Select text to annotate, Click play in YouTube to begin
00:00:00
well hello there so this week I built a bot that takes voice notes like the one that I'm recording right now turns into text and sends that text to my notes database in notion and by text I don't just mean a transcript I also get a full
00:00:12
summary a list of the main points in the text and also a list of action items today I'm going to show you exactly how to build this workflow for yourself it is surprisingly easy to set up and once you do have it built it is completely hands off and automated which makes it
00:00:26
awesome now there are two recent AI tools that are doing the heavy lifting in this automation first open ai's whisper model is going to take the audio file and turn it into a transcript basically turn it from audio to text and
00:00:38
secondly chat GPT is going to take that transcript and use it to generate the summary and those lists of action items and main points and so on of course what chat should be T actually does with our transcript is entirely dependent on the prompt that we give it which means if
00:00:51
you wanted to you could change up that prompt and do something different like say recording a note on your phone and getting a blog post draft as a result and for that reason they later on in the video I'm actually going to share some tips for writing consistently better prompts that are going to give you
00:01:04
better output which means that you can either replicate my workflow exactly or you could change it up and customize it to your own note-taking needs and it's that customization that makes me so excited for both AI Tools in general but
00:01:17
also this workflow in particular because I see it as really widening the pipeline between my actual brain and my second brain inside of notion in the past if I wanted to brain dump a thought into my second brain while I'm on the go I would
00:01:30
have to use my thumbs and my phone keyboard which is very very slow but now I can just grab my phone talk to it and get that perfect transcript inside a notion which is super cool by the way if you do want to turn notion into your own second brain as I've done my ultimate
00:01:43
brain template for notion is the best way to do that it's got a robust task manager it has a project management dashboard as goal tracking and most relevant to this video it has a full note-taking system in fact this is my
00:01:56
note-taking system and one customization that I made to it recently by I go down to my voice notes section here is all of my AI transcribed notes that I have made through talking to my phone are in this
00:02:07
section along with this page right here if I click one at random I've got my summary I've got my transcript all that good stuff so if you do want to learn more about ultimate brain you can go over to thomasjfrank.com brain to check
00:02:20
it out there is also a discount code in the description down below so check that out as well and I'll talk a bit more about the features near the end of this video but for now let's start building now as always I've got the entire videos table of contents right there and below
00:02:32
so skip around to your heart's content we are going to kick things off with a bit of an overview of the four tools you will need to actually get this to work the first one is going to be a notion account which I assume you already have but there is a notion Link in the
00:02:44
description down below if you haven't signed up you will also want a notes database inside of notion that can be any old database so it could be a template like ultimate brain you could use one of my free templates or you could just set up a database inside of
00:02:55
notion just make sure you have a database ready secondly you're gonna need an open AI account this is what's going to allow you to actually use the whisper API for transcribing your audio and then use the chat GPT API for the
00:03:07
summaries in the lists and so forth and you can get that over at platform.openai.com third you'll need a cloud storage provider somewhere to upload your audio that will also kick off for automation I'm going to use Google Drive in this tutorial but you
00:03:20
could also use dropbox you could probably use OneDrive just get the one that works best for you and fourthly you'll need a pipedream account which you can get over at pipedream.com and pipedream is the automation Builder that's going to allow us to stick all
00:03:33
these other apps together and that's going to drive the Automation and have it automatically kick off every time a new audio file is uploaded to Google Drive and on that note I want to go over to Whimsical which is one of my favorite
00:03:45
tools here to basically show you how this automation is going to work at a high level so basically we are going to be taking voice notes on our phone they're going to be uploaded to Dropbox or Google Drive whatever you want and then pipedream is going to kick off an
00:03:58
automation that is going to query The Whisper API the Open ai's chat GPT API and then we're going to send all of the stuff we get from openai to a brand new page in notion so that is like the broad
00:04:11
Strokes overview of what's going to happen uh from your perspective but what's actually going to happen behind the scenes well if we scroll down on the canvas to this little graph that I've created here we get a bit more of a
00:04:22
detailed look so this is where we're actually going to be building we are going to be uploading our audio files to our cloud storage and the moment an audio file hits that cloud storage in what we're going to call a watch folder
00:04:34
pipedream is going to see it and that is going to be the trigger for our automation at that point we are going to be downloading the audio file into what's called Temp Storage we're going to send the audio file to Whispers for
