Why am I getting an out of memory error for uploading 30-min voice recordings in my workflow with Google Drive, AWS S3, OpenAI Whisper, and Notion?

Help
61

This seems to be related to the filetype of your audio file. Are you uploading an mp3?

62

m4a

63

Hmm - I’m guessing it’s having trouble with some aspect of your M4A then

64

I don’t think Drive can do this, but Dropbox has an automations feature that can convert uploaded files to mp3

65

And maybe you can do it right on pipedream too

66

interesting let me see if converting to mp3 first helps

67

also how do you add it to notion? I’m doing something really hacky and I don’t feel good about it lol

image.png
image.png2006×764 215 KB

68

to be able to fit 2000 characters per component

69

I have a step where I chunk the transcription up

const chunks = [];
    let currentChunk = "";
    const words = steps.create_transcription.$return_value.text.split(" ");

    for (const word of words) {
      if (currentChunk.length + word.length + 1 <= 2000) {
        if (currentChunk === "") {
          currentChunk += word;
        } else {
          currentChunk += " " + word;
        }
      } else {
        chunks.push(currentChunk);
        currentChunk = word;
      }
    }

    if (currentChunk !== "") {
      chunks.push(currentChunk);
    }

    return chunks;
70

Is this 2000 character limit coming from Notion or Pipedream?

71

That’s a Notion limit

72

I think you can try creating an empty page, then use the Append Block action. I tested rn with way more than 2000 characters and it worked

image.png
image.png1654×916 70.7 KB

73

the 2000 character limit per block comes from notion

74

Cool let me try the append block action

75

ok so there seems to be a limit to how big of a file you can upload. When I converted my 81min m4a to mp3 it went from 140MB to 73MB and then the workflow worked. When I leave it as a larger m4a file it gives me this error

image.png
image.png2034×692 229 KB

76

I think it has to do with size, not the fact that it is m4a since I haven’t had trouble with smaller m4a files

77

Cool thanks, I’ll add a caveat to the description about large files

78

Hey I got an out of memory error again today (haven’t run the automation for a couple days). I’m using 4096MB in my workflow. Thoughts on why this might be the case?

79

Hey , from my tests, I can say that the memory error in this component doesn’t actually happen because the workflow consumed too much memory, but because something kept the workflow hanging and didn’t let it complete.

Let me ask what size/audio length is the file you are transcribing? If it’s toooo large, I suggest you manually split it into smaller chunks while we work on the rate limiting. @U036XC1NX61 is currently working on that