Cross-Language Voice Input
Double-tap right ⌘ to switch between native dictation and instant voice translation across Simplified/Traditional Chinese, English, Japanese, Korean, French, Spanish, and German.
Speak naturally.
Write in the language you need.
Saylane is a native macOS input method: hold right ⌥ to dictate or speak Chinese and output idiomatic English or Japanese directly into Slack, Cursor, Mail, or Discord; type Rime Pinyin when quiet; press ⌥T to translate any screen region in place. All recognition runs on your Mac.
01 · Hold Right ⌥
Speak in your mother tongue
A compact 38-bar waveform opens without stealing focus from Slack, Cursor, Mail, or Linear. Press Esc anytime to cancel cleanly.
You speak (Mandarin)“把周五的设计评审改到下午三点，顺便把 PR 链接发群里。”
02 · Release Key
Two-stage on-device accuracy
Live preview updates every 0.26s while speaking. On release, local Qwen3-ASR 0.6B or SenseVoiceSmall refines the transcript in ~0.3s and translates on device.
Committed into Slack / Cursor“Let's move Friday's design review to 3 p.m. and drop the PR link in the channel.”
03 · Double-tap Right ⌘
Four modes + Rime Pinyin
Cycle A→A, A→B, B→A, and B→B across Chinese, English, Japanese, Korean, French, Spanish, and German — or just type Pinyin on the keyboard.
Same input sourceNo switching macOS input methods between typing and speaking.
Live words as you speak.
Sharper text when you let go.
The system recognizer streams a live draft every quarter second. When you release the key, a local model rewrites the finished sentence from the full utterance.
- Numbers, acronyms, and model names land the way engineers write them:
M1,36G,ChatGPT,0.8.1. - Fix a misspelled name once or twice and Saylane remembers your spelling locally on this Mac.
- 0.26s
- Live preview cadence
- 0.3s
- Release to final commit
- ~2/100
- Character error rate (Qwen3-ASR 0.6B)
Measured across 150 real Mandarin FLEURS recordings on Mac (vs ~5/100 for the default system recognizer). Methodology ›
Four capabilities inside one lightweight menu-bar input method.
Translate Anywhere (⌥T)
Drag over any window, PDF, image, or video frame. Saylane measures stroke size (0.9% median font-size error), weight, and color, replacing text in place while preserving backgrounds.
Built-in Rime Pinyin
Backed by librime in a separate crash-isolated InputMethodKit process with whole-sentence composition, fuzzy Pinyin, mixed English candidates, and an optional 409 MB language model.
Dictation & Pinyin stay
free forever.
Local hold-to-speak dictation, hold-to-speak voice translation, and Rime Pinyin require no account, no API key, and have zero weekly word limits.
Translate Anywhere (⌥T) and optional AI proofreading & context screen translation are included in the iTool Yearly Membership: $26.49/year (¥99/year), unlocking all 7 iTool apps.
Compare that with $144–$180/year for cloud-only dictation subscriptions. Works offline for up to 7 days after each membership check.
Get iTool Yearly ($26.49/yr · All 7 Apps)Before you install
What Mac do I need?
A Mac with Apple Silicon (M1 or newer) running macOS 26 or later. Intel Macs and earlier macOS releases are not supported.
Is voice dictation really free with no word limit?
Yes. Hold-to-speak dictation, hold-to-speak voice translation (A→B), and Rime Pinyin typing are completely free with no account, no API key, and no weekly word cap.
What does the iTool yearly membership unlock?
Translate Anywhere (⌥T in-place screen translation) and optional AI editing / context-aware screen translation are unlocked with the iTool yearly membership ($26.49/year by card or ¥99/year via Alipay), which also unlocks all 7 iTool apps.
Does my microphone audio leave my Mac?
Never. Speech recognition runs on your Mac using Apple's on-device recognizer and optional local Qwen3-ASR 0.6B or SenseVoiceSmall weights. Audio is never saved or uploaded.