Audio Skill

Audio understanding - transcription and analysis using Gemini.

Description

Rhea can listen and understand audio input. Uses Gemini's native audio understanding for transcription, summarization, and analysis.

Actions

• •transcribe — Generate transcript from audio
• •analyze — Describe and analyze audio content
• •summarize — Summarize audio content

Usage

# Transcribe audio file
skill.execute("transcribe", audio_file="path/to/audio.mp3")

# Analyze audio content
skill.execute("analyze", audio_file="path/to/audio.mp3", prompt="What instruments are playing?")

# Summarize a podcast
skill.execute("summarize", audio_file="path/to/podcast.mp3")

Supported Formats

• •WAV (audio/wav)
• •MP3 (audio/mp3)
• •AIFF (audio/aiff)
• •AAC (audio/aac)
• •OGG Vorbis (audio/ogg)
• •FLAC (audio/flac)

Technical Notes

• •1 second of audio = 32 tokens
• •Max audio length: 9.5 hours per prompt
• •Audio is downsampled to 16 Kbps