At a Glance:
VoiceTypr is an open-source AI-powered voice-to-text dictation tool for macOS and Windows that works completely offline with local transcription across 99+ languages and supports optional cloud-based AI enhancement.
Overview:
VoiceTypr is an open-source, AI-powered voice dictation tool that converts speech to text directly on a user's device. It works system-wide on macOS and Windows, inserting transcribed text at the cursor position in any application. The tool performs all speech recognition locally using downloadable models with hardware acceleration, requiring no cloud connection for core transcription. An optional AI enhancement feature can transform transcriptions into formatted outputs like emails or commits using cloud-based services, but this capability is separate from the primary offline dictation workflow.
Key Decision Points:
Offline-first transcription: All speech-to-text processing runs locally on your device, with no voice data sent to external servers during dictation.
Cross-platform desktop app: Native applications are available for both macOS (signed and notarized) and Windows, built with Rust and Tauri.
Optional cloud enhancement: AI-powered text transformation uses external services like Groq or Gemini and requires an active internet connection, separate from the core offline dictation.
Hardware acceleration support: The tool leverages GPU acceleration through Metal on macOS or NVIDIA, AMD, and Intel GPUs on Windows, automatically falling back to CPU when GPU is unavailable.
System-wide hotkey workflow: Dictation is triggered via a global keyboard shortcut and text appears directly at the cursor location in whichever application is active.
Core Features:
System-wide voice-to-text with hotkey: Trigger recording from anywhere using a system-level keyboard shortcut, with automatic text insertion at the current cursor position in any application.
Local AI transcription: Runs speech recognition models entirely on-device across multiple model sizes that trade between accuracy and speed, with support for 99+ languages and automatic language detection.
Hardware-accelerated processing: Uses platform-specific GPU acceleration including Metal on macOS and NVIDIA, AMD, or Intel GPUs on Windows, with transparent CPU fallback.
Optional AI text enhancement: Post-transcription transformation of text into formats like prompts, email drafts, commit messages, or notes using cloud-based Groq or Gemini models.
Recording cancellation: Double-press the Escape key during recording to cancel the current dictation session without inserting text.
Use Cases:
Developers can dictate comments, commit messages, or code descriptions directly into editors or terminals without switching context.
Users who handle text across multiple applications can insert transcriptions into tools like Slack, ChatGPT, or any other text-accepting software without per-app configuration.
Privacy-conscious users who need dictation capabilities can use the tool without their voice recordings ever leaving the device.
Users who frequently compose emails, messages, or structured notes can use the optional AI enhancement to reformat raw transcriptions into polished text.
Open-Source Alternative Value:
VoiceTypr is available under the AGPL v3.0 license and can be built from source. The project positions itself as an alternative to Wispr Flow and SuperWhisper, offering a pay-once model with continued offline transcription capability rather than subscription-based cloud services. All core transcription runs locally on the user's device, keeping recordings and processing off external infrastructure. The optional AI enhancement feature is clearly separated from the core dictation functionality, allowing users to choose whether to incorporate cloud-based text transformation while maintaining local-only voice processing.
