If you only need basic dictation into common Windows apps, the built-in Voice Access and Win+H voice typing are good enough and cost nothing. You start needing a dedicated app like PipeVoice when you want to pick your own transcription engine, run fully offline, get AI cleanup of filler words and punctuation, or dictate reliably into terminals, code editors, and Claude Code where the built-in tools struggle.
What Windows Voice Access and Win+H voice typing actually do
Windows ships two related speech features. Win+H voice typing is a quick dictation panel: press Win+H, speak, and it types into the focused text field using Microsoft's cloud recognizer. Voice Access (Windows 11) is broader, adding hands-free navigation and on-device recognition so you can click, scroll, and control windows by voice, not just dictate.
Both are free, pre-installed, and require zero setup beyond a microphone. For replying to emails, jotting notes, or filling a search box, they work without you installing anything.
Where the built-in tools shine
Give Microsoft credit where it is due. The built-in tools are genuinely the right call in several cases:
- Free and already there. No download, no account, no API key.
- Basic dictation in Microsoft-friendly apps. Word, Outlook, Edge, and standard text boxes handle dictation cleanly.
- Accessibility navigation. Voice Access can drive the whole desktop by voice, which PipeVoice does not try to do.
- On-device option. Voice Access can run recognition locally for privacy and offline use.
If that describes your needs, you can stop reading. There is no reason to add software.
Where they fall short
The cracks show up the moment your work gets specific. Three gaps come up again and again.
App support is patchy
Voice typing assumes a well-behaved standard text control. Terminals, many code editors, custom Electron chat boxes, and developer tools often do not qualify, so dictation either does nothing or drops characters. If you have hit this, see Win+H voice typing not working for the specifics.
No engine choice
You get Microsoft's recognizer and only Microsoft's recognizer. You cannot swap in a more accurate model for hard audio, a streaming engine for live feedback, or a fully local model for privacy.
No AI cleanup
Built-in dictation transcribes literally. Your "um", "you know", and false starts land in the text exactly as spoken, and punctuation is hit or miss unless you say it aloud.
PipeVoice vs Windows voice typing: side by side
PipeVoice is a free, open-source, push-to-talk voice typing app for Windows 10 and 11. You hold a hotkey (default Ctrl+\, or Right Ctrl), speak, release, and it types real keystrokes into whatever app is focused. Here is how the two compare.
| Capability | Windows Voice Typing / Voice Access | PipeVoice |
|---|---|---|
| Price | Free, built in | Free, open source |
| Engine choice | Microsoft recognizer only | Deepgram, OpenAI Whisper, or Local Whisper |
| Live streaming text | No | Yes, with Deepgram |
| Fully offline path | On-device option (Voice Access) | Yes: Local Whisper + Ollama, no key |
| AI cleanup (fillers, punctuation, casing) | No | Yes, optional Flow mode |
| Types into terminals and code editors | Unreliable | Yes, sends real keystrokes |
| Per-app profiles | No | Yes |
| Accent / language picker | Limited | British, US, Australian, Indian, NZ English and more |
| Voice commands | Yes | Yes ("new line", "scratch that", "send it") |
| Bring your own API key | N/A | Yes (cloud engines) |
Engine choice and accuracy
This is the core difference. With PipeVoice you pick the engine that fits the moment, and you bring your own free or low-cost API key:
- Deepgram (streaming): words appear live as you speak, the fastest option, needs your own free API key, roughly pennies per day.
- OpenAI Whisper (batch): the most accurate option, needs your OpenAI key.
- Local Whisper / faster-whisper: runs fully offline on your PC, free, no key. First use downloads a model of about 150MB, and you can raise the model size for more accuracy.
The built-in recognizer gives you none of these levers. If Microsoft mishears your domain jargon, you are stuck; with PipeVoice you can switch engines or add vocabulary boosting for specific terms.
Typing into terminals, editors, and Claude Code
Because PipeVoice emits real keystrokes rather than relying on a dictation hook, it types into anything that accepts text: a terminal, VS Code, Cursor, a browser chat box, or the Claude Code CLI. That matters for developers dictating prompts, since one consistent hotkey works in the CLI and everywhere else. A second hotkey copies the result to your clipboard instead of typing it, which is handy when an app fights direct input.
Offline and privacy differences
Voice Access can run on-device, which is good. PipeVoice gives you a fully offline path too: Local Whisper for transcription plus local Ollama for AI cleanup means zero cost, no key, and nothing leaving your PC. If you choose a cloud engine instead, audio goes only to the provider you picked, on your own key, and AI polish sends text only, never audio. PipeVoice itself has no account, no telemetry, and no servers of ours.
Optional AI polish
Flow mode cleans filler words, fixes punctuation, and corrects casing after transcription. You choose the provider: OpenAI, Google Gemini (free tier), OpenRouter (free community models), or local Ollama for an offline, no-key setup. The built-in Windows tools have no equivalent, so your raw "uh, so basically" stays in the text.
Honest limitations of PipeVoice
To be fair both ways:
- Windows only. There is no Mac or Linux build.
- The installer is currently unsigned, so Windows SmartScreen shows an "unrecognised app" warning. Click More info then Run anyway. Code signing is in progress.
- Cloud engines require your own API key.
- Local Whisper is slower than cloud and wants a decent CPU for larger models.
Which should you use?
- Just basic dictation in Word, Outlook, or a browser, and you want zero setup? Use Win+H or Voice Access. Done.
- Need to navigate the whole desktop hands-free? Voice Access is built for that.
- Dictating into terminals, code editors, or Claude Code, or you want a faster/more accurate engine, AI cleanup, or a fully offline setup? Use PipeVoice.
The two are not really rivals so much as different tiers. The built-in tools cover the floor; PipeVoice picks up where they stop. If you want to dig into the broader landscape, see free voice typing software for Windows, our overview of speech-to-text on Windows, and a practical guide to dictating on Windows. You can also compare against the paid cloud option on our PipeVoice vs Wispr Flow page.
Download PipeVoice free and try the offline path, or read the docs first. Talk faster than you type.