Wispr Flow

If you're not using voice dictation, you're wasting a lot of time because no matter how fast you type, you'll speak at least three times faster. But voice dictation used to be pretty bad. It made a lot of mistakes, especially when you have an accent like me. It would mix up similar sounding words. It would put punctuation at the wrong places. And when you use any nomenclature or slang, it would get this wrong all the time. And in the end, you spend more time fixing mistakes than you saved in the first place. But AI has changed the voice dictation game completely because LLMs can understand the context of what you're saying and adjust spelling and grammar accordingly. For example, when I talk about a piece of code, it will automatically write function names in camelCase. Because the AI is smart enough to understand what I'm talking about. It can also automatically correct mistakes, remove uhms and uhs and stutters, and also apply the appropriate formatting. For example, in an email, you want more formal formatting, and in chat messages, more casual with less punctuation. I tried a bunch of AI voice dictation tools, and for me, Wispr Flow is the clear winner. It's available for Windows, Mac, iOS, and soon also Android. It's fast and accurate. It has the best context detection. What exactly this means, I will show you in this video. And they have a very generous free tier. In this video, I will walk you through all the features in Wispr Flow and show you why it's the best AI voice dictation app. We will also try it in different use cases like writing emails, chat messages, and even coding. And we will see how well it works with background noise and when we actually whisper to the mic. We will also compare it to free alternatives like Windows's inbuilt voice dictation and the speech-to-text button on ChatGPT. And if you have been on the fence about getting Wispr Flow, after watching this video, you will know exactly if it's the right app for you. My name is Florian Walther and this is AI Tool Corner, where I review the latest AI software to find out which ones can actually improve our lives and businesses. I will put the link to Wispr Flow into the video description. You can download it from there. But first, let's take a look at the pricing. So, when you download Wispr Flow for free, you also get a 14-day trial for the pro mode. But they also have a very generous free tier. You get 2,000 words for free every week. And even if you exceed that, you can actually still use Wispr Flow. You just get a slower model. But from my experience, this still works really well. You also have like 95% of the features in the free version. The only missing feature I noticed is the command mode, but everything else I could still use. Nevertheless, I pay for the pro version because I use it all day every day and it makes me so much more productive. And if you get the annual subscription, you only pay 12 bucks per month, which is not a lot to be like three times more productive. And if you are a student, you can even get a 50% student discount and then it's really cheap. So, Wispr Flow is available for Windows, Mac, iOS, and soon also Android. There's a wait list right now, which I'm on. We will take a look at the mobile app later. For now, let's download it to our computer for free via this button here. Go ahead and install it on your computer. And then you should see something like this. We will go through all the features and settings step by step. But for now, let's start with the minimal necessary setup to get started as quickly as possible. We have to configure two settings. So, we click down here on the settings button. The first is the keyboard shortcut. There are two ways to use Wispr Flow: either push to talk, which means that you press a keyboard shortcut, you hold it down, you speak what you want to say and then you release it and it will go ahead and turn your speech into text. Then there's also a hands-free mode where you click a keyboard shortcut once you speak and then you press it again which ends your dictation. I almost always use push to talk and you can configure the keyboard shortcut for both of them here. Just click on this little field and press the combination on your keyboard. I use alt and the plus button on the numpad. Use something that doesn't clash with other shortcuts. There are also a bunch of other keyboard shortcuts here, for example, for command mode, which we will take a look at later. The second setting we need to configure is, of course, the microphone. For this video, I use this microphone here, which is pretty high quality. But in day-to-day life, I actually use a cheap $20 Bluetooth headset with an abysmal mic quality, and it works just as well. So, the mic doesn't have to be expensive. Really, any mic will work. We will go through the other settings later. For now, let's try it out. So, I opened Notepad. I will dictate some text and I will on purpose add some uhms and uhs and mistakes. So, I hold down the push to talk keyboard shortcut and then I dictate. Hey everyone, it's Florian Walther from the uh AI Tool Corner YouTube channel uh channel. Um uh subscribe if you haven't yet. And the output is clean and free of uhms, uhs, and obvious mistakes. This is the strength of AI voice dictation. LLMs are smart enough to understand what belongs in your dictation and what's likely a mistake and it can clean it up for you, which of course saves you a lot of time. And this works in any app because it will put the text wherever your cursor is. It's that simple. And if you put the cursor in the wrong place and it didn't properly