Voice to Text for GitHub Copilot
GitHub Copilot is brilliant at turning comments into code, but typing those comments is the bottleneck. The more context you give Copilot, the better the suggestions — but who has time to type paragraphs of explanation? Blurt lets you dictate detailed prompts, code comments, and chat queries naturally. Hold a button, describe exactly what you want your function to do, release. Your words appear instantly as a comment that Copilot can read. Better prompts mean better code suggestions. Talk to Copilot like you'd explain to a colleague.
The Typing Problem
Copilot needs context you don't have time to type
You know that Copilot gives better suggestions when you write detailed comments. 'Create a function that fetches user data, handles pagination, retries on 429 errors, and caches results for 5 minutes.' But typing all that context takes longer than writing the code yourself. So you type 'fetch user data' and get generic suggestions that miss your requirements.
Your code comments are too brief to be useful
Comments should explain the why, not the what. But after fighting with code all day, writing 'This implements rate limiting because our API provider throttles after 100 requests per minute and we need exponential backoff' feels like too much effort. You write '// rate limit' and move on. Six months later, nobody knows why.
Chat mode queries require detailed explanations
Copilot Chat is powerful when you give it context. But typing 'Can you explain why this regex fails on Unicode characters and suggest a fix that handles emoji while maintaining the existing capture groups?' takes forever. You type something shorter, get a partial answer, and spend more time in follow-up.
Function descriptions become an afterthought
You should document what each function does, its parameters, return values, and edge cases. The information is all in your head right now. But typing JSDoc or docstrings for every function? That's hours of work. So you skip it, and the knowledge leaves when you do.
Explaining code intent interrupts your flow
You're deep in implementation when you realize Copilot needs more context. Stopping to type a detailed comment pulls you out of the zone. By the time you've explained what you want, you've lost the mental model of what you were building.
How It Works
Blurt works everywhere Copilot does — VS Code, JetBrains IDEs, Neovim. Anywhere you can type a comment or open Copilot Chat, Blurt can insert your dictated text.
Hold your hotkey
Press your chosen shortcut. A small indicator shows Blurt is listening.
Describe your intent naturally
Explain what the code should do, why it works this way, or what you need from Copilot. Speak like you're explaining to a teammate.
Release and code
Your words appear as a comment or prompt. Copilot reads the context and generates better suggestions immediately.
Real Scenarios
Dictating detailed comments that become Copilot prompts
Position your cursor, hold your hotkey, and say 'Create a TypeScript function that validates email addresses using RFC 5322 compliant regex, returns a boolean, and handles edge cases like plus addressing and international domains.' Release, and Copilot reads your detailed comment to generate exactly what you described. Rich context, precise output.
Describing what code should do before writing it
Before implementing, dictate your intent: 'This React hook manages WebSocket connections. It should auto-reconnect with exponential backoff starting at 1 second, max 30 seconds. It exposes connection state and a send method. Cleanup on unmount.' Copilot now understands the full picture and suggests complete, correct implementations.
Writing rich queries for Copilot Chat
Open Copilot Chat, hold your button, and speak: 'I have a performance issue where this database query takes 3 seconds on tables with more than 100k rows. The query joins users and orders tables. Can you suggest indexing strategies and query optimizations?' Full context, specific answers.
Explaining architectural decisions in comments
Just made a tricky design choice? Dictate the reasoning: 'We use an event-driven architecture here instead of direct function calls because this component needs to be decoupled from the notification service for testing purposes and to allow future replacement of the notification provider.' Future developers will thank you.
Documenting functions as you write them
Before the function signature, dictate: 'Calculates the optimal shipping rate based on package weight, dimensions, destination zone, and current carrier pricing. Returns the cheapest option that meets the delivery deadline. Throws if no carriers can meet the deadline.' Complete documentation in seconds.
Asking Copilot to explain unfamiliar code
Highlight confusing code, open Chat, and speak: 'Can you explain what this recursive function does step by step? I understand it's building a tree structure but I don't understand how it handles the edge case when parent ID is null.' Detailed questions get detailed explanations.
Providing context for code reviews
Adding review comments? Dictate: 'This approach concerns me because it creates a new database connection for each request instead of using the connection pool. In production with 1000 concurrent users this will exhaust the connection limit.' Clear, specific feedback without the typing tax.
Why GitHub Copilot users choose Blurt over built-in dictation
| Blurt | macOS Dictation | |
|---|---|---|
| Activation | Single hotkey, instant start | Press twice or click microphone icon |
| Speed | Text appears in under 500ms | 2-3 second delay typical |
| Technical terms | Handles coding vocabulary accurately | Mangles function names and syntax terms |
| Reliability | Consistent across coding sessions | Often fails silently or stops responding |
Frequently Asked Questions
Start Typing Faster Today
Free to try — no credit card required
Download Blurt