3 features that make Google Pixel’s “context-aware” voice typing better than any keyboard


You probably already know Gboard offers a great voice typing experience. What you might not know is that on a Pixel device, Gboard also supports voice editing. You can take a messy voice draft and use commands to make precise edits, formalize it, or rewrite it—all without touching the screen.

Here are three core features that make Gboard’s voice typing on the Pixel better than any other keyboard.

Pixel 10

Brand

Google

SoC

Google Tensor G5

Display

6.3-inch Actua OLED, 20:9

RAM

12 GB RAM

Storage

128 GB / 256 GB

Battery

4970mAh

Looking to upgrade to a Pixel but not sure if you need all the bells and whistles of the more expensive models? You won’t be disappointed with the standard Pixel 10 model. Coming in striking colors, Gemini features, and seven years of updates, you can’t go wrong with this purchase.


Use your voice to insert punctuation

Stop poking at your screen just to add a comma

Gboard on Pixel already does a solid job of placing punctuation automatically as you speak. Most of the time, you don’t have to think about it—it just figures out where the commas and periods go.

However, it’s not perfect. Sometimes it misses an obvious comma, and other times you might want to deliberately place punctuation in a specific spot. In those cases, you can simply say the punctuation out loud: “comma,” “period,” “question mark,” “exclamation point,” or “new line.” Based on your pauses and speech patterns, Gboard can usually tell whether you mean the word “exclamation” or the punctuation—and it’s very accurate in my experience.

Beyond punctuation, you can also use your voice to insert emojis into your message. Instead of pausing to find the right emoji, just say—“crying laughing emoji” or “fire emoji”—and it appears inline. Just don’t do this around other people. Saying “laugh emoji” four times in a row is a quick way to get some awkward looks.


A selection of Emoji on an blue background.


Stop using these emojis wrong: 16 emojis people constantly misunderstand

Official names tell one story. Real usage tells another. Here are the emojis most often misread, plus what people actually mean today.

Hands-free text editing

From voice typing to voice editing

If mistakes happen when you type by hand, they’re even more likely when you type by voice. Thankfully, Gboard lets you fix those mistakes using voice commands—so you don’t have to pick up your phone and switch to the keyboard. For example, if the last word you dictated was wrong, you can say “delete last word” to remove it. If the entire transcription is a mess, just say “clear” to wipe it. And if everything looks perfect, say “send” to fire off the message.

The more powerful layer is detailed editing commands which are available on Pixel 8 and later. These let you make precise edits entirely by voice. For example, say you dictated:

“John went to the cafeteria.”

But you meant “Johnny” instead of “John.” You can say, “Change John to Johnny” or “Replace John with Johnny,” and Gboard will fix it automatically. You can also say “delete John” to remove that word. You can delete specific words, capitalize text, and even spell out unusual names character by character. It might sound fiddly at first, but once you get used to it, you can compose and edit an entire message without ever touching your phone.

Here’s a list of all the commands:

  • Delete last word: Deletes the most recently spoken/typed word.
  • Delete [word]: Deletes the most recent instance of the specific word from the text.
  • Clear: Deletes the last sentence. Example: “Good morning. How are you?” > say “Clear” > “Good morning.”
  • Clear all: Clears the entire text field. Example: “Meeting at 5pm tomorrow” > say “Clear all” > empty field
  • Send: Sends the current message.
  • Stop: Turns off the microphone.
  • Insert before/after [word]: Inserts a word before or after a specified word. Example: “I want coffee and tea” > say “Insert hot before coffee” > “I want hot coffee and tea”
  • Change [word] to [word]: Replaces a specific word. Example: “Let’s meet on Monday” > say “Change Monday to Friday” > “Let’s meet on Friday”
  • Spell [word] as [letters]: Forces a custom spelling. Example: Say “The new app is called Nate. [pause] Spell Nate as K-N-A-T-E” > “The new app is called KNATE”
  • Capitalize [word]: Capitalizes a specific word. Example: “meeting with dr. Sharma” > say “Capitalize dr.” > “meeting with Dr. Sharma”
  • Lowercase [word]: Lowercases a capitalized word. Example: “Send this to HR Department” > say “Lowercase Department” > “Send this to HR department”


Holding hand because of writs pain over a keyboard.


