Dictate into any app
Email, WhatsApp, Word, a browser tab: hold two keys, or tap the bubble on Android with your own keyboard still there, and talk. Long messages are fine.
Your voice.
Your device.
A voice assistant and dictation app for Windows and Android. Ask it to open apps, set reminders and find files, or talk and watch your words type themselves into any app. It all runs on your own device.
- Works offline
- No account
- Free in early access
“”Opened YouTube
Hey Berry, open SpotifyRemind me to pay rent on FridayFind my passport scanText Ali I’m on my waySet a timer for 25 minutesWhat’s 15% of 3,200?Turn on the torchCall MumOpen the Downloads folderPlay some musicNavigate to the airportGo offline
WatchMeet Berry
Meet Berry
in about a minute.
Two short films, drawn in code: the anime cut walks through Berry’s three moves; the doodle is the gentle version. Both show why your voice never has to leave your device.
Laptop and phoneTalk instead of type.
Talk instead of type.
Everywhere you write.
Hold two keys and speak. Berry writes it out, with punctuation and without the “ums”, straight into the app you’re using. On your phone, its bubble floats above your own keyboard.
To Sara Malik
Subject Homepage feedback
Typed by Berry
9:41Offline ✓
You said
…
Reminder set for 6:00 PM
How it worksTalk. It listens.
Talk. It listens.
It’s done.
Three steps, all on your own processor. Nothing is uploaded, so it works the same in airplane mode.
01
01 · Talk
Press two keys, or just say “Hey Berry”
On a PC, tap Alt + Win to ask Berry something, or hold Ctrl + Win and talk to type. On Android, say “Hey Berry” or tap the floating bubble above any keyboard.
02 · Understand
Your words are understood on your device
A neural voice detector hears when you start and stop, then Whisper turns speech into text, right on your processor. Fillers like “um” vanish, punctuation appears, and your own names and words are spelled the way you like.
03 · Done
Berry gets it done
It opens apps, sets reminders and timers, finds files, calls and texts on your phone, answers questions with a local AI model, or types what you said into whatever app you’re in. Then it answers in a natural voice.
What it doesAn assistant and a keyboard,
An assistant and a keyboard,
in one small app.
Everything Berry does runs locally. Online chat is there if you want a bigger model, and off until you turn it on.
A real assistant
Open apps, set reminders, timers and alarms, find files, call and text, control music and the torch.
Hands-free “Hey Berry”
Say the wake word and ask. Listening for it happens on your device and pauses when your screen is off.
Chat that runs locally
Ask anything and a small AI model answers on your device. Want a bigger one? Switch on online chat.
Your words, spelled right
Teach Berry names and terms once (“Ammi”, “Kubernetes”, a colleague’s name) and it gets them right.
A voice you’ll like
Replies in a natural neural voice, with several to choose from, generated on your device too.
History that forgets
See what you dictated and the minutes you saved. After the days you choose, only word counts remain.
Light on battery
Berry sleeps until you need it and unloads its models when idle, so it stays cool and quiet.
Online only if you ask
Offline is the default. Add a free Groq key in Settings if you want a larger model for chat.
Why BerryThe others need the cloud.
The others need the cloud.
Berry doesn’t.
Wispr Flow is great at dictation, but it sends your voice to its servers and charges monthly. Siri and Google Assistant live on their own platforms. Berry does both jobs on the devices you already have, without uploading a word.
| Feature | Berry | Wispr Flow | Siri / Google Assistant |
|---|---|---|---|
| Works with no internet | Everything, always | Needs a connection; transcription runs in the cloud | Some requests, on recent devices |
| Your voice stays on your device | Audio is never uploaded | Sent to their servers to transcribe | Depends on the request |
| Dictate into any app | Windows and Android | Mac, Windows, iPhone, Android | Through the system keyboard |
| Opens apps, reminders, files | Built in | Dictation and text editing | On their own platform |
| Runs on a Windows PC | Windows 10 and 11 | Yes | Siri: Apple devices only |
| Price | Free during early access, no subscription | Free tier with a weekly word limit; Pro $12–15 a month | Included with the device |
Based on each product’s public pricing and help pages, November 2026. Features change; check their sites for the latest.
PrivacyWhat you say
What you say
stays with you.
