This is the full transcript of Free, local meeting transcription. Better than Granola, Fireflies & others, published on YouTube by The Next New Thing. Every paragraph carries the moment it was spoken, so you can click any line to jump straight to that point in the video, search the whole thing for a word, or copy it out.
0:00Look at this, a free meeting note taker that sits on your computer so everything is private as it transcribes and it also doesn't have one of those weird AI bots that sits in the meeting with you. There are a lot of benefits and a couple of pitfalls that we want to talk about. Let's get into it. Presented by Zapier,
0:17the AI automation company. [music] Nick, what are we looking at here? What is this? So, this is a meeting I had this morning, hour-long meeting for my real estate team and I was able to have this join to the meeting, transcribe everything for me, also provide a little summary. That way I know everything that happened without, you know, in case I get distracted or something like that. It's all local and it adds timestamps. I really like
0:42it. Meaning like it doesn't have to go into the cloud to transcribe, doesn't go into the cloud for anything, all happens on your computer and at the end you got that summary on the right. Yep. And just like you kind of saw at the beginning, there is like a live transcription mode. I do find that if people are continuously talking, it takes a minute to process, but if you're looking for something that's like a meeting note taker that takes the notes while you're in the meeting and then at the end you have a full transcription, I
1:07think this is this is really good. I had a huge problem with this, by the way. We we did in our company where someone was talking about someone else at the company and the notes were kind of shared in public and so it was it was shared and I it's really tough to come back from something like that. I also feel bad because my note taker goes on all the time including when I have a conversation with my urologist, which is kind of weird, but I I do and I want to know it so they can find out about it later, but I don't want the whole world
1:32to know it or anyone in the world to know it. So, I like that this stuff is all on the computer. Let's take a look at the settings because I think that explains a lot of what's going on here. So, what I like about this is this saves everything, the audios and the transcription locally to your machine. When you pop into settings, you can see I just have mine saved in like my music folder, but you can save it to any folder you want and then like point your AI or your agent added in the future to maybe pull to-do lists or activity or just maybe give a summary of like, "Hey,
1:59what did we do 3 weeks ago?" Um you can point them at whatever folder cuz all this is sitting locally on your machine. The other thing that's kind of cool is that you have different options for transcription. So, I'm running this on a laptop. It does not have a GPU, does not have crazy hardware. I'm able to run Parakeet. It is pretty accurate. There are a couple of problems we we can maybe talk about that in a second. Um but if you have a little bit of extra hardware, I do recommend maybe running Whisper. Um but you do need some extra RAM and and
2:28maybe a a lightweight GPU. Yeah, full disclosure, the intro took us a couple of times because I'm such a freaking fast talker, especially in the intros, cuz I want everyone hopped up to pay attention. And the transcript wasn't picking up everything that I had said. And so we looked into it and Parakeet and Whisper, which are the lighter options, are good, but if someone talks super fast, it will miss some things. Whisper medium or Whisper for large V3 in is the ideal model to use for it. And for that, you need 16 GB of RAM at least. You need discrete GPU or Apple Silicon. That's what it says on
3:06their on their site. Having said that, I think for most people for most Once I stop stop talking like a pitchman trying to get people to watch our video and instead talk like a regular person, it picked up on everything. In fact, we can show that in a moment as we talk. But let's go into the rest of it. So, what
3:22are the other settings? So, you have the transcription settings. This is like the straight transcript while you're talking in like the final product. But sometimes, like for example, the meeting I had this morning is an hour long. I might need a summary of it at the end. I don't need to reread this whole transcript. So, you do have a whole separate set of models to provide
3:38summaries. Um you have like the default, it's Gemma 3, uh 4B, 1B. You also have QWEN. Um and you can use any of these to download and um do the transcription. But you can also add an API if you want this like professionally. I want like the highest quality transcript or summary. You can send this to like even a llama or open
4:01AI, Claude, anyone else. Okay, and then let's look at the other settings up at the top. And then you this is kind of like a beta They have I've used this a couple times actually. So, I've fed this also like video files. So, I have like a large set of of old video files I don't have transcripts for. Um I was able to point this at those video files and get transcripts off of them. Um which I we can go over here, too. I have some examples. Uh like for example, this was just a real short video. I was just testing it out, but uh it gives you a little transcript of
4:32your video files, which is also kind of nice. Did you happen to give it a video file with two different speakers? No, so that is one thing I noticed. Um like in my meeting this morning, you can see on the transcript it does not pull the speakers. It just pulls timestamps in for like when it kind of breaks. But, in the actual summary, it will pull the speakers' names. So, you can see Gary here, Nora's there. Um and then this was kind of like the main speaker. So, on the summary, it will give you names, but
