Parrot Scribe 1.1.0 replaces 1.0.8. It is rebuilt in Rust and Tauri with new on-device speech models. Your library, supported settings, Pro license and Recovery Key carry over. Before you update, check the spoken languages, export formats, speaker recognition and deletion changes below. ### New - New speech models, and playback highlights each word as it's spoken. - Choose which microphone to record from, or follow the system default. - Import WebM, Opus and MKV files. - In Identify Speakers, mark two or more voices as Same Person. Set the name used for you in Settings > General > Your Name. - Reassign microphone paragraphs to another speaker, not only system-audio paragraphs. - Export a whole session's audio with Audio as WAV…, and save a limited summary of one person with Export Speaker Summary…. - Before you change audio retention, see how many recordings and how much audio the change would delete. - Find in a transcript highlights every match and steps through them. - An AI app can start Parrot Scribe in the background when it needs it. - Settings lists the spoken languages that Auto supports. ### What works differently - **Spoken languages.** Auto supports the 28 languages NVIDIA lists for the new model: Arabic, Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Hindi, Hungarian, Italian, Japanese, Korean, Mandarin Chinese, Norwegian Bokmål, Polish, Portuguese, Romanian, Russian, Slovak, Spanish, Swedish, Turkish, Ukrainian and Vietnamese. Accuracy varies by language. 1.0.8 also covered other languages, such as Greek, Hebrew, Catalan, Indonesian and Thai. If you record in a language that isn't listed, stay on 1.0.8: in its Updates settings, set Automatic Updates to Notify when available, then choose Skip This Version when 1.1.0 is offered. - **Speech models.** Before your first recording, Parrot Scribe downloads its models: about 0.88 GB with Auto or 0.83 GB with English Only. The 1.0.8 models aren't used, and there's no choice of model size. Models now download from downloads.parrotscribe.com, the same host as app updates, instead of a third-party site. Expected Languages and the advanced transcription tuning controls are gone. - **Speaker recognition is off after the upgrade.** Saved names carry over, but voice profiles don't. Turn on Recognize Saved Voices in Settings > Transcription, then save a new sample for each person. Saving a voice profile asks you to confirm. The 1.0.x consent-prompt setting, including when your organization manages it, doesn't carry over. - **Deleting.** Forget Speaker is now Delete Speaker and Sessions…. Deleting a session or speaker removes it from the 1.1.0 library, but it stays in the preserved 1.0.x library and in any backups, and 1.1.0 can't remove that copy yet. - **AI apps.** Reconnect each app in Settings > AI Apps, because earlier approvals don't work with 1.1.0. Claude Code and the Codex CLI need a new token from Advanced. The app shown as ChatGPT in 1.0.8 is listed as Codex. Access is Simple or Trusted, and each reconnected app starts with Simple: Simple refuses deleting instead of asking each time, so grant Trusted again where you used it. The Allow AI apps switch and the Disabled level are gone. Disconnect an app, or use Revoke Token under Advanced for a command-line client, to remove its access. Usage guidance now comes from Parrot Scribe itself, without a separate skill. - **Encryption.** Recordings, transcripts, settings, folders, speakers and AI-app grants use a new encrypted format under one library key. A small index lists session IDs and start times without encryption; it holds no titles, transcript text or names. - **Updates.** Include prereleases replaces the update channel. Your 1.0.8 automatic-update choice stays in effect but no longer appears in Settings. - **Audio retention.** To clear all retained audio, turn audio retention off; new recordings then keep no audio. - **Diagnostics.** You review the diagnostics report before saving it and attaching it to a support email. ### Not in 1.1.0 - TOON, PDF and Transcript Bundle export, the default export format setting, and Share Text or Share Audio Range. Transcripts export as Markdown or plain text, and audio as WAV. - Automatic transcript deletion and keeping uncompressed audio. If your settings used either, Parrot Scribe shows a notice until you choose Got it. Sessions stay until you delete them, and audio retention still applies. - Opening Identify Speakers automatically after a recording, Identify Speakers in the menu bar, and Skip All. - Forgetting, merging or moving several speakers at once, and Back to Speaker. - Clear All for retained audio while keeping retention on. ### Known issues - Overlapping microphone and system speech can stay grouped in long paragraphs. Audio and timing stay separate. - Activating a different license key while Pro is active leaves the old key's device slot in use. Before you switch keys, choose Manage devices, then Deactivate on the row marked This Mac. - An AI app with an older Parrot Scribe plugin may still load its Listening skill. Disable that plugin in the AI app. ### Upgrading from 1.0.x - In-app updates from 1.0.0 to 1.0.7 install 1.0.8 first, then offer 1.1.0. Choose Install and Relaunch and enter an administrator password when asked. - Before running a downloaded installer, quit Parrot Scribe from its menu bar icon and wait for recordings to finish saving. - The upgrade copies your library, including retained audio, into the new format next to the untouched original. Make sure you have free disk space for a second copy. - Sessions, transcripts, recordings, folders, comments, speaker names, supported settings, Pro access, the Recovery Key and Launch at Login carry over. The original library stays untouched as a recovery copy. - If the upgrade can't finish, Parrot Scribe says so and keeps your original library unchanged. Quit and reopen it to try again. ### Going back to 1.0.8 Turn off Launch at Login, quit from the menu bar icon and wait for saves to finish. Move Parrot Scribe out of Applications, then install 1.0.8 from parrotscribe.com/downloads. In 1.0.8's Updates settings, set Automatic Updates to Notify when available, and choose Skip This Version when 1.1.0 is offered. Turn Launch at Login back on in 1.0.8 if you want it. 1.0.8 shows your library as it was before the upgrade. Recordings and edits made in 1.1.0 stay in its library and don't appear in 1.0.8, and sessions deleted in 1.1.0 reappear. Recordings and edits you make in 1.0.8 after going back won't move to 1.1.0 if you update again, so export anything you need first.