FAQs

HoldSay is a desktop voice app for Mac and Windows. It turns speech into text wherever your cursor is, records your voice and the other people in an online meeting without adding a participant bot, and transcribes existing audio or video files.

Dictation, recording, transcription, speaker labels and summaries run on your own computer. There is no monthly word or transcription minute allowance, and HoldSay is sold as a one-time purchase per device seat.

Yes. HoldSay is built to work offline. After a one-time model download on first launch, dictation, meeting recording, transcription, speaker labels and summaries all run on your own Mac or Windows PC, with no cloud transcription server involved.

You can turn Wi-Fi off mid-meeting and everything keeps working, which also means flights, client sites and locked-down networks are fine. Your audio and transcripts never leave your computer. The app only goes online for small things such as license checks and automatic updates, never to process your voice.

No. There is no cloud transcription server behind HoldSay: the speech models live on your computer, and your audio, transcripts and summaries are processed and stored there. Nothing you record or dictate is ever uploaded, to us or to anyone else.

The app talks to the network only for housekeeping: license checks, usage and integrity events, and update checks. None of them carry your audio or your text, and you can verify that yourself by working with Wi-Fi off. More on how HoldSay works offline.

No. There is no sign-up, no login and no password to remember. You install the app and activate it with the license key from your purchase email; the key is checked against the number of device seats you bought, and that is the entire ceremony.

Because there is no account, there is also no profile holding your data on someone's server: recordings, transcripts and summaries exist only on your computer. If you replace your machine, email support and we move the seat.

On your own computer, as plain files you can open in Finder or File Explorer. Nothing is uploaded to us or any third party. The library sits in HoldSay Library by default. Move it under Settings → General → Local file library, and the Open button beside it takes you straight there.

Every note is mirrored into its own dated folder: Dictation/YYYY/MM/DD/ for hotkey dictations, Recordings/YYYY/MM/ for recordings. Each one holds the same set of plain files:

audio.wav · transcript.txt · speakers.txt · outline.md · meta.json

No database and no proprietary format, so your own scripts and AI agents can read your voice history directly.

HoldSay keeps all audio, transcripts and summaries on your own computer and sends nothing of them to the cloud, so the data-transfer and third-party-processing questions at the heart of GDPR largely do not arise in the first place.

That is exactly why privacy-sensitive professionals reach for it: journalists protecting sources, lawyers and therapists under confidentiality duties, and teams whose IT departments refuse cloud recorders. Your interview never becomes someone else's training data, and there is no processor agreement to chase, because no third party ever touches the recording.

No. HoldSay never charges a subscription: it is a one-time purchase, and after that the app is yours, with no monthly fee, no word or minute caps and no features held back for a higher tier.

Because the cost stops the moment you buy, HoldSay works out far cheaper over time than dictation and meeting tools that bill monthly; heavy use costs exactly the same as light use. One payment, and the meter is gone. Buy HoldSay.

HoldSay is a one-time purchase of $69.99 per device seat, currently $39.99 as an early-bird price. Each seat activates one Mac or Windows PC, and a single license key can hold 1, 2 or 3 seats in any mix of the two platforms.

There is no subscription, no word or minute cap and no feature tier above it: one payment covers unlimited dictation, meetings and transcription. Download HoldSay.

Each device seat activates HoldSay on one Mac or Windows PC. A key can include 1, 2 or 3 seats, and the seats can be any mix of the two platforms: a Mac at work plus a Windows machine at home on one key is fine. Seats are permanent, with no annual renewal and nothing that expires.

For bulk or team orders, email support@holdsay.com. Replacing a computer? Email us and we move the seat to the new machine.

Yes. Updates are free and automatic: HoldSay checks in the background, downloads quietly and applies the new version without you doing anything, so you are always on the latest build.

Updates carry improvements to dictation, meetings and transcription to the app you already own, and applying them never asks for money or an account. If an update arrives while you are working, it simply applies the next time the app restarts.

Ask within 14 days of purchase. Email support@holdsay.com from the address you used at checkout, or send a ticket from the support page and choose Billing & license.

We issue the refund through Lemon Squeezy, and the money goes back to the payment method you paid with; Lemon Squeezy says it can take up to 10 days to appear on your statement. The refund covers every seat on that license key, and the key is cancelled when the refund is issued. Your recordings and transcripts are files on your own computer and stay there. The full policy is in the Terms of Service.

Yes. HoldSay records without a bot. One click on the floating card captures your computer audio and your microphone together, locally, so it hears both sides of a Zoom, Google Meet or Teams call without anything joining the meeting or appearing to the other participants.

The same click covers in-person meetings and lectures, online or offline. When you stop, you get a transcript with speaker labels and a structured summary with the decisions and action items pulled out, and there is no cap on how many meetings you record. Here's how the card records a call.

Yes. Start a recording while a YouTube video, webinar or online lecture is playing, and HoldSay captures the computer audio locally and transcribes it as it plays.

When you stop, the recording becomes a note with the transcript, speaker labels and a structured summary. Nothing joins the session, the media does not need to be uploaded, and there is no monthly transcription-minute limit. See how to capture computer audio.