insert your dictation, you can always find your past dictations here in the history. You can copy them from here. or remember there's also another shortcut which automatically pastes the last transcript. This is very useful so you don't have to repeat your dictation. We will compare Wispr Flow to Windows's inbuilt voice dictation later, but for now let's stress test it. First of all, what languages does it work with? They're pretty much all of them. And the coolest thing is you can even mix languages. So I speak English and German. I don't speak anything else. So I disabled auto detect. And now I can even mix German and English in the same sentence and it will still work. For example, I can comfortably mix languages in one sentence. And again, this works really well. And the spelling and grammar is impeccable. But single sentences are easy. Let's try dictating a whole email. And this is where Wispr Flow really shines. But first, I need another coffee. So, I'm going to reply to this email using only my voice. And again, I will add some uhms, uhs, stutters, and mistakes on purpose. So, please don't assume I'm stupid. This is intentional. Hey, Steven. Sure. Um, I can give you my feedback on Serif. Uh, here are a few points. Uh, one, the bot often malfunctioned and got stuck. Uh, two, the standard plan is uh too expensive for me. And three, there were too many confusing settings. Um, if you want, we can schedule a Zoom call for Friday 3 p.m. No, actually, I mean 1 p.m. and uh talk about it. Regards, Florian Walther. And now a bunch of amazing things happened. First, it formatted the email properly, and it even added a numbered list here. Okay, there's one thing that's not perfect. I would have preferred regards in a new line, but the rest is pretty much perfect. Second, remember down here I made a correction. I said 3 p.m. No, actually 1 p.m. And Wispr Flow is smart enough to completely remove the 3pm and only leave 1 p.m. here. That's what I mean when LLMs understand what you're trying to say. They can intelligently remove your mistakes while keeping the full meaning of your dictation intact. But what's even more impressive is that it spelled difficult names correctly—Stephen (with "Ph") and Serif—the app name from the email thread—which generic dictation would often get wrong. How did Wispr Flow know how to spell them? It can actually access some of the information in the active window. So, it can read the sender of this email and the email body and it will use this information to generate our dictation. Now, from my experience, it cannot access the email address, which is important for security, right? Because you don't want to send random email addresses to a third party. This is a privacy nightmare. It cannot access the email address. I'm pretty sure of that because it gets the email address wrong when I try to spell it. But it can access less sensitive information like the sender name and the body of the email. And this is very useful because then we don't have to make corrections down here because it gets the spelling right. and other voice dictation apps I tried didn't have this feature. They would usually spell Steven with a V and get Serif completely wrong. Now, if you're still worried about privacy and you don't want to send the sender name or email body to any foreign servers, for example, because you want to use Wispr Flow within your company, then you can actually disable this feature. I will show you how to do that later. AI voice dictation also perfectly handles difficult words where I often forget how to spell them, which saves me a ton of time. For example, the word conscientious is notoriously difficult to spell. I'm glad that AI can handle it for me. And again, it's perfect. It's fast. I would have probably needed at least 10 times as long to spell this by hand because I would have to look up conscientious and maybe even notoriously. But with AI voice dictation, it takes me 2 seconds. In the app, we can also change the style for different apps, which changes how our dictation is formatted. Here are different categories. For example, in an email, we want more formal formatting with more punctuation and spaces. But in a chat app, we might prefer more casual formatting with less punctuation or even very casual where it starts sentences with lowercase letters. This depends on your preferred writing style. However, I think this only works with actual apps and not with the web browser because if I open Gmail in a web browser, it doesn't detect it as email. It detects it as other. So the other style is applied whether I'm in Gmail or in Discord unless you actually install the desktop app then it will use the correct category. This is a small limitation because I think it detects the browser as one single app. So when I change other to "Excited", go back into Gmail in my web browser and spell something. Hey Steven, thanks for reaching out. Sure. Let's schedule a Zoom meeting and talk about my experience with Serif regards Florian Walther. You can see it adds more exclamation marks which is the excited style that we set here. So again this is a limitation. I think it applies other to all websites running inside your web browser. So I usually set this to formal and email as well. We will try out Wispr Flow in more use cases in just a minute. But first let's take a look at a dictionary. Of course, there are still some difficult words that Wispr Flow will get wrong, like proper nouns and brand names and certain kinds of slang. You can add these words here to the dictionary. Just type in your word, save it, and the next time Wispr Flow