#Audio Import & Retranscription

Import audio files you recorded elsewhere, or re-transcribe existing meetings with different Whisper models or language settings.

#Import Audio Files

Bring audio from external sources into Clearminutes for transcription and summarisation:

1. Click the Import button in the sidebar (or File → Import Audio)
2. Select an audio or video file from your computer
3. Choose transcription settings:
   • Whisper model (tiny, base, small, medium, large-v3)
   • Language (or auto-detect)
   • Diarization (speaker labels)
4. Click Import
5. Clearminutes transcribes the file and creates a new meeting entry

#Supported File Formats

Clearminutes automatically resamples to 48 kHz and extracts the audio track from video files.

#Import Workflow

Select File
     ↓
Choose Settings (model, language, diarization)
     ↓
Whisper Transcribes (same pipeline as live recording)
     ↓
Meeting Created in History
     ↓
Optional: Generate Summary

Imported meetings work exactly like recorded ones — you can edit transcripts, generate summaries, export, and run action-item extraction.

#Tips for Best Results

• Higher quality audio transcribes better — prefer WAV or FLAC over MP3 when possible
• Longer files take longer — a 2-hour recording with the large-v3 model may take 15-20 minutes
• Use a smaller model for quick previews — import with base first, then re-transcribe with large-v3 if needed
• Set the language — if you know the language, setting it explicitly improves accuracy over auto-detect

#Re-transcribe Existing Meetings

Re-run transcription on a meeting you already recorded, with different settings:

1. Open a meeting in your history
2. Click the Re-transcribe button (or Meeting → Re-transcribe)
3. Choose new settings:
   • Different Whisper model (e.g., upgrade from base to large-v3 for better accuracy)
   • Different language (e.g., transcribe a meeting that was auto-detected incorrectly)
   • Enable/disable diarization
4. Click Re-transcribe
5. New transcript replaces the old one

#When to Re-transcribe

• Wrong language detected — Auto-detect sometimes picks the wrong language for short or multilingual meetings
• Poor accuracy with small model — Upgrade to a larger model for better results
• Need speaker labels — Enable diarization on a meeting that was recorded without it
• Updated Whisper model — Re-run with a newer or better model

#Re-transcription and Your Data

• The original audio file is not modified — re-transcription reads the same audio
• The previous transcript is replaced with the new one
• Any edits you made to the old transcript are lost (the new transcript overwrites it)
• Summaries are not automatically regenerated — generate a new summary after re-transcribing if needed

#Troubleshooting

#Import fails with "unsupported format"

• Verify the file is a supported format (see the table above)
• Try converting the file to WAV or M4A using FFmpeg or an online converter
• Very large files (>2 GB) may fail — try splitting into smaller segments

#Imported meeting has poor transcription

• Try re-transcribing with a larger Whisper model
• Set the language explicitly instead of relying on auto-detect
• Enable diarization if multiple speakers are present
• Check audio quality — background noise, low volume, or echo reduce accuracy

#Re-transcription takes too long

• Use a smaller model for initial preview (tiny or base)
• Enable GPU acceleration for faster processing (see GPU Acceleration)
• Close other applications to free system resources

#Related

• Diarization — Add speaker labels to imported or re-transcribed meetings
• Choose the Right Model — Which Whisper model to use
• GPU Acceleration — Speed up transcription