01.
Works in every app
Dictated today
Slack · #productCasual
Gmail · To SaraFormal
Cursor · commit messageCode
01.
Works in every app
Hold your key in Slack, Gmail, Notion or your code editor. Your words appear right where the cursor is.
Free & open-source · v1.2.0Say it, and it’s on the page.
TapWisper is AI dictation and meeting notes for Mac and Windows. Hold a key in any app and talk — your words land cleaned up, in the right tone.
- No account
- Works offline
- Mac & Windows
TapWisperThe real app, running in your browser — click the sidebar. Hold fn to dictate.
Works with the models you trust
- OpenAI
- Anthropic
- Together AI
- Soniox
- Whisper
- Llama
- Qwen
Why TapWisper
Typing, the way it should work.
Most dictation leaves you a mess to fix. TapWisper does the cleanup as you talk, so your words land ready to send.
02.
You talk. AI tidies up.
As you said it
um so the fix is merged uh deploying after lunch
The fix is merged — deploying after lunch.
02.
You talk. AI tidies up.
Filler words vanish, punctuation lands and the tone fits the app — before the text reaches the page.
03.
Your models, your machine
Speech engine
On this Mac · WhisperOffline
Cloud · your API keyBYOK
Self-hosted · OllamaLocal
03.
Your models, your machine
Run open-source models offline, bring your own OpenAI, Claude or Gemini key, or point it at your own server.
04.
Private by design
What leaves your device
No account to create
No usage tracking
History stays local
04.
Private by design
No account and no telemetry. Keys stay in your keychain and your history never leaves your device.
How it works
Hold. Talk. Let go.
Three times faster than typing, and nothing new to learn. TapWisper only listens while you hold the key.
Hold your key
fn on Mac, Ctrl + Win on Windows — in any app, any text field.
Just talk
Ramble, pause, correct yourself. Add Space to go hands-free.
Let go
Filler words vanish, punctuation lands, tone matches the app.
Slack
# product
fnHold to talk
Meeting transcription
No bot joins the call. You still get the notes.
TapWisper notices when a call starts and offers to take notes. It records your mic and the meeting audio on your computer — live transcript while you talk, a clean summary the moment you stop.
Project X review
Recording · 12:48
What did I miss?
Your microphone
Meeting audio
SummaryTranscriptMy notes
Others10:47
So if we ship phase one Monday, the migration guide has to be ready by Friday.
You10:47
I can take the guide. Sara, can you own the rollout email?
Others10:48
Yes — I'll draft it today and share it in the channel.
Sí, lo redactaré hoy y lo compartiré en el canal.
Listening
Live transcript
You and Others, separated — rename speakers, jump back to live.
Summary when you stop
Topics, decisions and a next-steps checklist with owners.
What did I miss?
A recap of the last five minutes when you drift off.
Ask about the meeting
“What did we decide about pricing?” — answered with timestamps.
Live translation
Translate each line as it's spoken, or the whole transcript after.
Your notes shape it
Jot what matters in My notes; the summary follows your lead.
Recognizes calls in
ZoomGoogle MeetMicrosoft TeamsSlack huddlesWebexFaceTimeDiscordWherebyAround
Transcripts, notes and summaries stay on your device. Export any meeting to Markdown. Let everyone know you’re recording.
How meeting notes workFeatures
What you get.
Dictation, AI cleanup, voice editing, meeting notes and more — in one free app.
Your machine. Your models.
Run open-source models on your computer, bring your own API key, or connect your own server. Free either way.
Cloud
46 models · OpenAI, Claude, Gemini, Together AI
On your device
42 open-source models · free and offline
Self-hosted
Any OpenAI-compatible server · Ollama, LM Studio, vLLM
Edit by voice.
Select any text, hold your key and say what to change. It's replaced right where it is.
Selected
I am writing to inform you that we have decided to postpone the launch until the second week of next month.
You said
“make it shorter and friendlier”
Output
Quick heads-up — we’re pushing the launch to mid next month. Thanks for bearing with us!
It learns your words.
Names, jargon and product terms the AI might miss. Add what it mishears once — it's right every time after.
tap whisper→TapWisper
cooper netties→Kubernetes
super base→Supabase
Zaid→Zeyad
Speak any language.
20+ languages with auto-detect, 99 with local Whisper. Switch mid-sentence — c’est pas un problème.
You said
“send the report to Omar, وقل له إن الاجتماع بكرة الساعة عشرة”
Output · translated to English
Send the report to Omar, and tell him the meeting is tomorrow at ten.
EnglishالعربيةEspañolFrançaisDeutsch中文日本語हिन्दी+ more
Sounds like you, in every app.
Filler words gone, punctuation fixed, and the tone set per app — casual in WhatsApp, formal in email.
Messageshey! running 10 min late, save me a seatCasual
WorkFix is merged — deploying after lunch.Casual
EmailHi Sara, thank you for the update. I'll review it today.Formal
Say less, type more.
Snippets expand a short phrase into the full text.
You said
“slash sig”
Output
Best,
Zeyad Elosherey
TapWisper · cal.com/zeyad/30min
Run anything by voice.
Triggers search the web, open apps, press shortcuts, run scripts and tile windows.
