Voice Input for GitHub Copilot

The short answer: open GitHub Copilot in VS Code, Visual Studio, or a JetBrains IDE with Copilot enabled, click into Copilot Chat or a comment block above your code, press ⌃+⌥+R (Mac) or Ctrl+Alt+R (Windows/Linux) to start recording, speak your full requirement, press the hotkey again, and AICHE inserts the prompt for Copilot to use.

This article is about GitHub Copilot in the editor: inline ghost completions and Copilot Chat. It is not about Microsoft Copilot in Windows or Edge. AICHE does not replace Copilot. It gets detailed natural-language context into the fields Copilot already reads.

How It Works

1. Install GitHub Copilot in your editor and sign in with a Copilot-licensed account.
2. Open Copilot Chat (VS Code: open the Chat view from the activity bar; JetBrains: open the Copilot tool window).
3. Click into the chat input, or place the cursor in a comment block directly above where you want code.
4. Press ⌃+⌥+R (Mac) or Ctrl+Alt+R (Windows/Linux) to start recording. Recording is toggle-based: press once to start, again to stop.
5. Speak for 30 to 90 seconds. Include types, error cases, libraries, and constraints.
6. Press the hotkey again. AICHE uses cloud transcription, optionally cleans up filler, and inserts text at your cursor. Audio is streamed for cloud transcription, processed, and discarded immediately after processing, within 1 second. No persistent audio copy.
7. For Chat: press Enter to send (or use your editor's send action). For inline: type the function signature or next line and let ghost completions appear.

Heads-up: AICHE inserts text only. It does not accept Copilot suggestions, run agent tasks, or modify files. You still review and apply Copilot output.

Copilot Chat: Panel, #selection, and @workspace

Copilot Chat is the sidebar or editor panel where you ask questions, request refactors, and generate tests.

Typical voice workflow:

1. Select the code you care about in the editor.
2. Open Copilot Chat and click the input field.
3. Press your AICHE hotkey and speak your request. Mention that the selection is the subject (in VS Code you can type or let Copilot attach #selection; for repo-wide questions use @workspace so Chat can search the project index).
4. Stop recording. AICHE inserts the prompt. Send it to Copilot.

Example spoken prompt (inserted as text): "Using #selection, refactor this handler so declined-card errors are not retried, add structured logging per attempt, and return a typed result instead of throwing. Keep the public method signature."

@workspace is for questions that need file context beyond the selection: "Where do we validate JWT issuers?" or "Add tests consistent with existing patterns in this service." Voice makes those prompts long enough to be useful without five minutes of typing.

Inline Ghost Completions and the Comment-Block Pattern

Inline suggestions appear as gray ghost text while you type. Copilot weighs nearby comments heavily. A one-line // todo helps a little; a short paragraph describing parameters, failures, and invariants steers completions much more reliably.

Comment-block-as-prompt workflow:

1. Place the cursor where the function will live.
2. Start a block comment (//, /* */, """, #, etc.).
3. Press ⌃+⌥+R and dictate the spec: inputs, return shape, error handling, performance limits.
4. Press the hotkey again. AICHE inserts the comment text.
5. Begin the function signature or the next line of code. Ghost completions should align with what you spoke.

This pattern works in TypeScript, Python, Java, Go, and other languages Copilot supports. You are not "prompting Copilot" through a special API. You are giving it the same context a thorough teammate would leave in a comment, produced at speaking speed.

Agent Mode (When Your Editor Offers It)

Some editors expose Copilot agent mode (or an agent-style chat mode) that can propose multi-file edits from Chat. Availability and behavior depend on your IDE version and Copilot plan. AICHE still only fills the chat input: dictate the goal, constraints, and files to touch, send the message, then review each edit Copilot proposes before accepting. Voice is for front-loading context; you keep control of apply/reject.

What You Get

• Copilot Chat prompts with selection and workspace context, spoken in full sentences.
• Comment blocks that steer inline ghost completions toward your real requirements.
• Fewer under-specified prompts that force three rounds of "almost right" code.
• No Copilot conflict - AICHE is a separate desktop app, not an extension.
• Custom Vocabulary for framework names, internal services, and acronyms Copilot should see spelled correctly in your prompt text.
• Software Development profile (Pro) for camelCase, snake_case, flags, and library names in speech.
• Cloud transcription; audio streamed for cloud transcription, processed, and discarded immediately after processing, within 1 second. No persistent audio copy.

Pricing

AICHE Personal starts at $3.99/mo (billed annually) with a 7-day free trial, no credit card required. Pro unlocks the Software Development profile for code-heavy dictation.

See plans and download

FAQ

Is this Microsoft Copilot?
No. This page covers GitHub Copilot inside VS Code, Visual Studio, JetBrains, and other supported coding tools. Microsoft Copilot in Windows or the browser is a different product.

Does AICHE trigger Copilot suggestions automatically?
No. After AICHE inserts text, you send Chat messages or continue typing so Copilot's own UI produces suggestions. AICHE does not press Tab or Accept for you.

Can I use voice in Copilot Chat and inline comments in the same session?
Yes. Dictate a Chat refactor, then dictate a comment block above the next function. Same hotkey, same toggle workflow.

Will long spoken prompts hit a length limit in Chat?
Very long dictation may need trimming before send depending on editor limits. For most tasks, 30 to 90 seconds of speech is enough and fits comfortably.

Is recording push-to-talk?
No. Toggle with ⌃+⌥+R (Mac) or Ctrl+Alt+R (Windows/Linux): press to start, press again to stop and transcribe.

What happens to my audio?
Audio is streamed for cloud transcription, processed, and discarded immediately after processing, within 1 second. No persistent audio copy.

Does AICHE work with Copilot in JetBrains IDEs?
Yes. Click into the Copilot Chat input or a source comment, use the same global hotkey, and insert text as in VS Code.

Related Features

• Voice Code for AI Coding Agents - Pro: pause-aware auto-send when you want the prompt to ship without pressing Enter.
• Software Development Profile - recognition tuned for code-heavy speech.
• Custom Vocabulary - correct spellings for repos, APIs, and internal terms in prompts.
• Smart Insert - text insertion at the cursor without clipboard round-trips.

Result: Copilot Chat gets prompts that mention selection and workspace context. Inline completions follow comment blocks you would never have typed at length. Less shorthand, fewer fix-up loops.

Try it now: download AICHE, select a messy function, open GitHub Copilot Chat, press your hotkey, and dictate a refactor with error-handling rules. Send it and review the diff Copilot proposes.