00:04:47
the transcription we're going to send the transcription basically the text to chat gbt for summarization in those lists then we're going to do a little bit of formatting to get all that stuff in a nice little bundle and we're going
00:04:59
to send it on to notion creating a brand new page so hopefully that sounds relatively doable to you because it is actually time to start building and the first step that you're going to want to take is signing up for a pipe dream account I'm going to assume you already
00:05:11
have one and once you have one you should see this little dashboard here or it might take you to a brand new workflow right away if it doesn't you will have this new button right here and I'm going to go ahead and click that and start building our workflow so this is
00:05:25
basically where everything is going to happen and we're going to create a series of steps much like you would in zapier or make.com and then when our automation actually triggers it's just going to go through all those different steps so I'm going to first name this
00:05:38
workflow and let's just name it uh Speech to Text to notion and the first step of workflow is always the trigger and the trigger is also the only step in each workflow that you cannot rename it's just always called trigger so in
00:05:51
this instance we want our workflow to trigger every time an audio file is uploaded to a specific folder in Google Drive so to do that I'm going to search for Google Drive and actually just right there on the screen I'll click that and now I have a list of different actions
00:06:03
that I can pick from and for Google Drive there is a new files action and says emit a new event anytime a new file is added in your linked Google Drive so I'm going to choose that as my trigger and now I need to actually connect a Google Drive account so I'm going to go
00:06:17
ahead and click to connect and once I'm connected you can see that the drive option here is set to my drive that is actually what we want we could choose like shared drives if they existed we do want my drive and I want only files that
00:06:29
go to a specific folder to be watched in this case I don't want it watching my entire Google Drive account so I'm going to click the folders optional field here and I'm going to search for a folder and I actually have a folder pre-prepped for
00:06:42
this that's just called audio upload test I believe there it is I'll check that and I will hit create source and now that I've got my trigger set up before I can start making additional steps in this workflow it says action
00:06:55
required and it wants me to generate a test event for this trigger well this this event involves uploading audio files to Google Drive now I could do that from my voice notes app but for demonstration purposes here I'm just
00:07:07
going to open up my Google Drive to this folder and I'm going to drag an audio file into it and now I can see that one new event has been detected by the trigger so I'll go ahead and open that up and I will select this new file that
00:07:20
it has detected and every time you get a successful result in a step in pipe dream you're going to get this green success message and also in the exports tab here you're going to get an exported object so basically what we're going to be doing throughout this entire process
00:07:33
is accessing properties of this object and referencing them in additional steps now you will notice that there is both a copy path option and a copy value option if I click copy value I'm going to be
00:07:46
copying the exact value from this particular run of the automation so it'll be the one percent rule sampleaudiotest.n4a usually you don't want to do that usually you want to copy the path because that's going to dine
00:07:59
dynamically reference the value that comes in every single time that the automation runs so to continue I actually want to go up to this ID property right here and I'm going to hit copy path so I can dynamically reference this ID so we're going to go ahead and
00:08:12
hit continue and then the next thing that we need to do is actually download the file from Google Drive into pipedream's temp directory the reason we need to do that is pipedream has been authenticated and can access your Google
00:08:25
drive but open ai's whisper cannot we have to actually upload the file to whisper and to do that we need to download it from Google Drive first so luckily Google Drive actually has an action for this so once again I'm going
00:08:38
to search for drive in this step I'm going to find Google Drive and there should be a download a file action you can either search for it or you can hit load more actions to find it and I'm just going to go ahead and search
00:08:50
there it is download file and once again we can see my Google Drive account is already connected and under the file property we're going to reference that Dynamic value from the previous step so to do that instead of choosing one of
00:09:03
these uh options here I'm going to go to enter a custom expression and I could either go here and hit select path or because I copied the path earlier I could just paste it right here so notice we're using double brackets here this is
00:09:15
the convention that pipedream uses for referencing these projects and if you went through my notion API guide it's not necessary for this tutorial but that could be useful you will probably recognize this format here essentially
00:09:27
we are accessing properties of an object called steps and those properties might have other objects as their values so if I go back up to results we can kind of see what's happening here there's this
00:09:40
steps.trigger object and then I can sort of drill down into that one of the properties is called event and that is an object itself and I can drill down into that and find the ID property which has this value so I'm basically telling
