This article is for the setup where Sprecho runs on the Windows PC in front of you and the applications you dictate into run somewhere else — a Terminal Server, an RDS farm, a RemoteApp window, or a full remote desktop.
Dictation works there. Some things behave differently than they do locally, and knowing which ones saves a lot of guessing.
The setup that works
- Install Sprecho on your local PC and sign in there. Nothing needs to be installed on the server for dictation itself.
- Connect to the remote machine as usual.
- Click into the text field inside the remote window.
- Hold your dictation key — Insert by default on Windows — and speak.
Sprecho records with your local microphone and delivers the finished text into the remote window the same way your keyboard delivers keystrokes. From the server's point of view, you typed.
Your microphone stays local — redirection changes nothing
Remote Desktop can forward audio in both directions, and the setting called Remote audio recording or "Record from this computer" is the one people reach for first. It is not needed here and it does not help.
Sprecho never asks the remote session for sound. It opens your local microphone directly and puts the text back into the window on your screen. Leave your audio-redirection settings as your IT team configured them.
This has a pleasant side effect: microphone quality does not depend on the remote connection at all. A congested link slows down how quickly text appears, never how well you were understood.
The plugin your remote-desktop client loads
Sprecho installs a small plugin that your remote-desktop client loads when it connects. Sprecho itself never loads it — the client does, per connection.
Its job is context. From your PC, a remote application is only a picture, so Sprecho cannot tell which app you are in or what is already in the field. The plugin carries that information back so per-app behaviour and correct capitalisation keep working across the remote boundary.
Two consequences worth remembering:
- After a Sprecho update, the new plugin takes effect on your next connection. A session that is already open keeps using the one it loaded.
- An open session blocks the update itself on older builds. That is the most common Terminal-Server complaint we get, and it has its own article: The Windows app will not update.
For Enterprise teams: the Terminal Server module
The plugin bridges the connection, but something inside the session has to read the cursor position. That is the Terminal Server module, which Enterprise teams can enable from the admin dashboard.
Your administrator installs a small MSI once, machine-wide, on the Terminal Server image. It starts automatically in every user session at logon, with no window and no per-user setup, and only a compact signal ever leaves the session — whether the cursor sits at the start of a sentence and the single character before it. Never text, never field contents, never window titles.
The module is advisory. If it is not installed or does not answer in time, dictation behaves exactly as it does without it. The download and the switch are both on the Downloads page of the Enterprise dashboard.
Known limits
- Already-connected sessions lag behind updates. Reconnect after updating Sprecho to get the current plugin.
- Citrix and other non-Microsoft clients. Plain dictation works, because Sprecho types into the window like a keyboard. The deeper context integration is built around Microsoft's Remote Desktop clients; if you are on Citrix and want the module, talk to support before rolling it out.
- Nested sessions. Connecting through one remote desktop into another leaves Sprecho two boundaries away from the app. Dictation still types, context does not survive the second hop.
- Sprecho inside the remote session is not the supported setup. It would depend on redirected audio arriving intact in the session, which is exactly the fragile part we designed around.
- Text arrives at connection speed. On a slow link, the finished text lands a moment later than it would locally. It is not lost.
Symptom → cause
| What you see | What it usually is |
|---|---|
| Recording works, nothing is typed into the remote app | The remote window lost focus, or the field is not a text field — the same rules as local dictation |
| Dictation ignores your dictionary in remote apps | Normal without the Terminal Server module — a remote window is a picture to your PC |
| A sentence continuing mid-thought gets capitalised | Same cause: no cursor information from inside the session |
| The update fails while you are connected | The plugin is loaded by the client — close the session and retry |
| The plugin behaves like the old version after updating | The session was already open; reconnect |
| Text appears seconds late | Connection latency, not recognition |
Still stuck?
If dictation types nothing at all into remote applications, first confirm it works in a local app such as Notepad. That single test tells support whether the problem is remote-specific or general — No text is inserted into my app covers the general case.
Otherwise contact support with your remote-desktop client and version, whether you use RemoteApp or a full desktop, and your Sprecho version from Settings → System → Updates.