Skip to main content

​
What can you do with it?

Transcribe video files to text with timestamps, speaker identification, and meeting summaries. Perfect for converting meetings, presentations, interviews, and other video content into searchable, readable transcripts. Automatically pre-processes video by extracting audio for optimal transcription quality.

​
How to use it?

​
Basic Command Structure

/video-transcription [prompt] [video-file]

​
Parameters

Required:
  • prompt - Instructions for transcription (timestamps, speakers, summary, etc.)
  • files - Video file to transcribe (supports MP4, MOV, AVI, and other common formats)
Optional:
  • model - Gemini model to use (defaults to gemini-2.5-flash)
  • output filename - Custom name for transcript file

​
Response Format

Asynchronous Response (all video transcription is async): 
{
  "message": "Process started. Results will be in Multimedia Artifact collection",
  "statusUrl": "https://skills.pinkfish.ai/llm/gemini/async/job-12345",
  "placeholderFile": {
    "fileId": "abc123",
    "signedUrl": "https://skills.pinkfish.ai/files/collection/transcript.txt",
    "fileName": "transcript_20251007.txt"
  },
  "responseId": "job-12345",
  "status": "queued"
}

​
Examples

​
Basic Transcription

/video-transcription
prompt: Transcribe this video with timestamps and speakers
files: meeting_recording.mp4
Creates a basic transcript with timestamps and speaker identification.

​
Meeting Summary

/video-transcription
prompt: Transcribe this video with timestamps and speakers. Also provide a summary of key topics discussed and any action items mentioned.
files: team_meeting.mp4
Generates transcript plus meeting summary with action items.

​
Detailed Analysis

/video-transcription
prompt: Transcribe this video with timestamps and speakers. Include a summary of key decisions made, technical terms discussed, and next steps identified.
files: project_review.mp4
output filename: project_review_transcript.txt
Comprehensive transcription with detailed analysis and custom filename.

​
Notes

Processing Details:
  • Always Async: Video transcription uses background processing due to file size and complexity
  • Audio Extraction: Automatically extracts audio from video for optimal transcription
  • Output Format: Plain text transcript with timestamps and speaker labels
  • File Storage: Results saved to Multimedia Artifact collection
  • Processing Time: Varies based on video length (typically 2-5 minutes for 30-minute videos)
Common Instructions:
  • “Include timestamps for each speaker turn”
  • “Identify speakers by name if mentioned”
  • “Provide a summary of key topics discussed”
  • “Extract any action items or decisions made”
  • “Note technical terms or tools mentioned”
  • “Highlight concerns or issues raised”
Output Format: The transcript file contains:
  • Timestamped dialogue with speaker identification
  • Meeting summary (if requested)
  • Action items and decisions (if requested)
  • Technical terms and references (if requested)

​
Supported Models

  • gemini-2.5-flash (default) - Optimal balance of speed and accuracy for transcription
  • gemini-2.5-pro - Enhanced accuracy for complex audio or multiple speakers
  • gemini-2.5-flash-lite - Faster processing for simple, clear audio
All models support video transcription with timestamp and speaker identification.
⌘I