What people use YazSes for¶
YazSes is one program — hold a key, speak, release, and your words are typed into whatever app has focus, transcribed entirely on your own machine. But people arrive at it from very different problems.
These pages start from the problem, not the feature list. Each one explains whether YazSes actually solves it, what to turn on, and — honestly — where it falls short.
-
Voice typing on Linux
Dictation that works on both X11 and Wayland, in any application, with no cloud service behind it.
-
Dictation on Wayland
Wayland blocks apps from typing into each other, which silently breaks most dictation tools. How YazSes gets around it on GNOME, KDE, sway and Hyprland.
-
Confidential & offline work
When the audio legally or practically cannot be uploaded — clinical notes, legal work, classified or air-gapped machines.
-
Writing code by voice
Spoken symbols, LaTeX maths, git phrases turned into git commands, and a safety gate before a destructive shell command runs.
-
Prompts for AI coding agents
Dictate long prompts into Claude Code, Codex or a chat app. Hold-to-talk never cuts you off mid-thought, and nothing leaves your machine.
-
Code and technical vocabulary
Get identifiers, jargon and punctuation right — personal vocabulary,
initial_prompt, vocabulary correction and spoken punctuation. -
Law, medicine & journalism
Dictation and transcription under a confidentiality duty, where the audio may not leave the building — what it does, and what it does not promise.
-
Controlling the computer
Undo, save, jump to a line, run the tests, or fire a multi-step macro — with the same key that types your words.
-
RSI & hands-free access
For repetitive strain injury, motor impairment, dysfluent speech, or eyes-free use — including non-keyboard triggers.
-
Transcribing recordings
Turn an existing audio or video file into text, with speaker labels, without uploading it anywhere.
-
Research interviews
A tool-neutral guide for qualitative researchers: why cloud transcription raises consent and ethics questions, and the free local options.
-
Meetings & minutes
Record a whole meeting hands-free and get a speaker-labelled transcript — and optional minutes — without a bot joining the call.
-
Dictating to a remote server
Speak into your laptop mic; the text is typed on the machine you are SSH'd into.
-
More than one language
Dictating in a non-English language with a multilingual Whisper model you choose — a how-to, not a support claim.
-
Chinese & Mandarin
Point YazSes at a multilingual model for Mandarin, and avoid the Simplified/Traditional trap.
-
Hindi & Indian languages
Configure a multilingual Whisper model for Hindi and other Indian languages, and what to expect when Devanagari is typed.
The common thread¶
Every use case above is served by the same install and the same locally downloaded model. Nothing on this site requires an account, an API key, a subscription, or a network connection after install.
If you are comparing YazSes against Dragon, Talon, Wispr Flow, nerd-dictation or cloud dictation, the comparison page is deliberately honest about which of them wins for which job.
Not sure where to start?¶
See also the FAQ and the install guide for your platform.