00:09:53
it I want to download the file that corresponds with that ID and then I need to provide it with a destination file path and we get a little bit of help here it says the destination path for the file in the slash TMP directory and
00:10:06
it gives you an example so what I could do here is I could type slash TMP and then just give it a name like slash recording dot MP3 the only problem with this is you might actually have different kinds of audio
00:10:19
files that you upload to your Google Drive for example if you're using an iPhone The Voice Notes app actually exports m4a files but if you're using a different app maybe it does MP3 files so instead of typing MP3 there which is
00:10:32
going to give us an error what we actually want to do is go back up to our results here and before I do that I'm going to go ahead and hit this little pin icon so this actually stays open but I'll toggle Open results I'll pin that as well and we can actually get the file
00:10:45
type of this file because it is one of the properties in this event object if I click more here I can find more properties and if I scroll way down there is this full file extension property and we can see m4a is displayed
00:10:57
right there so if I copy this path what I can do is come back to my destination file path and instead of dot MP3 I'm going to do Dot and there it is steps trigger event full file extension so now
00:11:10
if I do an m4a file or an MP3 file or really any kind of audio file that whisper supports it's going to work and I'm going to be able to dynamically set that extension so I'm going to hit test here and as long as I did things
00:11:22
correctly I should get a success message and then I should have that file in Temp Storage and there it is we've got a return value we have the name right there and we have the mime type all that is good so the next thing that we want
00:11:34
to do is actually send our audio file to whisper to open Ai and get it transcribed so to do that I'm going to hit the plus icon down here and then I'm going to look for the openai chat GPT
00:11:46
app and when you click that you're going to get a list of actions just like you would for any of the other apps and the one that we want to pick here is create transcription so if I click that it's going to ask me to connect to my openai
00:11:58
account and this is where I'm going to actually show you how to make an openai account if you don't already have it if you already use chat GPT you likely have one if you don't you're gonna have to sign up but either way you can go over to
00:12:09
platform.openai.com to log in or create an account and once you do if it is a brand new account you should get five dollars worth of free tokens that are going to expire after three months and if I'm doing my math correctly that should actually give you about 12 hours
00:12:22
of audio that you could upload and transcribe and get summarized which is a pretty darn good amount for a free trial anyway after you sign up you're gonna have this little uh dashboard here and if you go up to personal in the top right corner and you go to manage
00:12:35
account the first thing I will point out is that if you do want to upgrade to a paid account after that trial is over you can do so in the billing overview area and if you want to you can even set a hard limit that will ensure you don't
00:12:48
go over a certain amount of spending every single month which is pretty cool after that you're going to go to the user API Keys option here and you're going to create a brand new secret key now I am going to delete this key after I make this video so don't go trying to
00:13:01
use it make your own key pay for your own AI transcription please thank you but I'll hit the copy button right here and I'm going to go back to my pipedream workflow and paste set API key right here and hit save the next thing I'm going to select is my audio upload type
00:13:14
and because we have our file and Temp Storage I'm going to click file right here and then here I actually have to define the file path for our file well earlier we typed slash TMP slash
00:13:27
recording Dot and then what we actually have to do is once again get that file extension from our first step here so I'll kind of scroll down here to show you this you can actually go into any of
00:13:39
the objects for previous steps here and we know that our file extension was in the trigger steps object so we're going to go there and we're going to see if we can find it and actually if we search for it I believe it was full file
00:13:52
something full file extension there it is cool we'll select path and now this should access our recording dot in this case m4a perfectly well so once again we're going to go ahead and test so I will point out there are some optional
00:14:05
Fields here typically with transcription you don't need to set any optional Fields but some of these will actually set in the next step when we're working with chat GPT so if for now I'm going to go ahead and hit test and if things go well we should get a full transcript of
00:14:19
the audio file that I just uploaded and just like that we now have our transcription in this little return value transcription property here and we can see it is pretty much a perfect transcription of my audio file now this
00:14:32
is the part in the building process where you can run into a couple of different errors so I want to briefly cover those before we move on first you might get an error when you test this transcription step saying that you're recording no longer exists that can
00:14:45
happen because the temp directory doesn't keep files in it for very long now when you're actually running the automation that's not going to be an issue but if you're taking a while to build this automation while watching this video you might run into that problem so if it happens just go ahead
