C-e H s opens Speech settings for the captured buffer. g switches to global defaults; C-u C-e H starts there. Indentation shares this scope with its parents, including changes made inside its own menu. Opening/closing alone saves nothing.
Select Rate and adjust with Shift-left/right by the configured step, or use RET~/~r for an exact value. Speech must be running; Omnivox uses 0–100. Global rate changes also affect the current buffer. A named voice’s relative rate modifies this base as described in Voice Adjustments.
In Speech settings, press p for the punctuation menu. Choose n for none, s for some or a for all punctuation. That menu’s p offers the built-in levels and available named profiles. In buffer scope, r restores automatic mode policy. Press q to return to Speech settings.
The punctuation menu’s e opens the Omnivox pronunciation editor. In that editor, lowercase s saves overrides on the speech host and lowercase r restarts speech after confirmation. Its q returns to the same punctuation menu item, retaining the draft and selected character for reopening.
Capital S in the punctuation menu has a different purpose: it saves the selected level or profile through Customize. It does not save the pronunciation editor’s host-side draft.
S in Speech settings saves displayed rate and punctuation as startup defaults. Rate uses the backend’s startup option. Saving from buffer scope promotes those displayed choices to defaults. Mode punctuation policy and explicit buffer settings still override the saved punctuation default.
For word or phrase substitutions managed by Emacs, open C-e H s p d. This menu is separate from the Omnivox pronunciation editor on e. Its dictionaries are stored in an Emacs dictionary file, not Customize or the speech host, and changes need no speech restart. Opening the menu writes nothing and leaves the source’s pronunciation table unchanged.
Press d to define an entry at the captured source item. Choose buffer, file, directory or major-mode scope, then enter the term and spoken replacement. The existing prompt can load saved dictionaries first. Buffer entries last only in that buffer’s current pronunciation table. Other scopes update the in-memory dictionary collection and need a separate w Save to persist.
Use e to choose an existing dictionary and open its widget editor. TAB moves between fields and buttons; RET activates them. Field edits update the in-memory collection. Its Save button writes dictionaries to a chosen file. Return with C-c C-c from anywhere, or q outside editable fields; inside a field, q inserts text. Returning retains edits without saving or rolling them back, and reopening resumes the same editor buffer.
Back in the menu, r rebuilds and enables the captured source’s table from its dictionaries. This replaces temporary buffer-only entries. t turns substitution off or composes the table again when turning it on; turning off also loses buffer-only entries. Changes to file, directory and mode entries remain in the collection. To reverse an entry change, edit it back and refresh the source; save again if you changed the saved dictionary.
Finish and close existing widget editors before loading replacements: their
retained fields are not reloaded. The menu’s l loads a dictionary file into
the collection; refresh the source afterward to use those entries. w saves file, directory and mode dictionaries,
excluding buffer-only entries. Both actions prompt for a file; Load offers
emacsvox-pronounce-dictionaries-file, while Save offers its basename in the
Emacsvox user directory. Saving elsewhere does not change the configured
startup file. Use ? followed by an item’s key
for its scope and effects before applying it.
Open C-e H s i or directly C-e d I. In global-default scope, S saves indentation defaults and tone volume through Customize. Select Volume and adjust by five percentage points with Shift-left/right, or enter with RET. Plain left/right changes sections and SPC previews indentation. Other selected items give guidance rather than treating every shifted arrow as volume.
t opens pitch-tone settings: base pitch, rise per column, maximum pitch, and requested duration. Shift-left/right adjusts; RET enters exactly. SPC previews four columns without changing presentation or enabling feedback. s stops. Short durations show the effective blank-line-tone minimum. R restores the four bundled values in scope. Global changes preserve local overrides; S saves global defaults and tone volume. q retains live changes.
C-e H s c opens Capitalization. i toggles indication, p chooses spoken, tone, combined, personal-rule, or no presentation, and c independently toggles camel-case splitting. They share buffer/global scope on g. SPC previews reading; S saves these three startup defaults. Buffer overrides still apply.
Within Capitalization, l edits the capital-letter beep and a the all-caps beep. Each has a separate global draft. Select pitch/duration and adjust by 25 Hz/5 milliseconds with Shift-left/right, or use RET. SPC previews only the beep and s stops. Previewing neither changes reading settings nor enables capitalisation indication.
