The best offline speech recognition on a Windows PC starts with Voice Access, which is built into Windows 11 and handles dictation and PC control with no internet connection. For more accurate dictation, Handy is free and open source: hold a shortcut, speak, and Whisper transcribes on your PC and types into any app. Dragon Professional v16 is the paid option for people who dictate all day. To turn recorded voice notes and other files into text, Buzz and Vibe run Whisper locally for free. One thing to know up front: Windows voice typing (Win + H) is not offline. It sends your speech to Microsoft’s cloud.
Windows voice typing vs Voice Access: which one is offline? #
These two sound similar but work very differently.
Voice typing is the dictation bar you open with Win + H. Microsoft says it “uses online speech recognition, which is powered by Azure Speech services” and that you need to be connected to the internet to use it. Your speech goes to Microsoft’s servers. That’s a reasonable design for a free feature, but it’s the wrong tool for client notes, health details or a journal. On Copilot+ PCs, Microsoft adds “fluid dictation,” which uses on-device small language models to clean up text. That’s a layer on top of dictation, not a switch to on-device speech recognition for voice typing.
Voice Access is an accessibility feature in Windows 11, version 22H2 and later. Microsoft describes it as a way to “control their PC and author text using only their voice and without an internet connection.” It runs on the device.
If privacy or offline use is why you’re reading this, use Voice Access and leave voice typing alone.
Offline speech recognition apps for Windows compared #
| App | What it does | Works offline | Cost | Best for |
|---|---|---|---|---|
| Voice Access | Dictation and full PC control | Yes | Built into Windows 11 | Hands-free use, accessibility |
| Handy | Push-to-talk dictation into any app | Yes | Free, open source | Accurate dictation with Whisper |
| Dragon Professional v16 | Dictation, commands, file transcription | Desktop app | Paid; check Nuance for pricing | Professionals who dictate all day |
| Buzz | Transcribing audio and video files | Yes | Free, open source | Voice notes, recordings, subtitles |
| Vibe | Files and batches, many formats | Yes | Free, open source | Back catalogs, video |
| Spokenly | Hotkey dictation | Yes, with local models | Local models free | Dictation if you switch on Local Only Mode |
| Windows voice typing | Dictation | No, uses Azure | Free | Quick use when cloud is fine |
1. Voice Access: best built-in option #
Voice Access comes with Windows 11, so there’s nothing to install:
- Open Settings > Accessibility > Speech and turn on Voice access.
- Download the speech files for your language when prompted. Some languages are ready to use and others need a download.
- Click into a document or note and speak. Voice Access types what you say and also understands commands for editing and moving around.
It has a dictation mode that treats everything you say as text, and a default mode that mixes commands with dictation. You can open programs, click buttons, scroll, and select and correct text.
Good for: people who want to use the whole PC by voice, and anyone who needs offline dictation without installing anything.
Less good for: transcribing recordings, since it only listens to your microphone. It’s also more command-driven than pure dictation tools, which takes a little learning.
2. Handy: best free dictation with Whisper accuracy #
Handy is a small, free, open-source (MIT) app for Windows, Mac and Linux. According to its project page, you press or hold a keyboard shortcut and speak. It transcribes locally with Whisper models or Parakeet V3, a CPU-friendly model with automatic language detection, and pastes the text into whatever app has focus. Parakeet means it works on laptops without a dedicated GPU.
Good for: writers, developers and anyone who wants fast, accurate dictation into email, documents or code comments.
Less good for: PC control and file transcription. It does neither.
3. Dragon Professional v16: best for heavy dictation #
Dragon has been the professional dictation standard for decades. The current version, Dragon Professional v16, is optimized for Windows 11 and compatible with Windows 10. It covers live speech-to-text and transcription of pre-recorded audio files. It learns your vocabulary and supports custom commands, which is why legal and medical offices use it.
Good for: people who dictate for hours a day and want deep customization.
Less good for: casual users. It’s paid software with a learning curve, and it’s Windows only. Some Dragon products, such as the Anywhere mobile versions, are cloud services, so check which edition you’re buying.
4. Buzz: best for transcribing voice notes and recordings #
Speech recognition often means turning a recording into text rather than dictating. If you record voice notes with Windows’ Sound Recorder app, a USB mic or a phone, Buzz turns the files into text locally with OpenAI’s Whisper:
- Download Buzz from its GitHub page. It’s free and MIT-licensed.
