I type fast enough.
On a standard keyboard I hit around 80–90 wpm. That’s fine for writing code and replies. But over the last couple of years I’ve noticed a ceiling: my hands can’t keep up with my thoughts. A commit message, an architecture note, or an entire email forms in my head faster than my fingers can deliver it.
Worse, the keyboard becomes a wall in other places:
- On my phone, composing a long answer is a chore.
- In meetings, I want to capture a thought without stopping the conversation.
- Late at night, I’d rather not hammer keys while others sleep.
So I started taking voice input seriously. I tried the built-in dictation on macOS,讯飞, Doubao input, and finally Typeless. After two months, it’s the one I’m willing to pay for.
This is not a spec sheet. It’s what worked, what didn’t, and why I think other developers and writers should try it. There’s a Typeless referral link at the end: sign up through it and you’ll get $5 in free Pro credit.
What Typeless actually is
Typeless is a cross-platform AI voice keyboard. In plain English: it lets you speak into any text field on macOS, Windows, iOS, or Android, and turns your speech into clean, usable text.
The “any text field” part matters. Typeless is not a feature inside Notion or VS Code. It is a system-level input tool. I’ve used it in:
- VS Code commit messages and comments
- Cursor agent prompts
- Notion, Obsidian, and Google Docs
- WeChat, Telegram, Slack
- Gmail, Outlook, Linear
- Browser forms and search boxes
If you can place a cursor there, Typeless can probably dictate into it.
Five things that made me stick with it
1. It cleans up your speech, not just transcribes it
This is the biggest difference between Typeless and ordinary dictation.
Most voice tools give you exactly what you said, filler words and all:
“Uhh… this function mainly handles, you know, the user input string, and then it returns, like, a result object.”
Typeless strips the junk. It removes fillers, repetitions, and self-corrections, and keeps only your final intent. It also auto-formats spoken lists, steps, and bullet points.
The cleanup is not perfect, but it removes the cleanup tax that usually makes voice input more trouble than it’s worth.
2. It’s meaningfully faster when dumping ideas
Typeless advertises 220 wpm versus ~45 wpm for average typing. My own sense matches that benchmark.
Two places where the speed boost is obvious:
- Long-form drafts: I can speak an outline or a section and only edit transitions afterward.
- AI agent prompts: Giving a detailed multi-step instruction to Claude Code or Cursor by voice is far smoother than typing it.
3. “Ask Anything” edits selected text by voice
Select any text, hit a shortcut, and speak a command:
- “Make this more formal.”
- “Translate to English.”
- “Shorten to 50 words.”
- “Turn this into bullet points.”
- “Add a short summary.”
For docs, emails, and meeting notes, this combines voice input and AI rewriting into a single shortcut. No tab switching, no copy-paste.
4. Translation and 100+ languages
I write in both Chinese and English. Typeless handles code-switching better than any built-in dictation I’ve used. I can speak a sentence mixed with Chinese and English terms, and it usually gets both parts right.
The translation feature is also useful: speak in one language, get clean text in another.
5. Privacy claims that don’t sound hand-wavy
Voice input is always a little uncomfortable. Typeless states:
- Zero cloud data retention
- No training on your data
- History stored locally
I can’t independently verify the backend, but the product messaging is clearer than most competitors. That matters when I dictate code comments and internal notes.
Who Typeless is for
| Audience | Why it fits |
|---|---|
| Developers | Commit messages, comments, agent prompts, docs |
| Technical writers | Long drafts, voice-to-outline, quick rewrites |
| Heavy email users | Fast first drafts with tone adjustments |
| Multilingual users | Mixed-language input and real-time translation |
| Mobile-first workers | Much better than phone keyboards for long messages |
Who should maybe skip it:
- People who mostly write in open offices or cafes
- Users who only send a few short messages per day
- Anyone allergic to subscription tools
Pricing: Free vs Pro
- Free: $0, 8,000 words per week, core features included.
- Pro: $12/month billed yearly or $30/month billed monthly, unlimited words, better accuracy, priority access, team features.
- Enterprise: Custom pricing.
The free tier is genuinely useful for testing. My recommendation: use it for one to two weeks. If you find yourself reaching for the shortcut every day, Pro starts to make sense.
How to get started
- Install the client for your main device — desktop first if you write long text.
- Pick a shortcut and make it muscle memory. I use the default press-and-hold.
- Start with idea-dumping: don’t worry about formatting, just speak.
- Try the voice-editing shortcuts on a block of rough text.
- Add technical terms and names to your personal dictionary.
Honest downsides
- Location matters: You can’t comfortably dictate in a shared open office.
- Casual Chinese internet slang: Fine for ordinary speech, but very regional or meme-heavy phrases can slip.
- Pro price: Reasonable for daily users, too much for occasional use.
- Setup curve: Spending 30 minutes on shortcuts and dictionary pays off, but it’s not instant magic.
Try it with $5 free Pro credit
If you want to test it, here is my referral link:
Download Typeless for free and get $5 Typeless Pro credit →
It costs you nothing extra. After you sign up and hit 2,000 words of input, we both get $5 in Pro credit.
Tip: The free tier gives you 8,000 words per week, which is more than enough to reach the 2,000-word referral target.
Final thoughts
Typeless is not a tool that instantly changes your life. It’s a tool that quietly removes friction once you build the habit.
The value is not replacing the keyboard everywhere. It’s removing the lag between “I already know what I want to say” and “it’s on the screen.” For developers, writers, and heavy AI-tool users, that lag is worth eliminating.
If you’ve been curious about voice input, give it 2,000 words. That’s enough to tell whether it fits your workflow — and if it does, the $5 credit gets you a nice head start on Pro.