🤖

Voice Note

👋
This note is an example of the speech-to-text workflow I built. It allows me to record voice notes on my phone, then sends a perfect transcript and summary directly to Ultimate Brain.

If you want to set this workflow up for yourself, you’re in luck! It takes less than 10 minutes, and this video contains all the instructions:

🤖
This AI transcription/summary was created on @October 14, 2023. Listen to the original recording here.
🔥
Feel free to delete this note if you like! It’s simply here to demonstrate how a Voice Note will look in Ultimate Notes.

Summary

One of the easiest and most useful ways to take notes in Ultimate Brain is by using voice notes. By recording a voice memo on your phone and uploading the audio file to a cloud storage provider like Dropbox, Google Drive, or Microsoft OneDrive, an automation system will transcribe the audio into text. The text can then be summarized using ChatGPT to provide main points, action items, follow-up questions, references, and translations. The workflow is efficient and can be accessed using apps like RECUP on iOS.

A new app called Flylighter is also being developed to provide voice recording and summarization. The cost of using this workflow depends on the usage of OpenAI APIs, with an estimated cost of 40 cents per hour of uploaded audio using the default GPT 3.5 Turbo model. The audio file and tutorial video can be accessed through the provided links. Overall, this workflow is valuable for brain dumping ideas or transcribing scripts.

Transcript

👋
No modifications have been made to this transcript other than the links added to the Flylighter mentions. Whisper’s transcription model is just that accurate.

One of the easiest and most useful ways to take notes that end up in Ultimate Brain is to use your voice. I have a whole workflow where you can simply record a voice memo on your phone, be it on Android or on iOS, upload the audio file to your cloud storage provider of choice, be it Dropbox, Google Drive, or Microsoft OneDrive, and a whole automation that I spent months building will transcribe your audio into near-perfect text, then use ChatGPT to summarize that transcript, giving you lists of main points, action items, follow-up questions, references and citations, and all sorts of other useful things, and then send a brand new page to Ultimate Brain containing the transcript, the summary, those lists, and even a translated transcript if you would like to speak in one language or say record audio from another speaker in one language and then translate that to your language. It is a super useful workflow. I have a whole tutorial on how to build it.

And as you can see from the transcript and the summary in this sample document, it works really well. I personally use an app called RECUP on my iPhone. That's R-E-C-U-P. It is currently the only app that I know of that will automatically upload audio to any source, and the only source it uploads to is Dropbox.

As of this recording, the only other option I know of is to use the record audio action inside of iOS shortcuts and then save the file to Dropbox. Again, Dropbox is the only cloud storage provider that seems to have taken the time to create shortcuts actions, and hence is the only cloud storage provider that I know of that you can use to automatically upload audio. My team and I are working on fixing this problem. We are building a brand new app called Flylighter, which is not only the best web clipper for Notion on both mobile and desktop platforms, but will eventually have a voice recording workflow that will automatically upload your recordings and do the summarization for you.

So if you're interested in that, we have a waitlist over at flylighter.com. You can check that out. But for now, you can set up the free Pipe Dream workflow that will do this for you. It works today, and it is completely free to set up.

The only cost involved in using it is your usage of the OpenAI APIs, both for transcribing your text and then for getting it summarized. The numbers I ran came out to around 40 cents per hour of uploaded audio, as long as you are using the default GPT 3.5 Turbo model, which works pretty well for summarizing text. If you wanted to use GPT-4 instead, you can do that. It will slightly increase the cost of each run.

However, most of the cost is in the transcription step. And my team and I are working on making that even cheaper in Flylighter. So eventually this workflow should be much, much cheaper than it currently is. But still 40 cents per hour of uploaded audio is incredibly cheap.

And for my purposes, at least, being able to say, go for a half an hour walk and brain dump ideas or run through a video script that's in my head and get that perfectly transcribed and summarized is super, super worth it. So again, I'll have the links and the tutorial video in this exact same note that you're looking at. You can listen to the audio file by clicking the link to it. I will keep that public in Google Drive so you can see how it sounded, or I guess hear how it sounded.

And yeah, try it out. Hopefully you found it useful. And once again, thanks for buying Ultimate Brain.

Additional Info

Main Points

  • Voice notes can be recorded on a phone and uploaded to a cloud storage provider
  • An automation system transcribes the audio into text
  • Text can be summarized using ChatGPT to provide useful information

Potential Action Items

Set up the free Pipe Dream workflow for audio transcription and summarization (cost involved)
Check out the waitlist for the new app Flylighter for a more cost-effective workflow

Follow-Up Questions

  • Visit flylighter.com to join the waitlist for the new app
  • Explore the provided links to access the audio file and tutorial video

Meta

  • Transcription Cost: $0.022
  • Chat API Cost: $0.003
  • Total Cost: $0.025