- Open it, choose a Whisper model and let it download once.
- Drag in your recordings. Buzz transcribes on your PC and can use an NVIDIA GPU through CUDA, or other GPUs through Vulkan.
- Export to TXT for notes, or SRT/VTT if you need timestamps.
Buzz also identifies speakers, translates offline, and has a watch folder: point it at the folder where your voice notes land, and new recordings are transcribed automatically. That’s the closest thing on Windows to a hands-off private voice notes inbox. The Windows installer isn’t signed, so SmartScreen will warn you when you install it.
5. Vibe: best for batches and export formats #
Vibe is another free, open-source Whisper app. It’s stronger than Buzz at batch jobs and exports SRT, VTT, TXT, HTML, PDF, JSON and DOCX. Choose it when you have a folder of recordings to get through. For a walkthrough with both apps, see how to transcribe podcasts offline on Windows.
What about Spokenly and Superwhisper? #
Spokenly offers free, unlimited local Whisper and Parakeet models for hotkey dictation. It also sells cloud features, so switch on its Local Only Mode for private use. Superwhisper has a Windows version with local models and optional cloud AI. If you use it, keep confidential dictation on a local model.
What about Microsoft 365 and OneNote? #
Transcription features in Microsoft 365 apps process audio in Microsoft’s cloud and depend on your subscription. They’re convenient if your organization already trusts Microsoft 365 with its documents, but they aren’t offline. For the wider trade-off, see on-device vs cloud transcription.
What about voice notes you record on your phone? #
A Windows PC is a good place to write, but voice notes and meeting recordings tend to happen on a phone, on a walk or right after a meeting. You can move each file to the PC and run it through Buzz, or transcribe on the phone and move only the text.
Private Transcribe is an iPhone and Android app, not a Windows program, so it doesn’t replace the tools above. It covers the phone side: it runs Whisper on the phone with no account and no upload, and after a one-time model download it works in airplane mode. You record, and the text appears once you stop, with timestamps and a searchable history. It can also import audio or video files (m4a, mp3, wav, aac, mp4, mov) up to 90 minutes long. It doesn’t label speakers, and the free version shows ads. To get the text onto your PC, copy it, or with the one-time Pro purchase export a .txt file and move it by USB or a local transfer app rather than a cloud drive.
How to confirm a Windows app really runs offline #
- Download the app and its model while online.
- Turn on airplane mode, or disconnect Wi-Fi and Ethernet.
- Dictate or transcribe a short file.
- For extra certainty, block the app in Windows Defender Firewall and try again.
If it still produces text, the speech recognition is local. Plain Whisper apps pass this test, and voice typing (Win + H) fails it by design. The same test works for phone and Mac apps; there’s a cross-platform list in speech-to-text apps that work completely offline.
Tips for better accuracy with any Windows tool #
- Use a headset or USB mic. Laptop mics sit far from your mouth and near the fan and keyboard. A headset boom or a desk mic a few centimetres away makes a big difference.
- Dictate in a quiet room. Background noise causes more errors than accents do.
- Speak in full phrases. Most engines do better with a sentence than with stop-start words.
- Pick the model size to suit your hardware. Whisper’s small and medium models run on most laptops, and the large models are much faster with a GPU.
Frequently asked questions #
Does Windows 11 have offline speech to text? #
Yes. Voice Access, available in Windows 11 version 22H2 and later, handles dictation and PC control without an internet connection. Voice typing (Win + H) is a different feature and needs a connection.
Is Windows voice typing private? #
It sends your speech to Microsoft’s Azure speech service to be transcribed, so the audio leaves your PC. If you want dictation that stays on the device, use Voice Access or a local app like Handy.
What’s the best free offline speech-to-text software for Windows? #
For dictation, Voice Access if you also want to control the PC by voice, or Handy for the most accurate plain dictation. For recorded voice notes and files, Buzz. All three are free, work offline and need no account.
Can I transcribe an M4A voice note on Windows without uploading it? #
Yes. Buzz and Vibe open M4A, MP3 and WAV files and transcribe them on your PC. Voice Access and Handy can’t, because they only listen to your microphone.
Do I need a powerful PC for offline speech recognition? #
No. Voice Access, Parakeet and Whisper’s smaller models run on ordinary laptops. A dedicated GPU makes the largest, most accurate Whisper models much faster, but it isn’t required for voice notes or dictation.