Saypane features

Dictation into any app

Talk instead of typing, in every app with a text cursor. Free and unlimited. More about dictation.

Hold to talk, or toggle

Hold the hotkey while you speak and release it to insert the text. Or switch to toggle mode: press once to start and once to stop, so you never have to keep a key down.

Works in any app

Word, Outlook, Teams, Slack, a browser, an AI chat, a form field. Saypane pastes the text at the cursor and then puts your previous clipboard content back.

Terminals get typed characters

In Windows Terminal, the command prompt and PowerShell, pasting behaves differently, so Saypane types the characters one by one instead. You can also switch to this method yourself.

The first word is not cut off

The microphone opens the moment you press the key and keeps a short buffer, and the speech model stays loaded, so there is no warm-up delay before you can talk.

Punctuation and capitals

Saypane adds punctuation and capital letters from how you speak. There are no spoken commands such as "comma"; you just talk.

Custom vocabulary

Teach Saypane names and terms it should spell your way, such as a colleague's surname or a product name. 10 entries free, unlimited in Pro.

Searchable history

Every dictation is kept on your PC, so a lost text can be found and copied again. The last 20 are free; Pro keeps all of them.

Honest about admin windows

Windows does not let normal apps type into programs that run as administrator. Saypane notices, keeps the text on the clipboard and tells you to press Ctrl+V, instead of failing silently. Why this happens.

Spots a blocked microphone

If Windows privacy settings block the microphone, recordings come out silent. Saypane notices the silence and points you to the right setting. How to fix it.

Transcription of files and meetings

A transcription workbench in the same app, for interviews, lectures, calls and meetings. More about transcription.

Audio and video files

MP3, M4A, WAV, AAC, WMA, MP4, MOV and other common formats that Windows can play. Video files are read for their sound track.

Verbatim text with timestamps

What was said, in order, with a timestamp on every segment. No AI summary and no paraphrasing.

Speaker labels (Pro)

Each passage is labeled Speaker 1, Speaker 2 and so on, so questions and answers are easy to tell apart. Works best on clear audio where people take turns.

Meeting recording (Pro)

Records your microphone and the sound of your PC together, so both sides of a Zoom, Teams or Meet call are captured. No bot joins the meeting.

Batch queue (Pro)

Drop in a folder of recordings and let them run one after another.

Export

TXT in the free version. Pro adds SRT and VTT subtitles, DOCX for Word, and JSON for your own tools.

Long files without loops

Audio is split at pauses, each part is transcribed on its own, and the result is checked for repeated text, with a retry where a loop is found. This avoids the repeated sentences and skipped passages that whole-file Whisper decoding is known for.

A real free allowance

The first 15 minutes of every file are transcribed for free, as a complete transcript with no watermark and no limit on the number of files.

Languages and speech models

Saypane runs open speech models on your PC and picks the right one for the language you speak. Models and hardware in detail.

Parakeet for 25 European languages

NVIDIA Parakeet TDT 0.6B v3 covers English, German, French, Spanish, Portuguese, Italian, Dutch, Polish, Ukrainian and 16 more. On our test PC it ran about 13 times faster than real time on 4 CPU threads.

Whisper for Japanese, Chinese, Korean and more

For every other language Whisper supports, Saypane uses OpenAI Whisper: the small model in the free version and large v3 turbo in Pro.

Works offline from the first start

Whisper small and the voice activity detector ship inside the app. Larger models are downloaded once, resume if the connection drops, and are checked against a SHA-256 checksum.

Speed and hardware

Saypane is built for ordinary Windows PCs, not only gaming machines. Requirements.

GPU acceleration through Vulkan

Uses NVIDIA, AMD and Intel graphics through Vulkan, which is built into the app, so there is nothing extra to install. Without a GPU, everything still runs on the CPU.

Modest requirements

Windows 10 or 11 (64-bit), a 4-core processor with AVX2 and 8 GB of memory. On slower PCs, Saypane falls back to the smaller Whisper model for dictation.

Privacy

Your voice, your recordings and your text stay on your PC. Privacy policy.

Nothing is uploaded

Audio, transcripts, history and vocabulary are stored only on your PC. The app has no analytics and no tracking.

No account

There is nothing to sign up for. Install it from the Microsoft Store and start talking.

Network only for models and the Store

Saypane goes online only to download speech models and for Microsoft Store purchases and updates. After that, it works with the network cable unplugged.

Pricing

Free to download, with free and unlimited dictation.

Pay once, no subscription

Free download with free, unlimited dictation. Saypane Pro is a one-time in-app purchase of $24.99. Pro is bought inside the app through the Microsoft Store and is yours to keep.

No nagging

No trial timer, no watermark on free results, and no "rate us" popups.