Edit immediately after recording — Context is fresh, easier to spot errors
2. Listen to audio for unclear segments — Click timestamp to play original audio
3. Fix speaker labels first — Then edit text with correct context
4. Use Find & Replace for repeated errors — Faster than individual edits
5. Export before major changes — Keep a backup of the original
6. Lock important quotes — Prevent accidental future edits
7. Regenerate summary after major edits — Ensures summary matches edited text
8. Don't over-edit — Some filler words (um, uh) are authentic; leave natural speech patterns

#Keyboard Shortcuts for Editing

#Related

• View Transcripts & Summaries →
• Generate Summaries →
• Export Notes →
• Diarization (Pro) →