00:14:58
and basically upload another file to Google drive test the download file step again and then it should work the other potential problem is that by default pipedream workflows will time out after 30 seconds if they haven't finished now
00:15:10
most workflows are going to finish a lot faster than that but with whisper and particular if you're uploading a pretty long file it can sometimes take longer than 30 seconds so to deal with that you can go over this three dot menu up here go to your settings and I'm going to go
00:15:23
ahead and leave without deploying and you will find this execution control setting here and I'm just going to go ahead and set my timeout value from 30 seconds to 180 seconds I'm also going to mention that there is a lot more detail
00:15:35
around pipedream and this entire workflow in the written version of this tutorial I'll have that link in the description down below there is also what I'm going to call the code heavy method that I'm actually using
00:15:47
personally so if you want to do this the code heavy method with all these code blocks you can just copy and paste there are some advantages to doing that and the blog post kind of goes through them for this video I'm keeping things simple I'm just doing the no code method but
00:16:00
there are some details about the pipe dream settings and pricing and workflow all that's in there I'm not going to pad out this video with all that detail just in summary I'm setting my timeout value to 180 seconds and I'm going to go back
00:16:12
to my editor and we'll keep building so now that I have my transcription I can add another chat GPT step so I'm going to go with openai chat GPT once again and I'm going to choose chat this
00:16:23
time instead of transcribe so this is the point where we're actually going to be working with chat gbt via the API to summarize our transcript to turn it into lists all that good stuff so let's take
00:16:36
a look at the different configuration options we have here we're already authenticated with our openai account we don't have to worry about that if you happen to have beta access to gbt4 Via the API well number one lucky you and
00:16:48
how can I get that access because I don't have it right now but you could select that right here otherwise I recommend going with the GPT 3.5 turbo model this one right here is Frozen as of March 1st 2023 so you're going to
00:17:01
want to pick this one that isn't currently Frozen and then the user message is going to be your prompt and with that we are now going to briefly go to prompt school because your output that you get from chat gbt is very very
00:17:14
dependent the quality of it is very dependent on the quality of your prompt so there are three different parts to a prompt there's What's called the query or the user message there is the context and then there is the instructions or
00:17:27
the system message so basically the user message is what you would type into say chat gbt like write me a poem or summarize this email that I got the context is going to be the transcript in
00:17:41
this case an email or basically anything that you're basically asking uh the chat bot to analyze and use as context in your prompt and then there is the system instructions now if you're working with chat GPT via the actual website the
00:17:55
system instructions are one and the same with the user message it's going to kind of try to parse that out but when you're working with the API that is actually a different parameter it's called the system instructions and if you were to look at the code it would actually be a
00:18:08
different section with a completely different prompt and with the system instructions do is essentially instruct chat GPT on how to respond in terms of formatting and you can also give examples in the system instructions and
00:18:21
I have found if you give an example in the system instructions instead of the user message your output is almost always going to be a lot better okay prompts go over let's actually start filling out the user message and then we will handle the system instructions and
00:18:34
to fill this user message field out I'm actually going to go back over to the written version of this tutorial because I have the prompt pre-written so I will have this link in the description down below via the handy dandy table of
00:18:46
contents on the sidebar underneath the no code method which is what we're doing in this tutorial you can find this little summarize the transcript with chat GPT step and if you scroll down a little bit you're going to find this code block that has the exact prompt
00:18:59
that we're going to use so you can click copy the clipboard you can go back over to your user message field and I'll just go ahead and paste that in and we can look at what we're doing here so all of this is what we would call the user
00:19:12
message and I'm saying write a title for the transcript that is under 15 words I'm being specific I'm asking for a specific limit in the words of that part here and then I'm actually also asking it to
00:19:23
write dash dash summary dash dash so the reason I'm doing this is this is what I call a delimiter it's going to basically allow us to parse through the output that we get from chat gbt and do some
00:19:36
useful stuff with it and the reason that we're doing that is we actually want to separate the title the summary and then the lists and place them on our notion page in different spots obviously the
00:19:48
title is going to go in the title field right here and then the summary is going to go at the top and we want to have at least I want to have all of these lists beneath the actual transcript so to do that we're
00:20:02
