Speech to Text with Whisper

Professional voice input extension for VS Code and Cursor IDE with OpenAI Whisper API integration and AI text post-processing.

Core Functions

Key Features

Professional Audio Recording

• High-quality recording using FFmpeg • Cross-platform support: Windows, macOS, Linux
• Multiple formats: WAV, MP3, WebM, Opus • Automatic device detection

AI Transcription & Post-Processing

• OpenAI Whisper API with support for 40+ languages and auto language detection
• AI Text Post-Processing - Improve transcribed text quality using GPT models
  • Remove filler words (um, uh, like, you know)
  • Add proper punctuation and capitalization
  • Structure sentences for better readability
  • Maintain original meaning and technical terms

Smart Text Insertion

• Insert at cursor position or copy to clipboard
• Cursor IDE Integration: Direct send to AI chat

Important Note: Chat insertion functions use unofficial Cursor IDE APIs and may change in future versions.

Quick Start

1. Install FFmpeg (Required)

Windows: winget install FFmpeg | macOS: brew install ffmpeg | Linux: sudo apt install ffmpeg

2. Install Extension

Open VS Code/Cursor IDE → Extensions (Ctrl+Shift+X) → Search "Speech to Text with Whisper" → Install

3. Configure API Key

Get API key from platform.openai.com → Settings (Ctrl+,) → Search "Speech to Text with Whisper" → Enter OpenAI API key

4. Start Recording

Press Ctrl+Shift+N for AI chat or Ctrl+Shift+M for text insertion

Settings

Basic Settings

Post-Processing Settings

Available Models: Without post-processing, GPT-4.1 Mini (recommended), GPT-4o, GPT-3.5 Turbo, o1, o3, and others

Audio Settings

Supported Languages

Main Languages: English, Russian, Spanish, French, German, Italian, Portuguese

Asian Languages: Chinese, Japanese, Korean, Hindi, Thai, Indonesian, Vietnamese

European Languages: Dutch, Swedish, Norwegian, Danish, Finnish, Greek, Hungarian, Czech, Polish, Romanian, Ukrainian, Turkish

Other Languages: Arabic, Hebrew, Catalan

Total: 43+ languages with automatic detection

Commands

Access via Command Palette (Ctrl+Shift+P):

Recording

• Speech to Text with Whisper: Record and Insert at Cursor or Clipboard
• Speech to Text with Whisper: Record and Open New Chat

Settings & Tools

• Speech to Text with Whisper: Run Diagnostics
• Speech to Text with Whisper: Open Settings
• Speech to Text with Whisper: Select Audio Device
• Speech to Text with Whisper: Clear History

Extension Panel

Access via Activity Bar (microphone icon):

• Device Manager: Select audio input devices
• Recording Mode: Switch between "Insert Text" and "Copy to Clipboard"
• Settings: Quick access to configuration
• History: View and reuse past transcriptions with post-processing indicators
• Diagnostics: System health check

System Requirements

• VS Code: 1.74.0+ • FFmpeg: Installed in system
• OpenAI API: Key with Whisper access • Platform: Windows 10/11, macOS, Linux

Troubleshooting

"FFmpeg not found": Check installation with ffmpeg -version, add to PATH
"Recording in progress": Wait for completion or restart extension
"Recording stops automatically": Check silence detection sensitivity, increase to 30, 40, or 50 if needed
"No audio devices": Check DirectShow (Windows), Privacy settings (macOS), audio group (Linux)
"API key invalid": Verify format (starts with sk-), check credits

Run Speech to Text with Whisper: Run Diagnostics for automatic system check.

Development

git clone https://github.com/alexstich/vs-code-speech-to-text.git
cd vs-code-speech-to-text
npm install && npm run compile && npm run test

Support

• GitHub Issues - Report problems
• Discussions - Feature requests