In a beep editor, S saves that cue globally for tone/combined presentation. r restores bundled values in the draft; save to remove this menu’s override. D discards/rereads, q returns, and C-q closes menus while retaining drafts for this session. Unrelated rules and labels survive, and stronger rules may override the cue. Finish unsaved personal-rule editors before cue saving. If saved rules changed elsewhere, discard the cue draft and reopen it.
C-e H s f opens Spelling feedback. These settings are global. Toggle v for voice changes, t for a tone, and e for the help earcon. You can combine them or turn all three off without disabling Flyspell checking and highlighting. The default is voice plus earcon.
When you read text already marked by Flyspell, tones and earcons occur at the flagged word. Voice feedback uses the existing spelling face mappings. Tone and earcon feedback also report errors detected while editing. Character movement cues entry into a flagged word: its first character moving forwards, or its last character moving backwards. Moving within the same word or onto an adjacent space does not repeat the entry cue. Opening a mail reply checks and highlights inserted text without a burst of cues.
Use m to choose tone timing. Alongside speech is the default: it plays the tone while the word is spoken without adding a tone pause. Before speech finishes each tone before its flagged word, adding a pause. Previously saved choices remain in effect. Arrows browse without changing the setting; RET applies and returns to Timing, while q cancels. Timing leaves voice and earcon feedback unchanged. Compare it with SPC or by reading flagged text: a tone-only preview or editing cue has no accompanying speech to overlap.
Select p for pitch, d for duration, or l for spelling tone volume. RET enters an exact value; Shift-left/right adjusts by 25 Hz, 5 milliseconds, or 5 percentage points. Spelling tone volume ranges from 0 to 100 percent: zero silences the tone, and the default of 100 preserves its original level. V changes the shared volume of all main-stream tones, including indentation and spelling. The shared volume still scales the spelling tone after its individual adjustment; it is also available in Sounds and output.
These controls change the current session immediately. b previews the tone alone, even when tone feedback is off. SPC reads a sample containing one flagged word with the selected combination. S saves the spelling settings, including tone timing and shape, and the shared tone volume through Customize. q returns and C-q closes the menus without saving.
Press capital E in Spelling feedback, or use C-e H s f E, for a private draft of the complete tone. The title identifies its global spelling scope. This editor adds waveform, pitch sweep and envelope controls to pitch, duration and individual volume. Unlike the parent menu’s immediate adjustments, changes here affect only the draft until you apply it. Timing stays in Spelling feedback at C-e H s f m. Voice, earcon and tone enablement also remain in that menu; shared main-tone volume controls all main tones from Sounds and output.
The sections follow the editing task: the spelling tone’s basic controls, Pitch movement, Envelope, Preview, Presets, Apply and save, then Finish. Left/right changes sections; up/down selects an item. Shift-left/right adjusts only a selected number. RET enters an exact number, toggles Pitch sweep, or opens the selected choice list. In a list, up/down browses without changing the draft; RET chooses and returns, while q cancels. Waveform and Envelope are choices, so shifted arrows do not adjust them. ? followed by an item’s key explains its effect, units and dependencies. Use C-c l or the other menu review keys to inspect an announcement without activating it.
These previews privately enable the tone even if live spelling tones are off. They do not turn the live cue on, apply adjustments or save defaults. A tone alone cannot demonstrate Before speech versus Alongside speech; use capital P for that comparison. Shared tone volume still scales playback, so a shared volume of zero silences these previews too. After Apply, C compares against the newly applied tone, not the editor’s opening value.
p sets pitch; Shift-left/right changes it by 25 Hz. Choose w for waveform: sine is a smooth pure tone, triangle adds a softer harmonic character, and square sounds brighter. Pulse offers adjustable width; Sawtooth has a bright, buzzy character. Their perceived loudness can differ; compare at a comfortable individual volume.
For Pulse, select u for Pulse width. This is the percentage of each cycle spent high: 50 percent sounds like Square. Shift-left/right changes it by five percentage points within 5–95 percent, and RET enters an exact value. Only Pulse uses width. Trying another waveform keeps the width in this draft, so you can return to Pulse and continue comparing.
Toggle t for Pitch sweep to make pitch rise or fall during the tone. The ordinary Pitch control sets the start; f sets End pitch. Shift-left/right changes the end by 25 Hz and RET enters it exactly. Equal start and end values keep constant pitch. Turning the sweep off retains End pitch in the draft. The sweep occupies the existing duration and does not change timing relative to speech.
