For engineers
Voice to text for developers, running locally.
Ceylo is Voice Typing for Windows that transcribes on your own machine. It is built for the English around the code rather than the code itself: commit messages, pull request descriptions, review comments, design docs, incident notes and the long Slack reply you have been putting off. The Software pack teaches the recognizer the vocabulary, and the text lands in whichever window has focus, terminal included.
Free plan included · No account, no card · Windows 10/11
Updated
This pull request moves the retry out of the request handler and into the client. The endpoint does not change, so there is no migration to run. The deploy to staging has been up an hour and the latency is flat. rollback is one revert.
In the shipped Software packProse, not syntax: Ceylo cannot speak a brace or an identifier
What is in the Software pack?
Ceylo ships a Software vocabulary pack that loads engineering language before each dictation, covering the words a recognizer trained on ordinary speech tends to mangle. Your own My Terms entries, the service names and internal acronyms only your team says, always outrank the pack: Ceylo drops shipped terms first when the prompt runs out of room, and never drops yours.
- Delivery: deploy, staging, production, release, rollback, pipeline, CI/CD, DevOps, monitoring.
- Code and review: pull request, code review, merge, commit, branch, repository, refactor, debug.
- Architecture: backend, frontend, API, endpoint, database, cache, latency, microservices, Kubernetes, Docker.
How does Ceylo get text into an editor or a terminal?
Ceylo puts the transcript on the clipboard, sends Ctrl+V to the window that has focus, then restores whatever your clipboard held before. The whole block lands in one paste, so an editor with completion or an input with autocomplete has no gaps to interleave with and there is no letter by letter crawl. If the clipboard or the paste is unavailable, Ceylo falls back to sending the text as keystrokes so the dictation still lands.
Can I dictate code with Ceylo?
Ceylo dictates prose, not syntax. It types punctuated natural language and it has no command grammar: there is no spoken word that produces a brace, a camelCase identifier or an indent, and no way to say select, delete or undo. If you want to drive an editor by voice, a dedicated voice-control system is the right tool and Ceylo is not a replacement for one. What Ceylo is good at is everything written around the diff.
Does anything reach a server?
Ceylo runs the model locally, so a paragraph describing an internal system, a customer name in an incident note or anything you say about a private repository is never uploaded. Ceylo has no account, so there is no login, no token and no profile. What it does send is the traffic to fetch a speech model, check for updates and keep licensing honest, and none of it carries audio or text. The Privacy page keeps the full list.
What Ceylo is not, for a developer
The things engineers ask about first, answered before you download anything:
- Ceylo is not open source. The application is proprietary, and the speech models it downloads are third-party models fetched from their public host.
- There is no voice control and no editor plugin. Ceylo types text into the focused window and that is the whole integration surface.
- There is no API, no webhook and no command line. Ceylo is a desktop application, not a service you can build on.
- Windows 10 and 11, 64-bit, only. There is no macOS or Linux build and no web version.
Talk through the pull request instead of typing it.
The free plan needs no card and no account, and the installer takes about a minute.