Voice coding dictation for the writing that surrounds your code
Developers write far more prose than they admit: pull request descriptions, design docs, tickets, review comments, commit messages and long answers in team chat. Stenograph lets you speak that writing with identifiers, framework names and casing kept exactly as they appear in the codebase, in VS Code, the terminal and every other tool you use. It does not write your code for you. It writes what you would otherwise type about it.
Listening
A typical day for a developer using Stenograph
At 9:30 you pick up a ticket. Before touching the code, you open the design doc and talk through the approach for three minutes. The doc comes back with headings, a list of steps and "useOrderHistory" and "PostgreSQL" spelled correctly.
At 1:00 you finish the change. In the terminal you press the hotkey and dictate the commit message. The code aware style keeps it short and literal, with the identifier "retryWithBackoff" intact and no stray punctuation.
At 1:10 you open the pull request and dictate the description: summary, what changed, how it was tested. A custom PR style adds the sections automatically.
At 3:00 a teammate leaves a long review comment. You answer in the review thread by voice, then reply in Slack in two sentences.
At 5:00 you write a spec for next week and leave with your wrists rested.
Identifiers and framework names spelled exactly
The hardest part of dictation for engineers is the vocabulary. Stenograph includes a software vocabulary with common frameworks, libraries, cloud and DevOps terms, and you add your own service names, classes, functions and acronyms. Casing is respected, so camelCase, PascalCase, snake_case and kebab-case come out as written, without spaces inside them. Version numbers, file paths and flags keep their dots, slashes and dashes. The custom vocabulary speech recognition page shows how to import a list of terms from a CSV file.
- Function and class names such as "getUserProfile" and "OrderService"
- Framework names such as "Next.js", "FastAPI" and "Kubernetes"
- Paths and flags such as "src/api/routes.ts" and "--dry-run"
- Acronyms such as "gRPC", "OAuth" and "CI/CD"
Code aware style in VS Code and the terminal
In code editors and terminals, Stenograph switches to a code aware writing style. It keeps your words literal, does not add sentence punctuation where it would break something, and preserves symbols. In comments and docstrings it writes in a technical tone. Everywhere else, in issue trackers, docs and chat, the style adapts to the app, as described on the dictation writing styles page. Text is inserted at the cursor through the desktop app for Mac and Windows with a global hotkey, and in browser based tools through the Chrome extension.
Writing about code, not generating it
Stenograph is a dictation and transcription tool. It turns what you say into precise text. It does not generate source code, refactor files or act as a coding assistant. That is a deliberate boundary: the words in your commit message, spec or review are yours, spelled the way your codebase spells them. Use voice commands from the voice commands for dictation page to restructure or shorten what you said before you submit it.
Standups, design reviews and recorded calls
Design reviews and incident calls produce decisions that need to be written down. Record them on your device and Stenograph returns a transcript with speaker labels, a summary, the decisions and the action items. The ai meeting note taker page explains the meeting mode.
Which plan fits a developer or an engineering team
| Situation | Plan |
|---|---|
| Developer dictating PRs, commits and docs | Solo, 24 USD per month billed yearly, 1,500 dictation minutes, software vocabulary |
| Tech lead with many design reviews to summarize | Pro, 49 USD per month billed yearly, 900 transcription minutes, all vocabularies |
| Engineering team with shared service names | Team, 149 USD per month billed yearly, up to 5 seats, 10,000 shared terms |
| Engineering organization with SSO and SCIM | Enterprise, 499 USD per month billed yearly, up to 20 seats |
Teams usually share one dictionary of services and internal names, described on the shared dictionary for teams page. Prices are on the stenograph pricing page.
Questions developers ask
Does Stenograph write code for me?
No. It is a dictation tool for the text around your code: commits, pull requests, specs, docs, tickets and messages. Identifiers you say are spelled exactly, but it does not generate programs.
Does it work in the terminal?
Yes. The desktop app inserts text at the cursor in any terminal on Mac and Windows, using a literal, code aware style with symbols preserved.
Can it handle camelCase and snake_case?
Yes. Add the identifier to your dictionary once, or say it with the software vocabulary on, and the casing is kept exactly.
Can I try it before paying?
Yes. The one-time demo on the home page has a "Your terms" field for a few identifiers. Paid plans begin with Solo, 24 USD a month when billed for the year, and no tier costs nothing.
Spend less time typing about your code
Dictate your next pull request and see your identifiers land exactly as you wrote them.