I’m a professional writer, but I barely touch my keyboard—here’s what I use instead

Writing is about sharing your thoughts—not using a particular tool!

Proofread and rewrite entire messages with your voice

Just say “fix it”

This feature is available on Pixel 9 and later (excluding the 9a), and it’s where voice typing starts to feel genuinely powerful—and like a genuine keyboard replacement. After you dictate a message, you can say “proofread this” or “fix this,” and the keyboard’s built-in AI will clean it up for you. It corrects typos, grammar mistakes, and punctuation issues, leaving you with a polished, readable message.

As you’d expect, this completely streamlines the editing process. You don’t have to worry about editing mistranscriptions, missed punctuation, or filler words—all of it is handled for you. The main limitation is proper nouns the AI doesn’t recognize—those are usually the only things you’ll need to fix manually.

You can also use this feature to rewrite your message. If your voice drafts tend to be too casual to send to your boss or colleagues, you can say “formalize the text” to make it more professional. You can go the other way, too—make your message more casual, friendly, or even add emojis. You can also expand a short message or shorten a long one.

When you use the rewriter, a new dialog box appears with multiple versions of your message. You can cycle through them by saying “next” or “previous.” Once you find one you like, say “use this,” and it will replace the original text.

There are still some hiccups

It’s not perfect—but it’s almost perfect

Google Gboard illustration with keyboard layout and floating emojis around. Credit: Lucas Gouveia/How-To Geek

The first time you use Gboard’s advanced voice typing features, it can feel a bit hit or miss. That’s not because the feature doesn’t work—it’s because there’s a specific way to use it. However, once you get the hang of it, the workflow feels surprisingly intuitive.


Hands typing on a Gboard keyboard with the number row highlighted, flanked by a Google logo and a disabled spellcheck icon.


I hated Gboard until I changed these settings

Your Gboard experience can be a lot more pleasant with just a few tweaks.

You need to pause for the commands to register

The most important thing to understand is the pause. You need to pause for at least a second before saying a command—otherwise, it gets typed out as regular text. For example, if you say “John said to delete the word John” without pausing, the entire sentence will be typed. But if you say “John said to [pause] delete the word John,” Gboard will interpret the command and remove “John,” leaving “said to.”

That said, you don’t need to worry about accidental commands. If Gboard isn’t sure whether you meant something as a command or as text, it defaults to typing it out and highlighting it. You can then say “apply” to execute it as a command.

The AI can only ‘see’ the last couple of sentences

Voice commands are limited to the last few sentences from the location of the text cursor. So if you say “delete John,” it removes the most recent instance of “John.” I would’ve liked to have more advanced targeting options, like “delete the second instance of John,” but it isn’t supported—at least not at the time of writing. This limitation also applies to AI features like “fix it” or rewriting, where the AI only processes the last two or three sentences.

However, it’s not a dealbreaker. You can easily tap to select the word manually and then use voice commands to edit it. It’s a viable workaround, but it breaks the hands-free workflow.


The experience is optimized for short messages

Overall, Gboard’s advanced voice typing and voice command features work best with shorter messages. If you’re dictating a couple of sentences, everything feels seamless—the AI can process the full context, and voice edits work reliably.

But if you try dictating a 500-word block and then run “fix it,” the experience becomes a lot more janky due to the limited context window. A practical workaround is to dictate in chunks—two or three sentences at a time—run “fix it,” and then continue. It’s not ideal, but it works.



Source link

Leave a Reply

Subscribe to Our Newsletter

Get our latest articles delivered straight to your inbox. No spam, we promise.

