devonmochi/byetype

What it solves

ByeType is a voice-to-text and screen-capture tool that uses multimodal AI to provide highly accurate, context-aware transcriptions. It solves the common issues with standard dictation software, such as poor recognition of industry-specific terminology, incorrect handling of proper nouns, and the inability to format or clean up conversational filler words.

How it works

Unlike traditional tools that transcribe audio and then require a second pass by an LLM, ByeType sends raw audio directly to a multimodal model. You define custom rules, vocabulary lists, and output styles using Markdown files. The application applies these rules in real-time during the transcription process. It also features an "automatic learning" function that analyzes your manual corrections to improve future accuracy. For screen captures, it uses AI to interpret layout, remove UI clutter, and fix formatting issues like broken line breaks.

Who it’s for

It is designed for professionals and power users who frequently use voice input for documentation, coding, or communication and need high-precision output that understands technical jargon, specific brand names, and custom formatting requirements.

Highlights

• Customizable Rules: Use Markdown to define proprietary vocabulary, preferred formatting, and output styles.
• Multimodal Processing: Direct audio-to-text processing allows for immediate application of style and context rules.
• Smart Cleanup: Automatically removes filler words, formats numbers/units, and handles code snippets or text blocks from screenshots.
• Cross-Platform: Available on macOS, Windows, and via iOS Shortcuts.
• Developer-Friendly: Includes a local HTTP API for integrating voice transcription into other workflows or scripts.