Local-first voice typing

Speak your mind.Keep your flow.

Press a shortcut and speak. Saymore turns your words into text at your cursor, in the app you’re already using.

Free dictation · No Saymore account required

macOS 12+ · Windows 10/11 x64

Saymore voice typing illustration

Speaking
微信在线
Saymore 项目群4 位成员
周

Q8 的体积和内存数据已经跑完了,麻烦同步一下结论。

在这里输入消息

  1. Ready
  2. Speaking
  3. Transcribing
  4. At your cursor
  5. Saymore
Interactive illustration · Chinese dictation examplesThe text lands in the input field. You decide when to send it.

From a thought to a finished sentence

Stay in the app. Just start talking.

Default shortcutmacOS Right ⌘Windows Right Alt

  1. Place your cursor

    Focus the text field you want to write in.

  2. Press and speak

    Use your shortcut to start, then press it again to finish.

  3. Your words, written

    Saymore inserts the transcript. Optional refinement tidies the wording.

Your words, a little clearer.

Refinement only handles filler words, self-corrections, punctuation and structure. It doesn’t read your screen, answer questions, add facts or send anything. It runs on your own cloud API key.

  • Lists, paragraphs and punctuation
  • Leaves URLs, emails, paths, commands and version numbers alone
  • If refinement fails, you still get the cleaned-up transcript
Drops the filler, keeps your tone
1 / 3
What you said

So, um, I think this plan might, like, need another round of discussion. Not that we can’t do it.

What gets typed

I think this plan might need another round of discussion. Not that we can’t do it.

Pauses go. “Might” stays “might”.

Your data. Clear boundaries.

With local recognition, audio stays on your device. With a cloud speech provider, audio goes to that provider. With refinement on, only the transcript is sent.

Local recognitionAudio stays on your device
Your own keyOnly for cloud services, you pick the provider
Encrypted historyYou decide how long it’s kept

What it won’t do: read your screen, write replies, keep a knowledge base, run tasks or send messages.

Personal dictionary3 spellings
say moreSaymoreLearned
api keyAPI KeyAdded by hand
qwen three asrQwen3-ASRFrom CSV

Familiar names. The right spelling.

Add words by hand, import a CSV, or let Saymore learn from corrections it’s confident about. It only fixes the casing and character width of words you’ve confirmed, and never guesses at synonyms. The dictionary is free.

Dictation is free. Refinement is a one-time purchase.

Dictate with macOS Dictation or a local model, no key needed. Pay once to unlock refinement, which runs on your own cloud API key.

Dictation
Free
macOS Dictation or a local model. No key, no limit.
Personal dictionary
Free
Manual entries, CSV import, learning from your corrections.
Refinement
Paid
One-time purchase with a year of updates. Needs your own cloud API key.
See pricing and FAQ

At home in the apps you use.

We check app compatibility one by one in our release test matrix. The aim is the text fields you use every day.

EditorsCode review, prompts, issues
BrowsersWeb forms and online docs
Chat and emailMessages, email, customer replies
TerminalsCommands and technical notes

Find your pace, on your own machine.

Saymore keeps a local count of characters, dictation time and average speed. We’ll publish public comparisons only when there’s enough real usage data, together with the sample, the definitions and the method.

On this computerBuilds up as you use it

Characters, dictation time, characters per minute and a 7-day trend, all stored locally.

Public comparisonWaiting for real usage data
Published whenThe sample is big enough
Published withSample, definitions, method

Until then, you won’t see claims like “4× faster” on this site.

Every update, on the record.

New features, refinements and fixes, recorded version by version. See what’s changed in Saymore.

All releases
Saymore v0.2.2

Saymore v0.2.2 improves readability on the Models page and in local-model dialogs, with larger text and layouts adjusted to make model information, status, and actions easier to read.

Saymore v0.2.1

Saymore v0.2.1 improves model discovery and selection and refreshes security dependencies.

Rest your hands. Keep your train of thought.

Saymore runs on macOS 12+ and Windows 10/11, 64-bit (x64). Installers are on GitHub Releases. Dictation is free; refinement is a one-time purchase.