Voice Input on Windows in 2026
Windows users have always had more dictation options than Mac users — in part because Dragon NaturallySpeaking built its reputation on Windows, and in part because Windows' open ecosystem attracted more third-party tools. But the landscape has shifted considerably.
Dragon's dominance has eroded as AI-powered alternatives emerged. Windows Voice Typing has improved. And new push-to-talk tools built on Whisper and other modern models have entered the market. Here is a complete map of your options.
Windows Voice Typing (Built-in, Win+H)
Microsoft's built-in voice typing, accessed with the Win+H shortcut, is the zero-friction starting point. It works in most Windows text fields, supports auto-punctuation, and costs nothing.
How it works: Press Win+H, and a floating microphone widget appears. Speak, and text appears in the active field. Say "stop listening" or press the button again to stop.
Accuracy: Good for English in quiet environments. Microsoft has improved the underlying model significantly since the original Windows 10 launch. Handles conversational speech reliably.
Auto-punctuation: Available and works reasonably well. You do not need to say "period" and "comma" for most sentences — the system infers them.
Language support: Supported languages are more limited than Whisper-based tools. As of 2026, Windows Voice Typing supports around 25 languages.
Limitations: No AI enrichment. Works only in Windows text fields, not every application. No custom modes. Output quality is raw transcription.
Best for: Windows users who need occasional voice input without installing anything.
Dragon Professional (Nuance)
Dragon Professional remains the gold standard for Windows dictation accuracy, particularly for specialized vocabulary. At $699 one-time, it is a significant investment, but it comes with capabilities no other tool matches.
Accuracy: Excellent, especially after voice training. Dragon learns your voice patterns and vocabulary over time. For medical, legal, or technical terminology, Dragon's accuracy on trained vocabulary outperforms general-purpose models.
Custom vocabulary: You can add domain-specific terms, proper nouns, and specialized phrases. This is Dragon's biggest competitive advantage.
Integration: Deep Windows integration, including control of applications by voice. Dragon can navigate menus, click buttons, and control the OS — far beyond typing text.
Latency: Near-instant for trained voices. Response time is consistently under one second.
Limitations: High upfront cost with no subscription option for the main product. Software architecture is dated. No AI enrichment or text formatting — it transcribes exactly what you say. Mac version discontinued.
Best for: Professionals in fields like medicine, law, or finance who need the absolute highest accuracy for specialized vocabulary and are on Windows.
Telvr (Windows Version in Development)
Telvr is currently a macOS application, with Windows support actively in development. The core experience — push-to-talk with AI enrichment, system-wide text insertion, Whisper large-v3 accuracy — is planned for Windows.
What Windows users can expect: The same workflow that macOS users have today. Hold a hotkey in any application, speak, release, and get formatted text at the cursor position within about two seconds. Six enrichment modes covering email, meeting notes, summaries, dev tasks, and general cleanup.
Why it matters for Windows: There is currently no Windows tool that combines Whisper-level transcription accuracy with AI text enrichment and true system-wide insertion in a simple push-to-talk interface. Windows Voice Typing lacks enrichment; Dragon lacks modern AI formatting; Whisper tools lack integration.
Pricing: EUR 3/month infrastructure plus EUR 0.03 per minute — identical to the macOS version.
If you are on Windows and this workflow appeals to you, signing up for the waitlist on the Telvr website is the best way to be notified when the Windows version launches.
Whisper-Based Tools (Windows)
Several community and commercial tools bring Whisper transcription to Windows:
Whisper Transcriber / local CLI: Run Whisper directly on Windows. Requires Python setup and CUDA-capable GPU for fast inference (though CPU works for smaller models). Produces raw transcription; no enrichment.
MacWhisper equivalent tools: Several Windows apps wrap Whisper in a basic interface. Most are file-based (record audio, get transcript), not real-time keyboard replacements.
Limitations: All current Windows Whisper tools require manual integration work. None offer the push-to-talk system-wide insertion experience that Telvr provides on macOS. No enrichment layer.
Best for: Developers, privacy-focused users, or those comfortable building their own pipeline.
Google Voice Typing (Chrome)
Google Voice Typing is available in Chrome browser on Windows. It works in any contenteditable field within Chrome, with accuracy that benefits from Google's massive training data.
Limitations: Chrome-only. Does not work in native Windows applications. No enrichment. Privacy considerations.
Best for: Users who primarily work in Chrome and need free voice input for web applications.
Windows Speech Recognition (Legacy)
Older than Windows Voice Typing, Windows Speech Recognition (accessible via the Control Panel or search) offers more commands but worse accuracy than the modern Win+H implementation. It is largely superseded by Windows Voice Typing and worth skipping unless you need its application control commands.
Comparison Table
| Feature | Windows Voice Typing | Dragon Professional | Telvr (macOS) | Whisper (local) | |---|---|---|---|---| | Platform | Windows | Windows | macOS (Win coming) | Both | | System-wide | Most apps | Yes | Yes | Custom setup | | AI Enrichment | No | No | Yes (6 modes) | No | | Latency | 1-3s | Under 1s | Under 2s | 3-15s | | Language support | ~25 | ~15 | 50+ (auto-detect) | 99 | | Price | Free | $699 one-time | EUR 3/mo + usage | Free | | Custom vocabulary | No | Yes | Custom prompt | No |
Recommendations for Windows Users
The current reality: Windows users in 2026 do not have a single tool that combines modern AI accuracy, enrichment, and seamless system-wide integration. That gap is what Telvr's Windows version will address when it launches.
In the meantime:
For occasional, free voice input: Windows Voice Typing (Win+H) is the obvious starting point. Its auto-punctuation and improved accuracy make it workable for everyday tasks.
For specialized professional vocabulary: Dragon Professional remains the only real option for Windows users who need custom vocabulary training and high accuracy on domain-specific terminology.
For technical users who want the best transcription: Local Whisper via a community wrapper like Whispering gives you Whisper accuracy, but requires setup and produces raw output.
For Windows users who want the Telvr experience: Sign up for the waitlist. The macOS version demonstrates what the Windows release will deliver.
The gap in the Windows market is significant, and it is exactly the gap that modern push-to-talk tools with AI enrichment are positioned to fill.