audio-transcription

Transcribing audio and video into structured notes involves sending binary media to the Gemini model, which generates a Markdown transcript with summary and transcript sections for meetings, podcasts, and lectures.

Yes, Gemini can transcribe podcasts directly inside Obsidian by reading the audio file and saving the structured Markdown transcript as a new note in your vault.

The maximum file size limit for transcribing meetings is 20 MB per file, requiring read_file to feed the media into the Gemini model for processing.

Yes, the transcription process exports the transcript back to Obsidian as a new note, utilizing write_file to save the Markdown file with metadata for easy organization within your vault.

The Gemini model handles video transcription for lecture notes by processing binary media content and producing a structured Markdown output that separates the summary from the full transcript text.

You need to use read_file and write_file for audio transcription in Obsidian because the Skill requires read_file to feed binary media to the Gemini model and write_file to save the resulting transcript locally.