will get it right. Or you can use this correct a misspelling feature, which automatically changes a word into another one. For example, I could say every time I say ASAP, actually turn this into a as soon as possible. And now Wispr Flow will use that as a replacement. So when I say I will solve the problem ASAP, it turns ASAP into as soon as possible, which again makes it easier for us to dictate this. But you can see that there are words that have a little sparkle emoji next to them. Those were actually automatically added by Wispr Flow. And from what I know, no other voice dictation app has this feature. This happens when you do a dictation and then correct a mistake yourself. Let's try it with an example. I want to talk to Andre about a new idea. And now I change Andre to a different spelling. And this will cause Wispr Flow to automatically add this word to the dictionary. And there we go. As we can see from this message down here, it added Andre automatically to the dictionary. This is useful. It saves you a lot of time because you don't have to add all of these words manually. But if for some reason you don't like this, you can disable this in the settings here under system extras auto add to dictionary. You can turn this off, but I don't know why you would do that. Let's try out Wispr Flow in a chat app in Discord. Hey Chrome soldier and monkey drenchual, how are you guys today? And as you can see, it spelled both of these usernames correctly because it read them from the window from the context. And I can't emphasize enough how useful this is because if we have to go in here and make corrections all the time, we are not really more efficient. But with Wispr Flow, you don't have this problem. If we want to tag people, we can add these texts to the dictionary. For example, at everyone, just click up here on add new. Insert the word that you want to use and click on add word. And then we can use it down here. for example, hey at everyone who's going to join the meeting tomorrow and it uses the tag here which is useful for social apps and Wispr Flow also detects when you are in the middle of a sentence and it capitalizes it accordingly. So when I go before the question mark and I say morning at 11:00 a.m. it starts morning with a lowercase M because it knows that we are in the middle of a sentence. And again, this is very useful because now we don't have to correct this manually. Those are the little details where Wispr Flow really outshines other AI voice dictation apps. And this adds up to a lot of time savings every day. Wispr Flow is also the best voice dictation app for coding. Let me show you an example. Go into the MatchHeader.tsx file and remove the nominateSupplier function. So two things happened here. one, it spelled both the file name and the function name correctly and it even used camel case for the function name because that's how you write functions. But not only that, it even tagged the file here in Cursor which means that it added this link to the file which is necessary for the AI to know which file to change. This feature is unique to Wispr Flow. I'm not aware of any other voice dictation app that can do that. We can enable this here in the settings under Vibe coding, variable recognition and file tagging in cursor and Windsurf. But I have to say in reality this only has limited usefulness. It works really well when we focus on a single file like we did here. But when we want to change a bunch of different files, it often tags the wrong one or it doesn't tag it at all. Also, it's very common that you have many files with the same name like index.tsx and it's pretty much impossible to tell the AI which one you want to tag specifically. So it will always take the currently opened one. So for coding, I still often find myself typing by hand, but Wispr Flow still saves me a little bit of time here. Next, let's take a look at the snippets feature. Here we can insert text snippets that we can then trigger with a simple phrase or word. And those can be long. For example, I can put a whole email template here and say, I don't know, my email template. I can add the snippet. And now when I dictate my email template, it inserts the whole template. You can also add shorter ones. For example, my YouTube link. This one is really useful. And then I can say, "Hey everyone, please subscribe to my YouTube channel. Add my YouTube link." And it automatically replaces my YouTube link for the actual link. Again, super useful. You can also prepare AI prompts in here or at your social media links. Now, the name Wispr Flow is not picked randomly. It actually works when you whisper into the mic and with noise in the background. So, I have prepared the sound of a noisy office in the background. I'm going to turn this on and then I'm going to spell something while whispering into the mic and we will see if it still gets it right. Okay, let's go. Wispr Flow even works with a noisy crowd in the background—like an office or a coffee shop. Again, it nailed it—even when I could barely understand myself. And this is also useful to avoid strain on your voice from speaking all day. For example, I have a cold right now and it's actually more comfortable for me to whisper into the microphone. Next, let's take a look at command mode. And I think this is the only feature we have seen so far that's only available with a pro subscription, but it's very useful. We can enable this in the settings under experimental command mode. And we have to set a keyboard shortcut for it. Here under keyboard shortcuts, command mode. I set this to alt shift and plus. And with command