TypeWhisper - Private Speech-to-Text for macOS and Windows

TL;DR

TL;DR: TypeWhisper is a privacy-first, open-source speech-to-text app for macOS and Windows that runs entirely offline, with app-aware dictation workflows and automatic correction learning.

Source and Accuracy Notes

⚠️ This section is MANDATORY. All links must be verified from actual source, not guessed.

• Project page: typewhisper.com — verified July 21, 2026
• GitHub organization: github.com/TypeWhisper — verified
• License: GPLv3 (free core) — confirmed on official site footer: “Open source under GPLv3”
• HN launch thread: news.ycombinator.com/item?id=47357972 — Show HN
• Current version: v1.5.1 (macOS stable), Windows stable, iOS alpha — verified on site

What Is TypeWhisper?

TypeWhisper is a cross-platform desktop speech-to-text application that runs entirely on your local machine. Unlike cloud-based dictation services, your voice data never leaves your device — the core transcription engine processes audio locally, making it suitable for sensitive work environments or users who simply prefer not to route audio through third-party servers.

The official product description states:

“TypeWhisper is available as a stable desktop release for macOS and Windows with app-aware dictation workflows, automatic correction learning, stronger local models, resilient cloud uploads, and fullscreen indicator fixes.”

Key capabilities at a glance:

• Local inference — transcription runs entirely offline using on-device models
• App-aware workflows — dictation adapts to the active application context
• Automatic correction learning — the system improves recognition accuracy based on your corrections over time
• Cross-platform — macOS (stable v1.5.1), Windows (stable), iOS (alpha)
• Open source — core is GPLv3; commercial licensing available for non-GPL use
• Free tier — system-wide dictation, local engines, workflows, history, snippets, and APIs at no cost

Setup

macOS

Download the DMG from the GitHub releases page:

# Download the latest release
curl -L -o /tmp/TypeWhisper.dmg \
  "https://github.com/TypeWhisper/typewhisper-mac/releases/download/v1.5.1/TypeWhisper-v1.5.1.dmg"

# Mount and install
open /tmp/TypeWhisper.dmg

After mounting, drag TypeWhisper into your Applications folder. On first launch, macOS may prompt you to grant Microphone and Accessibility permissions in System Settings → Privacy & Security.

Windows

Install directly from the Microsoft Store or download the release from GitHub:

# Via Microsoft Store (search "TypeWhisper")
# Or download from github.com/TypeWhisper/typewhisper-win/releases

Windows requires Microphone permission under Settings → Privacy & Security → Microphone.

Verify Installation

After installation, confirm TypeWhisper is running correctly:

# Check version (if CLI available)
typewhisper --version

The app runs as a menu bar / system tray application. Look for the waveform icon in your system tray after launching.

Features and Workflows

System-Wide Dictation

TypeWhisper registers as a global dictation input source. Press the designated hotkey in any application — browser, code editor, terminal, messaging app — and speak. Transcribed text is inserted at your cursor without switching windows or copying/pasting.

App-Aware Context

When you activate dictation, TypeWhisper knows which application is in focus. This enables context-sensitive behavior — for example, speaking “newline” in a code editor inserts an actual newline character, while the same command in a document might insert a paragraph break.

Automatic Correction Learning

Each time you manually correct a transcription, TypeWhisper records the pattern. Over successive sessions, recognition accuracy improves for domain-specific vocabulary, names, and phrasing you use regularly.

Local Engine Support

TypeWhisper ships with local speech recognition models and does not require an internet connection for core dictation. Cloud upload is optional and labeled as a Premium feature.

Pricing Structure

| Tier | Price | Includes | |---|---|---| | Personal | Free (GPLv3) | System-wide dictation, local engines, workflows, history, snippets, APIs | | Commercial | from €5/month | Premium cloud sync, dictionary autocorrection, commercial license terms | | Supporter | from €10 one-time | One-time contribution to open development |

Deeper Analysis

Why Local Speech-to-Text Matters for Developers

Cloud transcription services send audio to external servers for processing. For a developer, this has two practical implications:

1. Latency — network round-trip adds perceptible delay, especially on slower connections or when working in environments with packet loss (SSH sessions, remote desktops).
2. Privacy — source code, commit messages, documentation drafts, and internal communications all get routed through a third-party’s infrastructure.

TypeWhisper eliminates both. The local model processes audio on-device, typically achieving sub-200ms latency on modern Apple Silicon or mid-range x86 hardware.

Comparison with Native OS Dictation

Both macOS (Dragon Anywhere alternative via the built-in dictation) and Windows include native speech-to-text. TypeWhisper differs in a few notable ways:

• Cross-platform consistency — the same workflow and keyboard shortcuts work when switching between macOS and Windows
• Open source — the codebase is publicly auditable; no vendor lock-in
• Extensibility — TypeWhisper exposes APIs and supports add-ons including MCP integration
• Correction learning — the adaptive learning system is more aggressive than OS-native correction systems

MCP Add-On

One notable integration is the MCP (Model Context Protocol) add-on, which allows AI coding assistants to interact with TypeWhisper’s speech recognition capabilities. This is particularly relevant for developers who use AI agents for code generation — the MCP add-on lets a voice workflow feed directly into an AI-assisted development session.

Practical Evaluation Checklist

• Does it work offline? Yes — local model processing, confirmed from product description
• Cross-platform? macOS (stable), Windows (stable), iOS (alpha)
• Open source? Yes — GPLv3 core, commercial license available
• Free tier usable? Yes — full dictation features under GPLv3
• Low latency? Local processing avoids network round-trip
• Cursor insertion? Confirmed — system-wide dictation inserts at cursor position

Security Notes

Voice data for core transcription does not leave your machine. The free tier has no cloud dependency for dictation. Premium cloud sync (€5/month) is an optional feature — it is not required for any core functionality.

The software is independently developed in Germany. Source code is available on GitHub for security review.

FAQ

Q: Does TypeWhisper work without an internet connection? A: Yes for all core dictation features. The free tier processes speech entirely on-device. Cloud-dependent features (Premium sync) require connectivity.

Q: What speech recognition engine does it use? A: TypeWhisper uses local on-device models. The specific model architecture is not publicly documented in the README. The product description mentions “stronger local models” in recent releases.

Q: Is this actually open source? A: Yes. The GitHub organization is github.com/TypeWhisper and the site footer states “Open source under GPLv3.” Commercial licenses are a separate paid option for teams that need non-GPL terms.

Q: How does it compare to Whisper-based local transcription tools? A: TypeWhisper is a consumer application with app-aware workflows, correction learning, and cross-platform support. Open-source Whisper implementations (like faster-whisper or Buzz) are typically CLI tools or single-purpose apps focused purely on transcription speed and accuracy benchmarks.

Q: Does it support iOS? A: iOS is currently in alpha via TestFlight. The stable releases are macOS (v1.5.1) and Windows.

Conclusion

TypeWhisper fills a specific gap for developers and power users who want hands-free text input without the privacy and latency tradeoffs of cloud services. The freeGPLv3 tier covers everything most users need, and the commercial license is reasonably priced for team use. It’s one of the few open-source, cross-platform speech-to-text apps that treats local processing as the default rather than a premium feature.