going to need a way to split the title the summary and the lists into three separate pieces of data that we can work with individually and the way that I'm going to do that is asking chat GPT for
00:20:14
these delimiters here summary and additional info beyond that I'm asking it to write summaries at heading 1 and then I'm asking for a summary and then down here I'm asking for a list of the main points action items follow-up questions potential arguments against
00:20:27
the transcript and you can get creative here ask for any kind of list you want or ask for basically anything you want it's chat GPT and then at the bottom I'm providing the transcript and once again I am dynamically linking and this is
00:20:40
kind of a weird bug here that I'm seeing but I'm dynamically linking to the return value of the previous step and I'm going to go ahead and delete that out of here and redo it just to sort of show you how it works so if I come back
00:20:52
up to my results here and create transcription I'm going to find this transcription property here I'm going to copy the path and once again this is how we dynamically reference values in
00:21:05
additional or subsequent steps here in pipedream so basically it's going to take this as the user message it's going to take the entire transcript from the previous step as the context and the last thing we want to set is system
00:21:17
instructions here so if I go back to my written version I can scroll down just a bit further and I can find my system instructions so this is the real Secret Sauce here for getting consistent output I tell it
00:21:29
you are an assistant that only speaks in markdown which is a formatting language that gets us like headings and bullet lists and all kinds of stuff like that do not write text that isn't formatted as markdown so that right there should
00:21:42
be enough to get chat gbt to always respond in well-formatted markdown but to be extra thorough you can actually give it example formatting so here I've said example formatting and I've given it exactly what I want including these
00:21:56
delimiters as I want them written so if I set that as my system instructions I set this as my user message and I pass it the transcription as the context I should be able to test it and I should get a good result one other thing that
00:22:10
I'm going to do though is I'm going to set a temperature setting so temperature is a value from 0 to 1 that basically determines or influences how creative the output is
00:22:21
going to be so if you want Chachi BT to be very creative you want to write like a poem or something or a blog post you might want to set a high temperature value but in this case I want well-formatted markdown and I also want it to be pretty straightforward in
00:22:34
summarizing and creating main point bullet lists out of the transcript that was provided so I'm going to give it a value of 0.2 and you can always experiment with this but I found that point two works pretty well the rest of these you don't have to worry about and
00:22:47
you can go ahead and hit test after that okay once again we have a success message and if we drill down into this choices property here which is actually an array we can keep on drilling down until we get to this little content
00:22:59
property here and we can see here's our title here's our summary delimiter here's our summary additional info main points all formatted as marked down just as we asked which is awesome and as a
00:23:12
bit of a demonstration for why we're formatting in markdown here I'm going to go ahead and hit copy value and I'm just going to go over to my notion workspace and create a brand new note uh we'll just make it in the inbox here and I'll just show you
00:23:25
that you can paste that markdown and it's going to come in with heading twos and with bolt lists just as we wanted so that's sort of why we are having an output markdown and one nice thing about pipedream is it will actually parse
00:23:37
markdown and send it to notion just like this however if we go back to one of our examples here we will notice that our summary is here and then our transcript is here and our lists are down here as I
00:23:51
mentioned before whereas our result from pipe dream has our summary has our main points our transcript is not in here and we have our title in there which we would want to put in the page title not in the page content so we are going to
00:24:05
add one small code step to this workflow which I guess technically makes it not entirely no code sorry for the click bait if I do title it no code here but it's going to make things a lot more elegant and it's going to save you a lot
00:24:19
more calls to open API because if you didn't do it this way you would have to make a call to open API AI for the title and then another one for the summary and then another one for your additional points and we don't want to do that so I'm going to add another step here I'm
00:24:32
going to choose node and then I'm going to choose run node code and I just want to choose any of the actions there because it's going to get me one of these code windows and this is really the power of pipedream make.com does not
00:24:44
allow you to write your own JavaScript code in your automations pipedream does so basically they're taking care of all of the execution environment for you they're taking care of the security the authentication all that kind of stuff
00:24:55
but you get the option of writing JavaScript or writing python if you know it and that's really what makes the platform so powerful so what we're going to do here is delete all of this code in here and we're going to just paste a code block that I have pre-written for
00:25:09
you you don't actually have to code anything and to do that I'm going to go over to the written guide once again and I'm going to go to format the title summary and transcript I will have a direct link to this section of the blog post in the description down below so