5:00on the transcript, it does not. That's really disappointing, actually. Um because you do want to know who said what within the transcript. Yep. You kind of want even if it just said person one, person two, would be nice. Right. Right. It doesn't need to actually check the the meeting information for that. Okay, so that's that's one of the issues. I do like that you can see the transcript in real time. Um you know, why don't you just hit record right now so people can see we're now using not Zoom, not any anything that has the ability to let an
5:31AI agent in. I'll just talk into it and people can see how fast it works. And one of my issues is that it doesn't happen immediately. Yep. It takes a second to process. Yeah. Now, it's not uh-huh. The other thing that I kind of like about this, too, is like personally in my business, I do a lot of like face-to-face meetings with clients. I have my laptop there as we're going over things, and I've started to test this out a little bit by actually recording those conversations with them so I can go back and kind of reference it. I give everyone more or less the same speech,
6:05but sometimes questions come up, and then I can add this and tag it into my CRM later. Uh but that comes back to just pointing the agent at the file and then just maybe have them pull those to be list out of the transcript at the end. Yeah, I noticed that it still didn't
6:20finish and it's listening. If you give it a second, it'll process. Now, for me, that actually that's a that's a real problem for me because I do want to see immediately what people see, almost like subtitles, especially as I talk to people who are like not English first speakers. Um but I do like how all the transcripts sit on my computer. I like how it has the audio recordings on my computer for me to work with. I think for somebody who doesn't need that immediacy of almost like closed captioning on the meeting, this is very helpful. And if you have a really fast computer, uh one with with it's not about fast. If
7:00you have the computer with the specs that we talked about earlier, I think that you'll get really solid results out of it. Um and what I like also is that like Grainola and Grain also has this feature, it's on your desktop. It's not in the meeting, and I find that it's kind of weird to have a meeting note taker in the meeting. And frankly, some people either don't know how to let it
7:20in or don't let it in. Um And then you I'm going to keep talking so people can see it on the screen. I also find that this that this called Meetly, and we'll have a link to it below. Meetly also um will work with FaceTime video, FaceTime audio. If you've got a computer where you you've got the ability to take calls, it it work with that and it'll transcribe the calls and record the
7:43calls, which is helpful. Yep. Uh And any any service, even Slack and Discord and anything else that you can talk into. Basically, it takes anything that's hitting I've noticed like the sound card and instead of like outputting it to my ear, it's grabbing a transcription of that. Which like you said, I like that too cuz it's cross-compatible across all apps, which I didn't fully understand at first, but once I did, I think that's real cool because it just grabs that
8:07output and then transcribes it. All right, it is free. It is open source. We actually the reason that this came up is I did a video showing the top 10 repos of the week. This was one of the top 10. It really is a very popular. You can go and grab the code yourself. And Nick said, "I'm going to try it and
8:26we'll talk about how well it works." What would you rate this? Um it depends what you're looking for. If you're looking for live transcription, I don't think it's the one for you. If you're looking for something as like an AI note-taker, you can run locally, especially if you're doing lots of meetings, probably four out of five. Transcription sometimes has some errors, but I only expect it's going to get better over time as models become smaller and and stuff just works a little bit
8:51better. And they do have a paid version where they will actually auto-detect meetings where they will where they'll do more of the processing for you and it's pretty inexpensive. We're looking at $10 a month uh to use it. But again, the free version is available for you. If you like it, we'll have a link to it below. Actually, we should have a link to both of them. How to get it from their website, how to pay if you want for more and also the GitHub repo. So, if you just want to throw Claude at it and say, "Do this, but make these
9:20changes." Claude can take it and and work on that. All right, Nick. Thanks. And we'll have a link of course to Nick if anyone wants to connect with you. Perfect. Thank you. AI is allowing people to create all kinds of cool apps. You just saw one and I've got another that I think you'll like. It's a free dictation app that will allow you to dictate into your computer. It's available on Mac and now,
9:40as of today, also available for Windows. Click the link below to see how well it works. See you there.
No line in this video contains that word.
Where these words come from. This is the caption track YouTube holds for this video, written automatically by YouTube rather than by the creator. We read it, tidied the line breaks and laid it out so it can be read. The plain text version is at https://viewrankai.com/tools/youtube-transcript/GB6S5cUqZrk.txt.
All rights in this video belong to The Next New Thing. Watch it on YouTube. If this is your video and you would rather this page did not exist, tell us and we will remove it.