Everyone's hyping free voice input tools. I just paid $144 for a year of Typeless, and my first reaction was: did I just waste money?
One quick try and I felt better. They're not even the same thing.
Recorder vs Secretary
Most voice input tools—Whisper, Apple Dictation, and the wave of new AI voice keyboards—do voice-to-text. You say something, it transcribes it, including all the "ums" and "you knows."
Typeless does voice-to-content. You can ramble at it however you want, jump from thought to thought, and out comes organized, written text. Spoken language becomes prose. Filler disappears. Paragraphs auto-structure.
One's a recorder. One's a secretary.
The difference is fundamental: transcription tools handle dictation, error correction, punctuation. Typeless takes everything you say—no matter how scattered—and automatically turns it into structured writing. You don't need to pre-organize your thoughts first.
Before I Found Typeless, I Tried These
Apple's built-in voice input: recognition was rough, mixed Chinese-English broke it entirely, punctuation was pure guessing. One day and I quit.
Local Whisper model: ran it on Mac, high latency, worse recognition than cloud, still pure transcription.
Various Chinese voice input apps: strong at Chinese recognition, but what you say is what you get—zero organizational help. Same category as basic dictation.
Ended up choosing Typeless. $144/year. Not cheap. But it solves a completely different problem.
40 Days of Real Numbers
Pulled straight from my Typeless dashboard:
- Personalization: 55.5%
- 30 hours cumulative voice input
- 256K characters total dictated
- 129 hours saved (Typeless's calculation)
- Average dictation speed: 207 characters/minute
- 84% used in the terminal—almost entirely talking to Claude Code
84% of my Claude Code conversations happen through voice.
Typing is slow. Describe a complex requirement involving strategy logic, data structures, and multiple files to modify, and typing takes forever to organize your thoughts. With Typeless, 30 seconds of talking gets the requirement crystal clear, and it structures it into something Claude understands.
The Features I Can't Live Without
Voice-to-content is the core. I talk at my computer constantly, stream-of-consciousness style, often completely illogical. Typeless outputs clean text—no verbal tics, logic sorted, paragraphs structured.
Here's what I actually said versus what it produced:
(a) I see quality content daily on X that I quote or retweet—some drives solid engagement.
(b) Through build-in-public, I create original tweets with insights.
2. Automation and publishing
(a) Automatically edit, design, format, and publish in the target platform's style.
(b) Hand the entire pipeline from sourcing to publishing to the bot.
If phase one succeeds, next we can productize it—MCP or Skill for agent-oriented products."
You speak human, it gives you prose.
The personalization dictionary is also gold. Over 40 days it auto-learned 80+ of my common terms: H12, H36, Codex, OpenClaw, PnL, Polymarket... Professional jargon it didn't know at first, but after I said it enough, it auto-added to the dictionary.
Manual adds work too. Strategy numbers like H43, H38—if it doesn't pick them up after a few tries, add them manually and it remembers forever.
Mixed language handling is solid too. My natural speech is code-switched between Chinese and English. Typeless doesn't try to translate technical terms into the wrong language like Apple's dictation does.
Privacy: I Checked the Local Database
Positives covered, now the uncomfortable part.
Someone reverse-engineered Typeless earlier, so I checked my local database. Inside:
- Every word I said (plain text)
- Current app name
- Window title
- URLs I'm browsing
- Even which messaging contacts I'm chatting with
A voice input tool collecting way more than just audio.
You can verify yourself:
sqlite3 ~/Library/Application\ Support/Typeless/typeless.db "SELECT refined_text, focused_app_name, focused_app_window_web_url FROM history LIMIT 5;"
Knowing all that, I kept using it. 30 hours of voice input, Typeless says it saved me 129 hours. Switching back to typing would cost roughly double the time. Time for privacy—I picked time.
If privacy matters to you, a few tips:
- Switch back to keyboard for passwords, private keys, sensitive credentials
- Clear history regularly—settings let you adjust retention
- Turn off Typeless when not in use, don't let it run in the background
- Run that query above before deciding—know what it's actually storing
Free Alternatives Coming—Switch?
Nope.
The workflows underneath are completely different. The operating logic and cost structure are totally different. Free products directly replacing paid ones doesn't happen that easily.
Sure, free tools have their place, especially on mobile. But I'm using Typeless everywhere now—coding, talking to AI, writing docs, replying to messages. The habit's set in. Typing feels slow by comparison.
Different categories. Not competitors.
About Annual Billing
A friend said "in the AI era, don't pay annual for SaaS—iteration moves too fast." Fair point. The "voice-to-content" capability Typeless offers will eventually become standard, maybe free.
But right now it saves me real time every single day. $144/year, under $0.40 a day. When a free tool can match it, I'll switch. Until then, no regrets.
Want to try? New users get 30 days of free Pro, unlimited. That's enough to decide if it's worth it. After that there's a free tier (8000 words/week) for light daily use.