Audio Understanding: Audio transcription and analysis with Gemini

File support

This skill supports audio analysis using Google Gemini models. Supported formats:

• Local audio files up to 9.5 hours long
• YouTube links (youtube.com/watch, youtu.be, youtube.com/embed)

Reference: https://ai.google.dev/gemini-api/docs/audio?example=dialogue

How to use

bash ${CLAUDE_PLUGIN_ROOT}/scripts/gemini.sh --file=AUDIO_PATH "YOUR QUESTION ABOUT THE AUDIO"

Arguments:

• --file – Required: Local audio file path or YouTube URL
• --model – Optional: Model to use (defaults to gemini-3-flash-preview)

Examples:

# Transcribe a local audio file
npx -y superconductor-gemini-skills --file=recording.mp3 "Transcribe this audio"
npx -y superconductor-gemini-skills --file=meeting.wav "Summarize the key points discussed"

# Analyze a podcast or YouTube audio
npx -y superconductor-gemini-skills --file="https://www.youtube.com/watch?v=dQw4w9WgXcQ" "Transcribe this audio"
npx -y superconductor-gemini-skills --file="https://youtu.be/dQw4w9WgXcQ" "What topics are discussed?"

# Extract specific information
npx -y superconductor-gemini-skills --file=interview.m4a "List all the questions asked by the interviewer"
npx -y superconductor-gemini-skills --file=lecture.ogg "Create a bullet-point summary of the main concepts"

API Key

The GEMINI_API_KEY environment variable must be set. Get your key at: https://ai.google.dev/gemini-api/docs/api-key

Models