00:25:21
you can easily zoom over to it and if we scroll down a little bit we're going to find this code block all we'll have to do is copy it and paste it in but before we do that I do want to give this step a more descriptive name because right now
00:25:35
it's just named node and if you will notice in all of our exports here the path to any of the values we're referencing in future steps is steps dot what is it it's the name of this step
00:25:48
here steps dot chat if we look at this one it's going to be steps dot create underscore transcription so the export value of this block right here by default would be step dot node and then
00:26:00
if we changed it later then anytime we're referencing that steps dot node path it would break so before we even build this I'm going to change this to formatter and now it's just a little bit more descriptive from there I'll go over
00:26:13
to my blog post I will hit that copy button on that block of code I'm going to go ahead and paste it in there like so so if I hit test what I'm going to get is an object that
00:26:25
basically splits that title that summary and all those lists into three different properties and here it is here's my return value I've got title I've got the transcript as well I've got summary and I've got additional info and I guess one
00:26:38
other thing that this bit of code is doing is turning the transcript from a wall of text into basically paragraphs that have no more than three sentences each and that have a blank space between them and this is going to cause
00:26:49
pipedream to send each of these paragraphs to notion as its own text block instead of having just a super big wall of text which is another reason why you might need this code block so now that we have that we are done with the
00:27:01
coding told you it would be easy and painless all we have to do to finish this up is send this bad boy to notion so I'm going to add one final step I'm going to search for notion as the app here and then the action I want is not
00:27:14
create page it's create page from database these are two different options here so I'm assuming you have a notes database that you're working with here we're going to choose create page from database and then the first thing we
00:27:26
have to do is connect a notion account so you've probably seen me do this before if you've watched my other tutorials I'm going to authenticate with my college info geek workspace here I will hit select pages and then I think I
00:27:39
actually have to drag this out because I'm kind of zoomed in I want to make sure that the notes database that I'm going to be using is selected here because if it's not pipedream is not going to have access to it but I can see
00:27:49
here all notes cig that is the database that I want to be using is already selected so I'm going to go ahead and hit allow access and we should be good to go from there from there I need to actually choose the database that I'm going to be using here and I'll set that
00:28:02
in the parent database ID field so I can click right here and it's going to give me a whole bunch of options based on what pipe dream has access to you could also manually enter a notion database ID that is also covered in the written version here one thing that I do want to
00:28:15
note is if you don't see your database in this list it might mean that pipedream actually doesn't have access to that database so I will show you briefly how to set that access manually if you're at your database and I'm actually going to open this as a full
00:28:27
page and then go one breadcrumb to the left to access the entire database you can go over to the three dot menu in the corner here and you can see all these apps that are connected to this database now I actually have two pipe dream
00:28:40
accounts which is why I have two here but if you didn't see pipedream here you could go to add connections you could search for it and because I already have it it's not there but you would see pipe dream there you could add it to that database and
00:28:52
that's how you would make sure that that access was granted but I do have that access so I'm going to go ahead and hit my all notes cig database right here set that as my parent database ID and now we can actually specify the page content
00:29:04
that we are going to send to notion and we can see here you can use markdown syntax which is pretty cool because we can basically dynamically decide how we want our page to be laid out but for the
00:29:16
sake of time and the sake of making things easier for you I'm just going to go back over to written version I'm going to go to send the transcript and summary to notion and I'm going to find my third little code block here which I
00:29:26
will copy and which I will paste into my page content like so so once again we are dynamically referencing values here here's our summary here is a transcript heading one here's the actual transcript
00:29:40
and finally there's an additional info heading and all of our lists inside the additional info object right there now we all also want to set a page title and we might also want to set some additional page properties as well so to
00:29:52
do that we have some optional Fields down here one of which is called meta types that's going to be for your icon or page cover and the other one is going to be setting property types in your database so first let's set a meta type because I do want to actually set a page
00:30:04
icon here and I can choose icon and then from experience working with this platform I have found that it's better to also enable the property types you want before setting your meta type
00:30:18
specifically so I'm going to go ahead and choose some property types and I think for this one I at least want to set the title which is by default going to be the name property in your database I've just changed it and I also want to
