Windows Speech Recognition is gone. What now?

What Microsoft actually changed

Windows Speech Recognition - the feature you reached through Ease of Access, trained with a reading passage, and drove with "open Notepad" - was retired on Windows 11 22H2 and later in September 2024. Voice Access, first shipped in 2022, is the designated successor. It is on-device, works without a connection, and does more of what most people used WSR for.

The retirement is version-scoped rather than universal. Windows 10 and Windows 11 21H2 still carry WSR. If you are on one of those and it works, nothing is forcing a change today, though planning for one is sensible since the feature has no future.

If a third-party tool of yours depended on the WSR speech API rather than on the WSR user interface, check that specific tool. Some applications used the underlying platform speech stack and were affected differently from the end-user feature.

Try Voice Access before you buy anything

Voice Access is free, already installed on a current Windows 11, and processes speech locally. It dictates, it corrects, it clicks controls by name, and it can put numbers or a grid over the screen so you can select something that has no useful name. For a large share of former WSR users that is the end of the search.

This site sells a competing product and still says this first, because recommending a paid subscription over a free built-in tool that would have done the job is how people end up distrusting every comparison page they read afterwards.

• Set-up is a download of the speech model on first use, then it runs offline.
• Dictation with punctuation, correction and spelling commands.
• Direct commands: "click Save", "scroll down", "switch to Chrome".
• Number and grid overlays for controls with no accessible name.

Where Voice Access stops

Voice Access executes an instruction you give it. It does not decide how to accomplish something you have described. "Click the Save button" works; "file last month's invoices and reply to the two that are overdue" is not a Voice Access request, and no amount of rephrasing makes it one.

That distinction is the whole reason a third category of tool exists. Fluent accepts the described outcome, has a planning model choose a sequence of bounded actions against the Windows accessibility tree, shows each step as it runs, and lets the run be stopped. It is slower and less predictable than a direct command, which is exactly why the progress display and the stop control are not optional extras.

The practical answer is usually both

These are not competing installs. Voice Access is the right tool for a single known action and for dictation, and it is free and offline. Fluent is the right tool for the multi-step chore you would otherwise click through, and it costs money because the planning does.

A layered setup also fails better. When a plan goes wrong, a direct command is the fallback that always works, and when speech itself is the problem - a bad throat day, a noisy room, a shared office - Fluent takes the same request typed.

Questions

Keep reading

• How do you type with your voice on Windows?Two of the three good answers are already on your machine and cost nothing. Here is which is which.
• Which should you choose?Windows Voice Access is the mature default. Fluent explores a different natural-language layer.
• Which Windows voice control comparison do you need?Five comparisons, each answering one question. Start with the one that names the tool you already have.

Sources

• Microsoft: use voice recognition in Windows — Official page covering the WSR retirement and Voice Access.
• Microsoft: set up voice access — Requirements, language packs and first-run download.
• Microsoft: interact with items on the screen — The command vocabulary and the number and grid overlays.
• Fluent vs Windows Voice Access — A longer comparison of the two approaches.