WH
SOFTWARE

Whispering

Whispering is a voice-input and transcription tool that records from a shortcut and copies or pastes text; the project has moved into the Epicenter desktop host.

Version 7.11.0(独立版归档)WebWindows/macOS/Linux 归档包Epicenter 桌面宿主AGPL-3.0-or-later

Current project path

Whispering started as an independent desktop voice-input application and is now presented as an application inside Epicenter. The current path can use a browser interface or an Epicenter host for global shortcuts, native paste and local GGUF transcription. The 7.11.0 desktop package should therefore be labeled as an archive rather than assumed to be a long-term standalone release.

Transcription and data paths

Local GGUF processing can keep audio on the device. Cloud providers receive audio when selected, and text polishing or conversion can send the transcript to a chosen model service. Login sync, microphone permission, clipboard access and history retention are separate settings that should be reviewed independently.

Archive boundaries

An archived package may preserve an old shortcut or workflow but can lose compatibility with current operating systems, APIs or sign-in services. Keep permissions minimal, remove API keys from test data and prefer a maintained host when a security or integration fix matters.

Maintenance note

This page was reviewed on 2026-08-23 against the archive metadata, current Epicenter path, transcription privacy boundaries and AGPL-3.0-or-later license. Test with a short non-sensitive recording before using real meeting or customer data.

SAVE TO CLOUD

Save to your cloud drive

Open the cloud drive to get the file directly, or save it for convenient access on another device.

Links checked 2026-08-06
Save first, access when you need itOn desktop, scan with the matching cloud-drive app. On mobile, tap the save button.
GUIDE

Whispering archive setup and current usage path

Decide whether the 7.11.0 archive is needed, then test microphone permission, a local or cloud transcription backend and the copy or paste workflow.

Before you start

  • Confirm whether the package is the standalone archive or the current Epicenter host, and choose the system architecture.
  • Prepare a short recording without passwords, meeting secrets or customer information.
  • Decide between local GGUF, a self-hosted endpoint and a cloud provider, including the data path and cost.
01

Installation steps

  1. 01

    Choose the current host or archive

    Use a browser path for a quick trial, or the current Epicenter host for global shortcuts, native paste and local models. Use the standalone package only for compatibility with an old workflow.

  2. 02

    Verify the system package

    Match Windows MSI or EXE, macOS Intel or Apple Silicon, and the appropriate Linux package. Do not treat an old archive as a substitute for current security updates.

  3. 03

    Grant the minimum permissions

    Enable only the microphone, shortcut and paste permissions needed for the test, then review whether any accessibility or startup permission remains unnecessary.

02

Quick start

  1. 01

    Configure the transcription backend

    Explicitly select local or cloud processing. For a cloud provider, verify the endpoint, key handling and billing before recording real content.

  2. 02

    Record a short phrase

    Use the recording shortcut, speak a short phrase and inspect the text, punctuation and language detection before testing a paste target.

  3. 03

    Review history and retention

    Check whether audio, transcripts and transformed text are stored, then clear the test data or disable history before regular use.

Usage tips

  • A browser interface does not provide the same global shortcut and native-paste behavior as a desktop host.
  • Cloud transcription, text polishing and sync are different data paths and should be reviewed separately.
  • Continued operation of an archive does not imply continued maintenance; check compatibility when login or API behavior changes.
Troubleshooting and uninstall

Why does the shortcut work but no audio appear?

Check microphone permission, the selected input device and other applications using the microphone. A browser page also needs its own microphone permission and does not provide a system-wide shortcut.

Why does automatic paste fail after transcription?

The desktop host may need an accessibility permission, and the target application may block simulated input. Test copying to the clipboard first and grant only the smallest required permission.

  1. Export or clear historySave needed text, remove recordings, transcripts and API keys that should not remain, and confirm whether sync is enabled.
  2. Remove the archive and permissionsUninstall the archived application and revoke microphone, accessibility and startup permissions. Do not remove shared Epicenter data when the current host is still in use.
FAQ

Frequently asked questions

Is the standalone Whispering desktop package still being updated?

The current project path places the Whispering interface inside Epicenter, so the standalone package should be treated as an archive and its maintenance status checked before use.

Does Whispering always upload recordings?

Local GGUF processing can keep audio on the device; a selected cloud provider receives audio, and text conversion may send the transcript to another service.

Which path supports global shortcuts?

The desktop host is the path intended for global shortcuts and native paste. A browser interface is more limited to the current page.

What should be checked before using a real recording?

Confirm the backend, permissions, storage and provider data policy with a short non-sensitive test, then decide whether history and sync should remain enabled.