Pitch and End pitch range from 50 to 10000 Hz. Duration ranges from 5 to 1000 milliseconds and steps by 5; individual Volume ranges from 0 to 100 percent and steps by 5 percentage points. Zero volume silences this tone. Waveform changes can alter perceived loudness even at the same volume.
Choose e for the envelope, which describes how loudness changes over the life of the tone. Classic uses the original short fades. ADSR uses four controls, with times in milliseconds and sustain in percent of peak level:
For example, with Duration 100 milliseconds, Attack 10, Decay 20 and Release 30, the tone holds its Sustain level for the remaining 40 milliseconds. A Sustain value of 40 means 40 percent of peak loudness during that hold, not a 40-millisecond duration. A longer Attack softens the onset; a short Attack followed by a longer Decay can make a plucked sound.
Shift-left/right changes each timed stage by 5 milliseconds and Sustain by 5 percentage points. RET accepts stage times from 0 to 60000 milliseconds and a Sustain level from 0 to 100 percent. The duration of the whole tone remains the limit, even when a stage requests a much longer time.
Total duration includes the whole envelope. If attack, decay and release would exceed it, they shorten proportionally and leave no sustain hold. Zero skips a timed stage; endpoint smoothing still limits clicks. Classic ignores these four controls, retaining their values while this draft remains open. The initial waveform is sine with Classic fades.
With ADSR, k chooses the envelope curve. Linear uses straight ramps and preserves the original ADSR behaviour. Exponential starts the attack more slowly and starts decay and release more quickly, while retaining the same stage durations and endpoint levels. Arrows browse the choices; RET selects and returns, and q cancels. Classic ignores the curve but retains its choice in the draft.
Preview after each small change with SPC or its equivalent, b. Capital S stops immediately without waiting for the Release stage. Compare a short tone with the spelling sample too: a shape that is clear alone can be masked by speech or make the word harder to hear.
Capital L opens presets. Arrows browse without changing the draft; RET loads a copy and returns, while q cancels. Soft tick, Rounded beep and Short pluck are starting points with different pitches and envelopes, each at 35 percent individual volume. Use ? followed by a preset’s displayed key for its character. After loading it and returning to the editor, press SPC to hear the draft. The preset list does not preview while you browse. Further adjustments leave the preset itself unchanged.
Capital N saves the draft under a name in your personal preset collection through Customize. This does not apply the tone or save the spelling cue defaults. Reusing a personal name asks before replacing it; bundled names cannot be replaced. Saved presets appear alongside the bundled choices at L.
Use A to apply the draft for this session. Capital W saves and applies its tone defaults through Customize. These actions leave timing, voice, earcon, tone enablement and shared volume unchanged. R resets only the draft to factory defaults for preview and deliberate application. q returns to Spelling feedback and discards unapplied edits; applied and saved changes remain. Reopening starts from the current tone. To reverse an applied adjustment, edit it back and apply again, saving too if you changed the saved defaults.
Apply and Save store the active tone shape. Unused Pulse width, End pitch and ADSR values remain available while editing this draft, but reopening reads the applied shape; it need not recover those inactive values. A named preset stores the complete draft for later reuse.
The keys have different jobs in the two menus: capital S stops playback in the tone editor, but saves settings in the parent Spelling feedback menu. Capital W in the editor saves only tone defaults. To keep a choice such as voice off and tone on, return with q and save with S in Spelling feedback. That parent Save also includes timing, earcon selection and shared tone volume. Saving a preset with N stores neither those choices nor the live cue defaults.
Individual volume below 100, harmonic waveforms and ADSR require an Omnivox worker with richer-tone support. Pulse, Sawtooth, sweeps and exponential curves require extended-tone support. Older workers omit an unsupported tone while retaining speech and earcons, and the feedback explanation records the omission. A preview reports missing support instead of playing an inaccurate substitute. Your settings remain intact. Use an updated worker, or restore volume 100, sine and Classic and turn off Pitch sweep to hear the tone on an older worker.
C-e H s t opens Typing and interaction. Character, word, completed-line echo, point indication/presentation, and spoken messages share buffer/global scope; global changes also affect the source buffer. Delayed phonetics and delay are global. Shift-left/right changes delay by 0.1 seconds; exact entry through RET accepts at least 0.1 seconds.
SPC previews the captured line with point settings, P speaks its character’s phonetic name immediately, and s stops. S saves displayed settings through Customize, promoting buffer choices to startup defaults. Closing retains live changes without saving.