Recent Reviews


Vibe coding has taken the development world by storm—and it truly is a modern marvel to behold. The problem is, the vibe coding rush is going to leave a lot of apps broken in its wake once people move on to the next craze. At the end of the day, many of us are going to be left with apps that are broken with no fixes in sight.

A lot of vibe “coders” are really just prompt typers

And they’ve never touched a line of code

An AI robot using a computer with a prompt field on the screen. Credit: Lucas Gouveia / How-To Geek

Vibe coding made development available to the masses like never before. You can simply take an AI tool, type a prompt into a text box, and out pops an app. It probably needs some refinement, but, typically, version one is still functional whenever you’re vibe coding.

The problem comes from “developers” who have never written a line of code. They’re just using vibe coding because it’s cool or they think they can make a quick buck, but they really have no knowledge of development—or any desire to learn proper development.

Think of those types of vibe coders as people who realize they can use a calculator and online tools to solve math problems for them, so they try to build a rocket. They might be able to make something work in some way, but they’ll never reach the moon, even though they think they can.

Anyone can vibe code a prototype

But you really need to know what you’re doing to build for the long haul

For those who don’t know what they’re doing, vibe coding is a fantastic way to build a prototype. I’ve vibe coded several projects so far, and out of everything I’ve done, I’ve realized one thing—vibe coding is only as good as the person behind the keyboard. I have spent more time debugging the fruits of my vibe coding than I have actually vibe coding.

Each project that I’ve built with vibe coding could have easily been “viable” within an hour or two, sometimes even less time than that. But, to make something of actual quality, it has always taken many, many hours.

Vibe coding is definitely faster than traditional coding if you’re a one-man team, but it’s not something that is fast by any means if you’re after a quality product. The same goes for continued updates.

I’ve spent the better part of three months building a weather app for iPhone. It’s a simple app, but it also has quite a lot of complex things going on in the background.

It recently got released in the App Store—no small feat at all. But, I still get a few crash reports a week, and I’m constantly squashing bugs and working on new features for the app. This is because I’m planning on supporting the app for a long time, not just the weekend I released it, and that takes a lot more work.

Vibe coders often jump from app to app without thinking of longevity

The app was a weekend project, after all

A relaxed man lounging on an orange beanbag watches as a friendly yellow robot works on a laptop for him, while multiple red exclamation-mark warning icons float around them. Credit: Lucas Gouveia/How-To Geek | ViDI Studio/Shutterstock

I’ve seen it far too often, a vibe coder touting that they built this “complex app” in 48 hours, as if that is something to be celebrated. Sure, it’s cool that a working version of an app was up and running in two days, but how well does it work? How many bugs are still in it? Are there race conditions that cause a random crash?

My weather app has a weird race condition right now I’m tracking down. It crashes, on occasion, when opened from Spotlight on an iPhone. Not every time does that cause a crash, just sometimes.

If a vibe coder’s only goal is to build apps in short amounts of time so they can brag about how fast they built the app, they likely aren’t going to take the time to fix little things like that.

I don’t vibe code my apps that way, and I know many other vibe coders that aren’t that way—but we all started with actual coding, not typing a prompt.


Anyone can be a vibe coder, but not all vibe coders are developers

“And when everyone’s super… no one will be.” – Syndrome, The Incredibles. It might be from a kids’ movie, but it rings true in the era of vibe coding. When everyone thinks they can build an app in a weekend, everyone thinks they’re a developer.

By contrast, not every vibe coder is actually a developer, and that’s the problem. It’s hard to know if the app you’re using was built by someone who has plans to support the app long-term or not—and that’s why there’s going to be a lot of broken apps in the future.

I can see it now, the apps that people built in a weekend as a challenge will simply go without updates. While the app might work for the first few weeks or months just fine, an API update comes along and breaks the app’s compatibility. It’s at that point we’ll see who was vibe coding to build an app versus who was vibe coding just for online clout—and the sad part is, consumers will lose out more often than not with broken apps.



Source link