Unlike most VDI platforms, Amazon WorkSpaces ships microphone support turned on — AWS enables audio-in for all new WorkSpaces. The catch is what listens on the other end: Windows voice typing, inside a desktop your admins control. Pithflow dictates from your local machine instead and delivers cleaned-up text into the WorkSpace, audio-in or not.
Mostly, yes — and this page won't pretend otherwise. Per the Amazon WorkSpaces FAQ, "audio-in is enabled for all new WorkSpaces," with support on the Windows, macOS, Android, and iPadOS clients. That makes WorkSpaces the friendliest of the big VDI platforms for in-session voice input, and if plain Win+H inside your WorkSpace covers your needs, you may not need anything else.
The fine print starts in the WorkSpaces Group Policy admin guide: audio-in is a policy admins can switch off (the DCV template's "Enable/disable audio-in redirection" setting under Computer Configuration › Administrative Templates › Amazon › WSP), and on Windows WorkSpaces the feature depends on users keeping local logon rights — AWS notes that a Group Policy restricting local logon stops the audio-in capability from being applied. Managed fleets restrict exactly these things.
Three reasons, in the order people hit them. The recognizer: with audio-in working, what you get inside the WorkSpace is Win+H — raw speech-to-text with no cleanup pass, no tone control, and weak handling of Spanish or mixed English/Spanish. The fleet: if your admins disabled audio-in or restricted local logon, the microphone chain is gone and no amount of client-side fiddling brings it back. The alternative: installing a dictation app inside the WorkSpace means another package for IT to approve, license, and patch in the corporate image — and your audio gets processed from inside the company desktop rather than your own machine.
Everything moves one hop closer to you. Pithflow captures your voice on the local machine, runs AI cleanup on the transcript — punctuation, capitalization, filler removal, your chosen tone — and delivers the result through the keyboard and clipboard forwarding the WorkSpaces client already does: text goes onto your local clipboard and gets pasted into the focused field, with a typed-keystroke fallback. The WorkSpace never handles audio and nothing installs inside it. That's the same local-capture architecture Pithflow uses for RDP and Citrix, and it works whether your fleet's audio-in policy is on or off.
The dependency to be straight about: pasting into the session rides on clipboard redirection. AWS ships the DCV policy "Configure clipboard redirection" enabled for two-way copy and paste, per the admin guide — but the same policy offers Copy Only and Paste Only modes, and a fleet set to Copy Only blocks text from entering the session via paste. Pilot one seat on the free tier before you conclude anything about your environment.
| In a WorkSpace | Win+H (audio-in path) | Pithflow (local capture) |
|---|---|---|
| Depends on the audio-in Group Policy | Yes — off means silence | No |
| Depends on local logon rights (per AWS docs) | Yes, on Windows WorkSpaces | No |
| Where your audio is handled | Inside the corporate WorkSpace | Your machine + Pithflow's pipeline; only text enters |
| Transcript quality | Raw speech-to-text | AI cleanup, tones, bilingual ES/EN |
| Blocked when clipboard is Copy Only | No | Paste path yes — pilot one seat first |
Skip the policy archaeology. Install Pithflow on the machine you run the WorkSpaces client from — the free tier is 2,000 words a week with no card — open your WorkSpace, click into any text field, hold the hotkey, and speak one sentence. Text lands: your clipboard direction is open and you're done evaluating. Nothing lands: your fleet restricts paste into the session, and you learned that in five minutes without filing a ticket.
AWS's FAQ lists audio-in support for the Windows, macOS, Android, and iPadOS clients — the web client isn't in that list, so don't count on in-session dictation there. Pithflow changes the question: the browser tab running WorkSpaces Web is just another window on your local machine, and Pithflow types into whatever window has focus.
With AWS's defaults, no — the DCV policy 'Configure clipboard redirection' ships enabled for two-way copy and paste, and pasting into the WorkSpace is all Pithflow's primary path needs. If your fleet restricts clipboard to Copy Only (session to client), that paste path is blocked and the typed fallback is what's left. One dictation on one seat tells you which fleet you're in.
No. Pithflow never talks to the streaming protocol — it hands finished text to your local machine's input path, and the WorkSpaces client forwards typing and pasting over whichever protocol your bundle uses. If you can type into the WorkSpace, dictated text arrives the same way.
No. Audio is captured on your local machine and transcribed and cleaned up by Pithflow's cloud pipeline under your own account — it never enters the WorkSpace, the corporate image, or the streaming protocol. Only the finished text lands in the session. Audio is processed in the cloud, not stored, and the app does not work offline.
Free tier: 2,000 words a week. Installs on your local Windows machine — the WorkSpace image stays untouched.
On Citrix, Horizon, or plain RDP? The remote desktop dictation guide covers the whole family.