00:30:30
set a property called type to a value called AI transcription so I'm going to choose those I'll choose icon and now I have the option of actually enabling these so let's go ahead and add them all
00:30:42
for emoji you can actually just search for emojis here so if I do I think it's Windows key semicolon if I can search for the robot Emoji which looks kind of like a football player on Windows I like
00:30:54
the one on Mac better for title once again we're going to dynamically reference an object and I'll actually show you how to do that so first I'll pin this configuration option here and I'll go back to the formatter success
00:31:07
message and there is our title property for which I will click copy path come back down to title and paste just like that and then for type we actually are going to get all of these different
00:31:19
select options I'm going to click AI transcription which I've set in my notion database and from here we can finally click test so to prepare ourselves let me go back over to my
00:31:31
notes area I'm going to go to my voice notes because this is a filtered view that I can show you here is going to show notes where the type is AI transcription which means that once we hit test we should see our brand new
00:31:44
transcript and summary show up in that exact View and there we go create a page successfully we have all of our exported return values from the notion API but more importantly if we go back over to notion we should have our brand new
00:31:58
transcript and summary and there it is there's our summary there's our transcript and all the way down here we have main points action items follow-up questions all that good stuff so from here all we
00:32:10
have to do is hit deploy and now our workflow is live which means it's going to be waiting for events and anytime you upload a brand new audio file to that folder in Google Drive this is going to kick off and you're going to get that
00:32:23
same transcript and summary inside an ocean now I will note again that this is the no code version and there are some limitations to the no code version namely uh whisper has a 25 megabyte file
00:32:36
limit right now however if you want to get around that if you want to say transcribe very long podcast episodes or you want to talk to your phone for an hour and transcribe that the code heavy version of this tutorial in the article
00:32:47
is going to help you get around that so if you are comfortable with at least copying and pasting code into pipedream maybe copying and paste it into chat GPT if you want to change it up and you get it explained for you then check out that
00:33:00
section of the tutorial it is just as comprehensive if not more than this no code version and that I will note is the version of this tutorial that I'm actually using myself in my own workflow I just felt that for the video the code
00:33:13
version might be a little bit too complicated for a lot of people who just want to get this working for simple voice notes and that is what this no code version is going to do for you now once again if you want to get a note-taking system just like the one I
00:33:24
have here which is going to be perfect for this AI transcription and summarization workflow you can get it inside of ultimate brain this is my truly all-in-one productivity template for notion that includes task management
00:33:36
it includes getting things done or GTD workflows if you want to use those you don't have to use those there's a project management dashboard there's goal tracking there's a recipe tracker and there is this all-in-one notes
00:33:48
dashboard where every single one of these Pages has its own dedicated page as well so it works really really well on your mobile phone and in addition to all these features ultimate brain is one of the only notion templates out there you're going to find that has active and
00:34:01
dedicated support we have a whole support team that answers every single question and we also have a community full of customers that are constantly sharing their own customizations and changes to the template so if you want to turn notion into a second brain
00:34:12
ultimate brain is the best way to do it you can get it over at Thomas jfrank.com brain there's even a discount code in the description down below so check that out if you want to get a pretty hefty discount and as always thank you so much for watching if you want to get notified
00:34:26
when I make new notion tutorials or release new templates you can always get on my notion tips newsletter as well there's going to be a link in the description down below for that and there's also a full notion fundamentals course on my website that is completely
00:34:37
free in fact I can show that to you right now we've got this fancy little sticky menu you can go to notion fundamentals and everything you need to learn notion from the ground up is on this page from page Basics to the writing system all the way down to
00:34:49
databases and there are new lessons being developed right now as well and of course right here is where you can sign up for that notion tips email newsletter if you have questions leave them in the comment section down below or hit me up on Twitter I'm Tom frankly over there
00:35:02
I'm very active on Twitter so definitely interact with me over there if you are on Twitter and once again thanks so much for watching uh click one of those videos there or there if you're looking for something else to watch but really go off and implement this workflow I
00:35:15
think this is going to be super helpful for anybody that has a lot of ideas when they're on the go so get it built for yourself try it out let me know what you think thanks for watching I'll see you in the next one foreign
End of transcript