# How to convert notes from analog to digital ## Current approach for paper notes My favorite approach to convert analog notes to digital consists in using [[Gemini]] live streaming to show my notes (video), and ask Gemini to convert those to Markdown automatically. Steps: - Go to https://aistudio.google.com/live using my phone or desktop computer - Go to "Run settings" - Set "Output format" to `Text` - Set the system instructions to `When I say or write "go", analyze the text and convert it to markdown. Write nothing else but the Markdown text. Please do not use Python. Use your own vision capabilities` - Start recording (audio + video) - Show a page and say or write "go" - Repeat for all the pages ![[How to convert notes from analog to digital - gemini flash config.png]] Once done, I have all my notes converted to Markdown 🎉 An improvement consists in analyzing a number of pages (e.g., a whole chapter), and generate a single Markdown document with the content and additional information (e.g., a summary, key points, etc). Here's an example: ``` Analyze each image I will paste and convert the text to Markdown. Don't output anything until I say "done". Identify the key ideas and keep those in mind. Then, output a single Markdown document with the following structure: Page 1: <Text of page 1> --- Page 2: <Text of page 2> --- ... --- Summary: <Summary> Key points: <20 key points> Ideas for separate notes: [[<Idea 1>]] <3 paragraphs to include in that note> [[<Idea 2>]] <3 paragraphs to include in that note> ... ``` To further analyze and enrich the generated book notes, you can use something like: ``` You will analyze those, and generate a summary (two paragraphs), as well as a list of 10 key points. You will format your answer as a markdown document ``` ## Alternatives for paper notes ### Using any LLM that accepts images as input Use the following prompt: ``` Analyze each image I will paste and convert the text to Markdown. Don't acknowledge. Don't say you understand. Don't comment. Don't output anything until I say "done". Then, output a single Markdown document with the following structure: ``` ``` Page 1: <Text of page 1> --- Page 2: <Text of page 2> --- ... ``` ### Using Voicenotes (audio recording + transcription) 1) Recording myself reading the notes using [[Voicenotes AI]] 2) Letting it transcribe everything for me 3) Importing, transcription into my Obsidian Daily Notes, or using the [[Voicenotes sync plugin for Obsidian]] to automatically import transcribed notes into [[Obsidian]] 4) Getting rid of the analog notes 5) Extracting those into Atomic Notes during my weekly reviews ### Using ChatGPT on the phone - Take a picture of the analog note - Upload it to ChatGPT and ask it to `Convert the image into a Markdown document` - Using [[Brian Petro]]'s [[Transcribe GPT]] ### Using OCR Use OCR tools such as Textra: https://github.com/freedmand/textra ## Current approach for voice notes I currently record voice notes using [[Voicenotes AI]], which automatically transcribes audio to text. In combination with that, I use the [[Voicenotes sync plugin for Obsidian]], which automatically import the transcribed notes into my [[Obsidian]] vault. This makes it a breeze for me to integrate those notes into my knowledge base. ## Related - [[The value of going from analog to digital]] ## References - Different approaches: https://news.ycombinator.com/item?id=40530719