Back to blogs
Developers
VS Code
Developers

Voice Dictation in VS Code: A Complete Setup Guide

How to use SERP Transcribe for voice input in VS Code — for comments, commit messages, documentation, and anything else you'd normally type in your editor.

ST

SERP Transcribe Team

Product

Voice Dictation in VS Code: A Complete Setup Guide cover

VS Code doesn't have built-in voice dictation. But because SERP Transcribe works system-wide — it's not a plugin, it works in any application — it works in VS Code just like it works anywhere else on your computer. This guide covers the setup and the specific workflows where voice input adds the most value for developers.

Basic Setup

There's no VS Code-specific configuration required. Install SERP Transcribe, set your hotkey (something that won't conflict with VS Code's extensive shortcut library — more on that below), and it works immediately in any VS Code text field.

  1. Click into any text area in VS Code — a file, the terminal, the search field, the commit message box
  2. Hold your SERP Transcribe hotkey
  3. Speak
  4. Release
  5. Paste with Ctrl+V / Cmd+V

That's it. The text appears wherever your cursor is.

Choosing a Hotkey That Doesn't Conflict

VS Code uses an enormous number of keyboard shortcuts, which means hotkey conflicts are a real consideration. A few combinations that work well:

On Mac:

  • Option+Shift+Space — rarely used by VS Code
  • Fn+F13 — if your keyboard has F13
  • Ctrl+Option+Space

On Windows:

  • Ctrl+Alt+Shift+Space
  • Windows+Shift+Space

To check for conflicts: open VS Code's keyboard shortcut editor (Ctrl+Shift+P → "Open Keyboard Shortcuts"), search for your intended combination, and confirm it's unbound.

Where Voice Input Helps Most in VS Code

Code comments and docstrings. This is the highest-value use case. Explaining what a function does, why a particular implementation decision was made, what a parameter expects — these are natural language descriptions that are faster to speak than type. Click at the comment location, hold hotkey, speak the explanation.

# Click above this line, hold hotkey, speak:
# "Calculates the compound interest given principal, rate, and number of periods.
# Rate should be expressed as a decimal, e.g. 0.05 for 5%. Returns the total
# amount including principal."

Commit messages. The VS Code source control panel's commit message field is a perfect voice dictation target. Click into it, hold hotkey, speak your commit message with the full subject and body. Commit messages that would take a minute to type take 10 seconds to speak. See our dedicated dictating git commits guide for more on this.

README and documentation files. When editing markdown documentation, voice handles paragraphs of prose faster than typing. Speak sections of your README, API documentation, or changelog directly into the editor.

The integrated terminal. You can dictate into the VS Code terminal too, though this requires care — terminal commands are precise and a transcription error can cause problems. Better to type commands and use voice for comments, commit messages, and documentation only.

TODO comments. The notes-to-future-self variety: "TODO: refactor this once the API is finalized, current implementation assumes X but that will change." Dictated in 5 seconds, typed in 30.

PR descriptions. VS Code's GitHub extension shows the PR description field within the editor. Dictate your PR description directly there — summary of changes, what to review, testing notes.

What Voice Input Doesn't Work Well For in VS Code

Actual code. Syntax, variable names, brackets, function calls — code requires precision that voice transcription doesn't reliably produce. The workflow is typing code, speaking documentation. Don't try to dictate const fetchUserData = async (userId: string): Promise<User> => { — type that, speak the comment above it.

Single-letter variable names. Voice struggles with individual characters and short identifiers. Again, type code, speak explanations.

Precise formatting. Markdown formatting like **bold** or [link text](url) is awkward by voice. Speak the prose, add formatting manually.

A Practical Developer Workflow

Here's how a developer's session might use voice throughout:

  1. Start working on a function → type the code
  2. Function is getting complex → hold hotkey, speak a comment explaining the approach
  3. Something non-obvious about a parameter → hold hotkey, speak a docstring
  4. Finish the feature → source control panel, hold hotkey, speak the commit message
  5. Raise a PR → hold hotkey, speak the PR description
  6. Reviewing someone else's PR → hold hotkey, speak your inline review comments

The pattern: type the code, speak everything that surrounds it. That split covers the majority of where developers spend their writing time and where voice provides the most benefit.

Terminal-Adjacent Tip

If you use the integrated terminal heavily, one voice workflow worth knowing: dictate notes to yourself about what you're about to do before running commands. Hold hotkey, speak "going to restart the dev server and check if the auth middleware changes fixed the 401s," paste into a scratch comment or terminal comment line. This creates a small running log of your debugging process that's useful when you need to retrace your steps.

It's not a major feature — it's a 5-second habit that occasionally saves 20 minutes of "wait, what was I trying to do here."