8.7 Edit A Named Voice

Open a palette through C-e H v p, select a voice, and press e. To tune the voice used by the captured source item, use C-e H f t. This changes the shared named entry, not a rule for that one item. q offers to retain unfinished work and returns to the menu.

Saving an inherited voice creates a complete entry in the destination palette. Editing a built-in palette requests a personal child. Opening and previewing do not change live feedback or save settings.

default controls ordinary reading and labels without an explicit voice. Choose it to change normal text’s physical voice, fallbacks, or tuning. Each palette has an effective default, inherited until locally edited. Switching palettes changes it too. Other entries such as indent and bolden retain their own settings; editing default is not a whole-palette voice change.

8.7.1 Adjust and compare one voice

The editor reads field labels in the normal navigation voice. Move between fields with n, p, the up/down arrows or TAB. Left/right adjusts a numeric field and moves between other fields. Press RET to activate a control or enter a value.

Lowercase d, or blank numeric input, requests the adapter’s default for that adjustment. Entering zero instead requests an explicit value, such as the lowest pitch or leftmost pan. A default and zero may sound alike for a particular effect, but they remain different settings.

More adjustments and effects exposes filters, gain, pan, and optional controls. Point stays on its button and speech says “More adjustments shown below” or “More adjustments hidden”. Move down or use TAB to enter exposed controls, and activate the button again to hide them. Changed fields state earlier values; unsupported settings stay visible and retained. Field navigation stops at its ends without wrapping, and numeric limits announce minimum or maximum.

P previews the edited voice; B compares captured original and edited versions with the same sample and spoken labels. T changes sample text and S stops. An automatic sample follows adjustments unless Automatic sample after adjustment is off. The value precedes it, for example “1. The quick brown fox…”.

That toggle announces on/off with distinct cues. More adjustments and effects and Fallbacks use open/close cues and optional state labels on both toggling and navigation. Navigation stops a sample. u undoes an edit, including reset. q returns to the originating view, offering to retain or discard unsaved work.

Physical voice searches engine or voice names. Audition with working adjustments before accepting; cancellation keeps the earlier choice and returns to the field. Fallbacks allows replacement, reordering, removal, individual auditions, appending, or returning to Automatic. Unavailable, engine-default, and property choices are retained. Shared adjustments supply inherited fields; Customize this voice provides optional per-choice settings.

8.7.2 Tune through the focused menu

From C-e H v p, enter a palette with RET, select a voice and press e to open its editor. Press m there for focused tuning. This menu edits the same retained draft. Up/down selects controls, left/right changes sections, and Shift-left/right adjusts relative rate, pitch, pitch range, or gain by one unit. RET enters an exact value. d restores the selected adjustment to adapter default, or asks which one to reset. Zero remains explicit.

f selects an existing fallback, h returns to shared settings, and the title identifies scope. For a fallback, i inherits the shared value and d requests adapter default. Play and Compare use the selected scope.

P previews, B compares original/edited with one sample, S stops, and T changes text. These previews are private and exclude contextual rules. Adjustments speak values without automatic samples here; the full editor’s automatic-sample preference is retained. u undoes the last draft change.

e opens engine-specific settings, waiting for a qualified runtime catalogue before permitting edits. Its q returns to the common editor; m reopens focused tuning on the same draft and scope. Physical experiments, individual fallbacks, and engine controls remain available in the full editor.

w saves and applies; c saves to the collection without selecting. Existing destination prompts, inheritance, and recovery remain. q stops playback and returns to the editor with the draft. Leaving that editor still offers keep, discard, or continue choices.

8.7.3 Save the draft and check application

w or C-c C-c saves and applies a complete named-voice draft. Save buttons follow settings, previews, and change details. Save to collection without selecting saves an inactive palette without changing selection. The header separates unsaved changes, partial saves, applying, failures, and saved/applied.

Retry keeps the proposal and records which saving steps already succeeded. If you make more edits while an earlier Apply is still waiting for a reply, that reply does not save the later edits. They remain unsaved. If a file or destination has changed, review the proposal again rather than expecting the editor to merge those changes automatically.

Complete-choice previews require Omnivox’s complete-voice-preview capability. They do not select the draft or change live settings. Older hosts report this unsupported; individual physical auditions remain available. Playback evidence identifies accepted audio when reported, including unsupported adjustments, without inferring a successful fallback from inventory or claiming exact native parameter measurements.

q returns to the originating row and column. Keep, discard, or continue with unsaved work. Retained drafts can be resumed through C-e H p d during this Emacs session only. They do not survive restart, and leaving does not undo a completed save.

Save and apply selects the palette for this session. U, Use at Startup in Voice Palettes, retains its startup selection separately. Saving can involve multiple files; partial failures retain the proposal for retry. See Recovering Unfinished Changes. Physical experiments first request a destination; see Keep an Installed Voice Experiment.