mode, we can do two things. For one, we can highlight some existing text and let AI modify it. For example, I can say rewrite this into a professional sounding email with proper formatting and empty lines and add regards Florian Walther to the end. Now, Wispr Flow rewrites this selected text into a properly formatted email. There we go. The second thing we can use this for is highlight some text and then ask Perplexity about it. For example, who is this? So it automatically opened Perplexity in another browser window and it sent this message. Who is this Florian? Obviously those are all impostors and the real Florian Walther is this one here. But we don't even have to highlight text. We can also just say something like what the hell is looksmaxxing and again it will open Perplexity and send a search request. So this is the command mode that is only available with a pro subscription in Wispr Flow. Now the only feature I find missing in Wispr Flow is an option to insert a custom prompt in VoiceType which is another AI voice dictation app. For example, you have this custom system prompt where you can add your own custom instructions for the AI which would be taken into account for every dictation. It would be cool if Wispr Flow had something like that but currently it doesn't. Now, let's compare Wispr Flow to Windows's inbuilt voice dictation. I will try to dictate the same email as before, but this time I press the Windows key and H, which starts Windows's own voice dictation. Hey, Steven. Um, sure I can uh give you my feedback on Serif. Uh, here are a few points. One, the bot often malfunctioned and got stuck. Uh, two, the standard plan is too expensive for me. And three, there were too many confusing settings. If you want, uh, we can schedule a Zoom call for Friday, 3:00 p.m., no, actually, I mean 1:00 p.m., and talk about it. Regards, Florian Walther. As you can see, this is terrible. Literally unusable. There's no formatting. Everything is spelled wrong. It has no context awareness. It got Steven wrong. Now it got Serif mostly right, but “Serif” started with a lowercase letter. And apparently my name is Florian Walther. And it even added extra letters here. That's another mistake. So this is no comparison to Wispr Flow, obviously. Now I'm using Windows 10 here. And I think the one in Windows 11 is a bit better, but still not as good as Wispr Flow, and it doesn't have all these special features. Now, this little dictate button here in ChatGPT is actually pretty good because they also use AI to clean up your input. Let's try this out as well. So, I click this button. Allow the microphone. Can you tell me what um makes uh Wispr Flow the best AI voice dictation tool? So, it removed the uhms and uhs. It cleaned up our input, but it spelled Wispr Flow wrong. And there's no dictionary here we can use to add special words. It has no context awareness. And of course the other problem is that we can use this one here only inside ChatGPT and not in all apps like Wispr Flow. So this is not a replacement for Wispr Flow. In Wispr Flow you can even share your snippets and settings with team members. You can invite your team members here and share snippets the dictionary unify your billing admin controls and so on. The other AI voice dictation apps I've tried don't have these team sharing features, but I haven't tried them out myself. And when you use Wispr Flow in a company, of course, data and privacy becomes an important topic. So, you have some settings for this as well. You can basically always activate privacy mode, which means that your dictations won't be used to train any models. I keep this on all the time. You can disable context awareness which is the feature that reads some of the information of the active window like the email sender or the email body which are then used to make your dictations more exact. I keep this on but you can disable this if you are worried about privacy. If you need HIPAA compliance which I think is for medical situations like doctors then you can enable this here and they also offer SOC 2 type 2 compliance in the enterprise plan. Lastly, let's take a look at the mobile app. So, currently the Wispr Flow app is only available on iOS. The Android version is coming soon, but it's not out yet. So, I couldn't try this out because I don't have an iPhone because I'm poor. But apparently, this is just a keyboard with a button that turns your speech into perfect text, exactly like we saw on desktop. It has 4.8 stars, which seems pretty rare for a mobile app, so I assume it works really well. And of course, it's nice to have the same subscription for desktop and mobile because then you don't have to pay for multiple apps. As soon as the Android version comes out, I'm going to give this a try. If you want to see another video on the Android app, let me know in the comments below and I will make one. And also leave a like on this video so I know that there's interest. And I will also leave a link to the Android waitlist in the video description below if you want to be one of the first to get access to this. Again, I will put the link to Wispr Flow into the video description below. You can try this out for free. The free tier is amazing. And if you want to see how Wispr Flow compares to other AI voice dictation apps, I have a really elaborate comparison video which I will put here into this card. Check it out next. Because even though Wispr Flow is great, maybe there is another app that's an even better fit for you. Subscribe to the channel for more AI tool reviews in the future. Then I hope I see you in the next video. Take care.