Push-to-Talk

Push-to-talk is the primary way you speak to Silk. It’s a press-to-toggle control: press the key once to start listening, speak, then press it again to stop. Silk transcribes what you said, captures your screen, and responds.

How it works

1. Press your push-to-talk key once to start listening.
2. Speak naturally. There’s no time limit.
3. Press the key again to stop. Silk registers your query and immediately begins processing.
4. Listen — a voice response plays through your speakers within a few seconds.

That second press is the step people most often miss — it’s how Silk knows you’re finished. Silk takes a screenshot of your screen at the moment you stop, so it has context for your question.

Caution

Push-to-talk is a toggle, not a hold. Press once to start and once to stop — don’t hold the key down while you speak.

Hotkey

The push-to-talk hotkey is Ctrl+Space on both macOS and Windows by default.

Changing the hotkey

1. Open the Silk panel and go to Settings → Voice hotkey.
2. Click the hotkey button. It shows Press a key combination…
3. Press the combination you want, for example Alt+Space or Ctrl+Shift+Space.
4. To go back to Ctrl+Space, click Reset.

If you don’t see Voice hotkey in Settings, update Silk: older versions couldn’t change it.

Windows: Ctrl+Space beeps or flickers on and off

On Windows, Ctrl+Space is also the shortcut for switching keyboard language or input method. If you have more than one keyboard language installed, Windows can grab the shortcut before Silk does: you hear a beep and listening flips on and off, so nothing is ever sent. It isn’t a microphone problem.

Fix: change Silk’s hotkey to Alt+Space or Ctrl+Shift+Space (see above). Or, if you’d rather keep Ctrl+Space, go to Windows Settings → Time & language → Typing → Advanced keyboard settings → Input language hot keys and set the switching shortcut to Not assigned.

Tip

The push-to-talk key is a global hotkey — it works even when Silk’s panel isn’t focused or visible. You don’t need to click on anything first.

Languages

Choose the language you speak in Settings → Language. It sets how Silk transcribes you and the voice it answers with. Spoken conversations are supported in:

• English
• Spanish (Español)
• French (Français)
• Chinese (中文)
• Japanese (日本語)
• Korean (한국어)

Typing works in any language. Typed requests (including through Telegram) go straight to the AI model, which understands almost every language, and Silk can read screens and apps in any language too. If your language isn’t in the list above, type your requests and Silk will work well; for voice, use one of the listed languages.

Conversation continuity

Silk remembers the entire conversation within a session. You can ask follow-up questions naturally:

“What does this function do?” (Silk explains) “How would I test it?” (Silk builds on the previous answer)

The conversation context is shown in the panel’s history view.

When Silk doesn’t respond

• Make sure the microphone permission is granted (see Permissions).
• Check that your system microphone is working, and that the correct mic is selected in the audio settings (try the test button).
• Make sure you pressed the key a second time to stop — Silk only processes your query once you toggle recording off.

Interrupting a response

If Silk is speaking and you press the push-to-talk key again, the current response stops and a new recording begins immediately.