Guides
Guide cloud voice selection and speech preparation
Choose inherited or authored cloud voices, understand preparation status, and resolve credit, storage, and length problems.
An authored Cloud voice copies a personal Voice Profile's voice and speed into a Guide. You need Guide editing access and an available Voice Profile. Create one in Settings > Accessibility > Voice profiles.
Choose a voice
- Open Guide settings > Cloud voice, or the Cloud voice section in Sequence settings, Page settings, or a block's Speech tab.
- Choose a named profile.
- Wait for the selection to save.
- Save any separate wording changes and test Preview.
The selection saves immediately, independently of Save guide, Save sequence, Save page, or Save block. Canceling the surrounding dialog does not undo it. Choose another profile or the Default option to change it back.
Default clears the choice at that level and inherits the next applicable choice. For new generated speech, priority is Block, Page, Sequence, Guide, then the learner's default Voice Profile, Platform Default, and device speech. Unavailable choices can be skipped. A ready authored recording is separate and takes priority for its speech.
Later edits or deletion of the personal profile do not update the Guide's copied choice. Select the updated profile again when needed. Retiring a Platform Voice can prevent new generation without invalidating already-ready audio.
Prepare saved speech
Saving relevant Guide changes marks speech for preparation. It can run after an editing pause or when you leave editing; changing Cloud voice also requests preparation.
Open Guide settings > Cloud voice to inspect the status and Speakables resolved count. Reopen the section to refresh an old display. Try again on a selector error reloads its configuration; it is not a dedicated regenerate-all button.
| Status | What to do |
|---|---|
| Speech changes pending / Speech queued / Preparing speech | Save edits, allow preparation, and check again |
| Speech ready | Test the actual speech and offline readiness |
| Insufficient AI credits | Restore the Guide workspace's available AI allowance |
| Insufficient storage | Make storage available in the Guide workspace |
| Voice unavailable | Select an available profile or return to Default |
| Speech needs attention | Read the detail and correct the affected source |
After correcting a preparation failure, leave editing to request another attempt, then reopen and check status.
Understand limits and readiness
Generated authored speech uses the Guide workspace's storage and can consume AI credits when new audio is needed. Reusing cached audio can still require storage for the Guide's copy.
Each generated speech item is limited to 5,000 characters. Shorten a long override or divide narration into separate blocks. A saved text field can accept more wording than one cloud speech item can generate.
Speech ready and the resolved count do not mean every possible utterance has a prepared audio file. Recordings, learner/device speech, and dynamic text are handled differently. Words depending on answers or other runtime values cannot all be prepared as static authored audio.
Publish and use offline
Catalog publishing can block while static authored speech is unready. Force-publish without pending generated speech, when offered, leaves those items to learner/device fallback in that immutable version. Later preparation does not fill the old published snapshot.
For offline use, refresh the Guide package after speech or default-voice changes. Check that device's readiness and test disconnected playback; Guide preparation status alone is not an offline guarantee.