HoldSay writes captured audio to disk while it records. If the app or your computer closes unexpectedly, HoldSay detects and recovers the interrupted recording the next time it opens, so the audio already captured can still be transcribed and turned into a note.

The recovered audio and transcript stay on your computer like every other HoldSay recording.

Yes. Drag an audio or video file into HoldSay, or a whole folder of them, and everything is transcribed on your own computer. On Mac you can also right-click a file in Finder and send it to HoldSay from there.

Imported files get the same treatment as live recordings: speaker separation, a clean transcript and a structured summary, in any of the 30 supported languages. Because transcription is local and unlimited, clearing out years of old voice memos or interview recordings costs nothing extra.

Yes. HoldSay separates speakers automatically and labels who said what, so interviews, meetings and lectures come out as a labeled script instead of one wall of text. It recognizes your own voice automatically, and you can enroll a colleague's voice once and they stay labeled in every recording after that.

Renaming a speaker updates the note in your library, and separation also works on audio files you import. Speaker identification runs on your own computer like everything else, so nobody's voice is uploaded anywhere to be recognized.

Yes. Transcripts open as editable text right inside HoldSay: click in, fix a word, and the change saves automatically. Find and replace is built in, and undo works the way you expect, so cleaning up a long meeting is quick.

The text is yours in the most literal sense: select and copy it into any app, or open the plain files from your library folder, where every note keeps its transcript as an ordinary text file no database can hold hostage.

Yes. Put the cursor where you want the text, then use HoldSay in either of two ways: hold the shortcut while you speak and release it when you finish, or tap once to start and tap again to stop. Your words land in the active app with punctuation already in place.

HoldSay works in the apps you already write in on Mac and Windows, including ChatGPT, Claude, Notion, Gmail, Google Docs, Word, Slack, Discord, Telegram, WhatsApp and iMessage. There are no word counts or monthly caps, and every dictation is also saved to your HoldSay library.

No. HoldSay supports both hold-to-talk and hands-free dictation. For a quick sentence, hold the shortcut while you speak and release it to finish.

For a longer thought, tap the shortcut once to start, let go, and tap it again when you are done. Both interactions use the same on-device transcription and place the finished text at your cursor, so you can choose whichever feels natural each time.

Yes. You can hide the floating voice card whenever you want and bring it back from the HoldSay menu bar or system tray. Hiding the card does not quit HoldSay, so dictation remains available from your keyboard shortcut.

This is useful during screen sharing or whenever you want a distraction-free workspace. See how to hide and restore the card.

HoldSay transcribes 30 languages on-device, with no language locked behind a higher tier. Here's the full list, by region:

The same language support works across dictation, meeting recording and file transcription.

Yes. HoldSay can recognize multiple supported languages in one dictation, including language changes within the same sentence, without making you switch languages manually.

The same mixed-language recognition works for dictation, meetings and imported files. You can also add names, company terms and specialist vocabulary to the custom dictionary.

Yes. HoldSay works with your computer's built-in microphone, external microphones and Bluetooth headsets like AirPods.

AirPods have a classic failure mode with dictation apps: the moment an app uses their microphone, Bluetooth drops into phone-call mode and everything turns muffled. HoldSay avoids that automatically, leaving your AirPods in high-quality playback and taking speech from a better microphone instead, so your dictation stays clear and your audio does not collapse mid-sentence.

Yes. HoldSay has a custom dictionary for names, jargon, brand terms and anything else the world spells unusually. Add entries one by one in Settings, or import hundreds at once from a CSV file.

The on-device speech engine reads your dictionary while it transcribes, so a colleague's surname, a drug name or your own product comes out right every time, whether you are dictating or transcribing a recording. The dictionary stays on your computer with everything else, and you can edit or remove entries whenever you like.

Yes. HoldSay's Polish pass removes filler words like um and uh, fixes punctuation and tidies the phrasing. It works as a faithful reviser rather than a rewriter: it never summarizes, reorders or invents anything, so the polished text still says exactly what you meant.

The raw transcript is always kept alongside the polished version, and you can copy either one. That is the difference between text you have to clean up before sending and text that is ready to paste the moment you stop talking.

HoldSay runs on Apple Silicon Macs (M1 or later) with macOS 14 or later, and on x64 PCs running Windows 10 or Windows 11. Intel Macs are not supported. Windows on ARM is not supported in version 1.

On first launch the app downloads its speech models once; after that everything runs locally, and an offline laptop works fine. Both platforms include the same core workflow: dictation, meeting recording, file transcription and summaries.

No. HoldSay requires an Apple Silicon Mac (M1 or later): the on-device speech engine is built for Apple Silicon and does not run on Intel processors.

This is a hard requirement rather than a testing gap, so there is no Intel build coming for version 1. If your main machine is an Intel Mac but you also have a Windows PC, the Windows 10/11 x64 version runs the same core workflow, and one license seat can activate either platform.

Yes, where the Mac supports it. HoldSay can use Apple's on-device foundation models to produce grounded meeting summaries, keeping the whole summarization step inside your machine and consistent with Apple's own privacy design.

On Macs without Apple Intelligence, and on Windows, HoldSay's bundled local model does the same job, so summaries stay on-device everywhere. Either way, nothing about your meeting leaves your computer to be summarized.