“google best coffee shops”
“open spotify”
“press command k”
“left half”
“deploy staging”
“ask ai to summarize this”
100 AI mini apps.
Translate, rephrase, format or fix grammar in one click — or add ready-made templates for emails, posts, study notes and more.
Productivity
13 templates
Meeting Summary, Daily Planner
Work
13 templates
Email Draft, Slack Update
Social Media
13 templates
LinkedIn Post, Tweet Generator
Creative Writing
13 templates
Story Starter, Song Lyrics
Education
12 templates
Study Notes, Flashcard Maker
Communication
12 templates
Thank You Note, Apology Message
Marketing
12 templates
Ad Copy Writer, Product Description
Personal
12 templates
Travel Itinerary, Meal Planner
Voice notes.
Record a thought and get a transcribed note you can title, edit and copy.
History & stats.
Every dictation saved and searchable — replay it, copy it, see your pace and time saved.
Private by default.
No account, no telemetry. Keys stay in your keychain; history stays on your device.
Free & open-source No account Works offline with local models macOS & Windows
Models
Open-source or closed. Your pick.
46 cloud models and 42 open-source models that run on your computer. Pick one for speech and one for AI cleanup — they don’t have to match. Read more about offline local models and cloud and self-hosted models.
Your key
Closed 13
Frontier models from OpenAI, Anthropic, Google and Soniox. Requests go straight to the provider.
GPT-5.2GPT-5Claude Opus 4.6Claude Sonnet 4.5Gemini 3 ProGemini 2.5 FlashGPT-5 MiniClaude Haiku 4.5
Together AI
Open-source, hosted 27
Open-weight models, run fast in the cloud with one Together AI key. No GPU needed.
Kimi K2.5DeepSeek V3.1DeepSeek R1Qwen3 235BLlama 4 MaverickGPT-OSS 120BGLM-5MiniMax M2.5
Free · offline
On your device 24
Open-source models you download once and run on your computer. Nothing leaves it.
Qwen3.5 4B2.7 GBGemma 4 E4B4.9 GBLlama 3.2 3B2.0 GBPhi-4 Mini2.5 GBgpt-oss 20B12.1 GBMistral 7B4.4 GBQwen3.5 0.8B550 MBGemma 4 E2B3.1 GB
Or bring your own server.
Any OpenAI-compatible endpoint on your network works, with any model it serves.
OllamaLM StudiovLLMLocalAIspeaches
Shortcuts
One key. That’s the interface.
Prefer another key? Presets include Ctrl + Option, Right ⌘ or Option + Space — or record any combination in Settings.
Hold to talk, release to pastefn
Go hands-freefn+Space
Stop hands-freePress it again
Edit by voiceSelect text, then hold fn
Cancel or dismissEsc
Download · v1.2.0
Start talking in a minute.
Free and open-source, no account. Setup walks you through permissions, your microphone and a speech engine — then you just talk.
FAQ
Good questions.
What is TapWisper?
TapWisper is a free, open-source AI voice dictation app for Mac and Windows. Hold a key in any app, speak, and your words are transcribed, cleaned up and pasted at your cursor. It also takes bot-free meeting notes, edits selected text by voice, and runs on cloud, local or self-hosted AI models.
Is TapWisper really free?
Yes. The app is free and open-source under the MIT licence. With cloud models you pay your provider directly for what you use. With local models it costs nothing.
Can TapWisper transcribe meetings?
Yes. It recognizes Zoom, Google Meet, Microsoft Teams, Slack huddles, Webex, FaceTime, Discord, Whereby and Around calls, records your mic and the meeting audio on your computer, shows a live transcript and writes a summary with decisions and next steps. No bot joins the call.
Do I need an API key?
Only for cloud providers. Choose Local in onboarding and TapWisper downloads an open-source model that runs on your computer — no key, no account, works offline.
Which AI models does TapWisper support?
OpenAI (GPT-5.2, GPT-5, GPT-4o Transcribe, Whisper), Anthropic Claude, Google Gemini, 27 open-weight models on Together AI, Soniox, 42 local open-source models such as Whisper, Parakeet, Qwen, Gemma and Llama, and any OpenAI-compatible server like Ollama or LM Studio.
Which apps does it work in?
Any app with a text cursor: email, chat, docs, code editors, browsers and terminals. TapWisper pastes at the cursor, and Style adjusts the tone to the app you're in.
Is TapWisper an open-source alternative to Wispr Flow?
Yes. TapWisper covers the same hold-to-talk dictation in any app, AI cleanup, edit-by-voice and bot-free meeting notes — free and open-source under the MIT licence, with offline local models and your own API keys as options.
Does TapWisper work offline?
Yes, with local models. TapWisper bundles whisper.cpp and llama.cpp and runs speech-to-text and AI cleanup entirely on your computer.
What does it need permission for?
On macOS: microphone, accessibility (to paste and listen for your shortcut) and screen recording (for meeting audio). On Windows, just the microphone. It only listens while you hold your key.
Is there a Linux version?
TapWisper runs on Linux (X11, or Wayland with wtype) — build it from the GitHub repo. Prebuilt downloads are available for macOS and Windows.