iPhone Voice Memo -> Obsidian

Hi folks, just wanted to quickly share something I worked on today.

This works on any iPhone that supports Apple Shortcuts.

I know there are some other people who have tried similar things, but I could never find something that has the 3 following properties:

  • it should reliably transcribe audio
  • it should be practically transparent (don’t want to interact with another app)
  • should also add a summary of you

Here’s a small demo on how you actually use it (sorry no sound):

If there’s enough interest i might go a little bit of detail how i did the apple shortcut but the gist is:

  1. Get file from share clipboard
  2. Send to openai whisper
  3. get transcription and send it to ChatGPT to format the text & summarize
  4. save on iCloud dir where obsidian is located

IMO would be nice that the experience is even more seamless. e.g. you could record audio as you normally would without any extra action and it would upload and transcribe automatically to your obsidian.

You can check the demo audio here if you’re curious:

6 Likes

Thank you for sharing. Looks very interesting!

Do you mind sharing the shortcut? Also, would it be possible to achieve the following:

  • Insert the link of the audio into the Obsidian note, e.g., ![[Audio]]?
  • Allow the user to customize the template of the note.

This is some exciting stuff indeed!
Is there any way of sending it to a local whisper server? I’m running a whisper addon on my Home Assistant, and I’m sure there are ways to do this somehow locally…

@dragnar @Wen

sorry a bit late response:
https://www.icloud.com/shortcuts/e0623cdc4fcf476993f700f3b9909c02

this one is a bit scrappy tried to adapt it so that it’s configurable if you know iOS shortcut i’m sure you can modify a few things to make it work

there’s two important vars:
a) your openai api key
b) the folder where the text is saved

3 Likes

This is amazing! Thank you for creating it.

1 Like

btw i created something that (i believe) is a big improvement.

but the downside is you need to have an extra plugin installed

You can replace the apple shortcut by a clicking on “Share” then choose “Save to Files”.

1 Like

My guy, I have been working on something similar to this. Would you be interested in working together to develop something? Someone has yet to crack the code on AI note-taking and I think we can do it.

happy to have a chat, dm here https://twitter.com/hotkartoffel1 @ianjamesburke

that’s really cool. would love to hear a bit more about your setup i find there’s something interesting we could be doing there if you’re in for it would you be up to send a dm to https://twitter.com/hotkartoffel1

Hey! I am very interested in what you guys are doing here. I have hundreds of voice memos on my iphone full of story ideas and worldbuilding notes that I would love to get transcribed. However, I am leery of AI services that might use my intellectual property. Is there a way to do this safely in that regard? Unfortunately my knowledge of this stuff is very limited, and I know almost nothing about programming. Any advice would be appreciated!

OK, so I found a partial solution. I found an app called Aiko that runs whisper entirely offline on my iphone so that the data doesn’t leave my device. It saves the transcription onto an icloud folder as a txt file. What I am trying to do is somehow automatically convert these txt files into markdown and import them into obsidian with a template I have already set up for raw transcripts. I dont see a way to do this with folder actions or automator on my Macbook but I am also very inexperienced with these functions.

2 Likes

Great information in this thread.

I have a similar situation to thebobcat. I take voice memos here and there to avoid forgetting good ideas when I’m running. I wanted to be able to get these into obsidian and process them into my vault by fixing typos, linking notes, bolding, etc. I don’t need all the functionality in benjaminshafii’s AI File Organizer 2000 (great work btw). I wanted something simpler so I made a plugin similar to the one benjaminshafii shares above. Though, there are two important differences.

  1. It uses Aiko to locally transcribe as thebobcat mentioned
  2. It is on RoutineHub, which means I can version it and add improvements over time.

Here’s a quick demo. It’s two clicks to transcribe from the Voice Memo app.

You can grab the shortcut and see the documentation here: https://routinehub.co/shortcut/18505/