- No uploadsAudio is turned into text on your processor, then thrown away.
- No account, no trackingThe apps have no sign-in, no analytics and no ads. Berry doesn’t know who you are.
- History you controlKept in a file on your device. Phrases are deleted after the days you pick; only word counts stay.
- Open about what’s insideBuilt on open-source models (Whisper, Silero, Kokoro, Qwen), each listed with its licence.
Download
Get Berry.
Straight from here: no store, no sign-up. Installing takes about two minutes.
Windows
Version 1.1.3 · 89 MB · Windows 10 or 11, 64-bit, 8 GB RAM recommended
Download Berry-Setup.exe- Open Berry-Setup.exe. If Windows says it protected your PC, choose More info → Run anyway (Berry is new, so Windows doesn’t know it yet).
- Berry downloads its speech, voice and chat models once (0.7–2 GB), then works offline.
- Press Alt + Win to talk, or hold Ctrl + Win to dictate.
Android
Version 1.2.0 · 94 MB · Android 8.0 or newer, 64-bit phone
Download Berry.apk- Open the downloaded Berry.apk. If asked, allow your browser to install unknown apps; if Play Protect warns you, tap Install anyway.
- On Berry’s Home tab, allow the microphone and download the speech model (about 160 MB, once).
- Turn on the extras you want: the dictation bubble, “Hey Berry”, the Berry keyboard.
Free while Berry is in early access. Later, new downloads will cost a small one-time price, never a subscription. Updates appear inside the app; download the new version from this page.
Read this before you downloadWindows and Android ask a couple of extra questions for apps from outside a store. See every message and what to tap, including Play Protect and Android’s “Restricted setting”.FAQQuestions,
Questions,
answered.
Does Berry really work without internet?
Yes. After the one-time model download, speech recognition, commands, chat and Berry’s voice all run on your device. Turn on airplane mode and try it. The only optional online parts are a once-a-day update check (which you can switch off) and online chat, if you add your own Groq key.
Why does Windows warn me when I open the installer?
Windows SmartScreen warns about any new app that hasn’t been downloaded many times yet. Click “More info”, then “Run anyway”. Berry installs just for your user account and doesn’t need administrator rights. The setup guide shows every message you might see.
How do I install the Android app outside the Play Store?
Open the downloaded Berry.apk. Android asks once whether your browser may install apps; allow it, and the install continues. If Play Protect warns you, tap “More details” → “Install anyway”. The setup guide walks through each step, including the “Restricted setting” message for the dictation bubble.
Will my PC or phone keep up?
Most PCs from the last five years are fine; 8 GB of RAM is comfortable. If chat feels slow, pick the lighter chat model in Settings, or use online chat. On phones Berry needs a 64-bit Android 8 device; it unloads models when idle so it doesn’t drain the battery.
What languages does it understand?
English works best, including Indian and Pakistani accents. Whisper also understands Urdu, Hindi and many more languages; choose yours in Settings.
How do updates work?
Berry checks this site once a day for a newer version and shows “Update available”. Download it from this page and install over the old one; your settings and history stay.
Is there an iPhone, Mac or Linux version?
Not yet. Berry is made by one person, and those versions need help. If you can build for iOS, macOS or Linux, see how to contribute.
Will Berry cost money?
It’s free during early access. Later, new downloads will cost a small one-time price. There will never be a subscription, and the version you have keeps working.
ContributeHelp Berry reach
Help Berry reach
more devices.
Berry runs on Windows and Android today. iPhone, Mac and Linux versions need people who know those platforms. If that’s you, or you want to help in another way, get in touch.
Made by Abdullah Nadeem
Berry is built by one person. If you'd like to help it reach more people, whether that's a new version, testing on your phone, better speech in your language or design, I'd love to hear from you.
- Not started
iPhone and iPad
Needs someone who knows Swift and iOS keyboards. The speech models already run on Apple chips.
- Not started
Mac
Berry's speech, chat and voice are plain Python and carry over. A Mac version needs its own keys, typing into apps and a menu-bar icon.
- Not started
Linux
Same Python core as Windows. What's needed: keys and typing on X11 and Wayland, a tray icon, and a package for the main distros.
Already here
Windows and Android. Bug reports, testing on different phones and translations help these too.