Skip to content

Releases: Nick6489/Eloquence64RS

Eloquence64RS 19.1.3-RS

Choose a tag to compare

@Nick6489 Nick6489 released this 19 Aug 17:16

Eloquence64RS 19.1.3-RS

Eloquence64RS 19.1.3 is a stable hotfix so a cancelled Speech Sequence cannot
start a new Speech Generation after Stop.

Fixes

  • Re-checks the Speech Generation before each synth-worker command, so cancel
    during the blocking voice-parameter reset no longer sends leftover text,
    indexes, or synthesize commands.
  • Stops the Host Channel from opening a new generation after Stop unless the
    next utterance explicitly prepares one. Leftover addText from a cancelled
    item can no longer become a new utterance.
  • Still starts the following utterance normally after cancel.

Installation

Remove or disable the IBMTTS add-on before installing. Eloquence64RS replaces
Eloquence64 and should not be installed alongside another Eloquence add-on.

The built-in updater follows only the production metadata maintained on the
repository's stable branch. Publishing this release promotes that metadata
from 19.1.2-RS to 19.1.3-RS.

Eloquence64RS 19.1.2-RS

Choose a tag to compare

@Nick6489 Nick6489 released this 19 Aug 02:58

Eloquence64RS 19.1.2-RS

Eloquence64RS 19.1.2 is a stable hotfix for NVDA configuration-profile
isolation when profiles use different audio output devices.

Fixes

  • Keeps Eloquence voice, variant, rate, pitch, inflection, volume, voice-shape,
    dictionary, pause, and audio-quality settings isolated between NVDA
    configuration profiles when an audio-device change recreates the synthesizer.
  • Preserves corrected profile settings across NVDA restarts instead of saving
    the outgoing profile's values into the incoming profile.
  • Supports manually activated profiles, application-triggered profiles, and
    stacked manual and triggered profiles.
  • Prevents raw boolean profile values from interrupting Eloquence speech.

Installation

Remove or disable the IBMTTS add-on before installing. Eloquence64RS replaces
Eloquence64 and should not be installed alongside another Eloquence add-on.

Eloquence64RS 19.1.1-RS

Choose a tag to compare

@Nick6489 Nick6489 released this 12 Aug 11:11

Eloquence64RS 19.1.1-RS

Eloquence64RS 19.1.1 is a stable hotfix for pronunciation dictionary profiles.

Fix

  • Pronunciation dictionary selection now follows NVDA configuration profile
    changes immediately. Switching profiles no longer leaves the dictionary from
    the previously active profile loaded in the Eloquence Host Process.

Installation

Remove or disable the IBMTTS add-on before installing. Eloquence64RS replaces
Eloquence64 and should not be installed alongside another Eloquence add-on.

Eloquence64RS 19.1-RS

Choose a tag to compare

@Nick6489 Nick6489 released this 08 Aug 14:16

Eloquence64RS 19.1-RS

Eloquence64RS 19.1 is the stable release of the enhanced-audio and dictionary
profile work tested throughout the 19.1 beta series. Standard 11 kHz output
remains the default, and Eloquence's built-in pronunciation behavior remains
available without a custom dictionary.

Highlights

  • Adds an optional Enhanced 22 kHz mode with stateful 2x interpolation,
    suppression of mirrored high-frequency energy, gentle high-frequency
    emphasis, and a subtle vocal-body lift.
  • Keeps Standard 11 kHz output bit-exact and selected by default.
  • Restores the selected audio mode correctly after changing away from and back
    to Eloquence.
  • Replaces the old shared dictionary updater with isolated pronunciation
    profiles for Eloquence's built-in behavior, alternative dictionaries,
    community dictionaries, and retained custom dictionaries.
  • Allows users to switch back to Eloquence's built-in pronunciations at any
    time, without leaving stale custom dictionaries active.
  • Downloads complete provider snapshots atomically, so updating one profile
    cannot merge it with or damage another profile.
  • Activates downloaded or selected dictionary profiles immediately and updates
    their displayed download state without reopening NVDA Settings.
  • Preserves existing custom dictionary files across migration and add-on
    upgrades.
  • Expands Python and Rust regression coverage for audio quality, dictionary
    migration and updates, live dictionary switching, native-host protocol
    behavior, and add-on installation.

Installation

Remove or disable the IBMTTS add-on before installing. Eloquence64RS replaces
Eloquence64 and should not be installed alongside another Eloquence add-on.

This release updates the native-host protocol. After updating, users who need
Eloquence on logon or other secure screens must open NVDA Settings >
Eloquence
and run Copy Helper to System Config (for Logon Screen) again.

The built-in updater follows only the production metadata maintained on the
repository's stable branch. Publishing this release promotes that metadata
from 19.0-RS to 19.1-RS.

Eloquence64RS 19.1-RS Beta 2

Pre-release

Choose a tag to compare

@Nick6489 Nick6489 released this 18 Jul 11:06

Eloquence64RS 19.1-RS Beta 2

This beta refines the optional Enhanced 22 kHz audio mode in response to user
feedback. It adds a broad, subtle 0.4 dB vocal-body lift centred at 1.4 kHz,
restoring a little fullness while preserving the Rust implementation's
articulation and high-frequency image suppression.

The filter remains stateful across audio callbacks and resets between
utterances and quality changes. Standard 11 kHz audio remains bit-exact and is
still the default. This build also includes the synth-reselection fix from
Beta 1.1.

The production updater remains pinned to the stable branch and does not offer
this beta automatically.

Eloquence64RS 19.0-RS

Choose a tag to compare

@Nick6489 Nick6489 released this 17 Jul 12:06

Eloquence64RS 19.0-RS

Eloquence64RS 19.0 is the stable release of the new 64-bit Rust-hosted driver.
It improves Say All reliability, modernizes the text-to-engine boundary, and
makes native-host deployment safer.

Highlights

  • Runs the 32-bit Eloquence engine behind a statically linked Rust host for
    current 64-bit NVDA releases.
  • Improves Say All progress, cancellation, and end-of-stream reliability.
  • Builds voice-appropriate engine text before the transport layer, keeping
    language changes and queued text correctly scoped.
  • Corrects punctuation pause handling and preserves NVDA chunk boundaries.
  • Warns when the helper copied into NVDA's systemConfig is older than the
    helper shipped with the installed add-on.
  • Includes expanded regression coverage for text encoding, pause policy,
    language scoping, native-host protocol behavior, Say All, and secure screens.

Installation

Remove or disable the IBMTTS add-on before installing. Eloquence64RS replaces
Eloquence64 and should not be installed alongside another Eloquence add-on.

After updating, users who need Eloquence on logon or other secure screens must
open NVDA Settings > Eloquence and run Copy Helper to System Config (for
Logon Screen)
again.

The built-in updater follows only the production metadata maintained on the
repository's stable branch. Development and prerelease work is not offered to
installed users until it is deliberately promoted there.

Eloquence64RS 19.1-RS Beta 1.1

Pre-release

Choose a tag to compare

@Nick6489 Nick6489 released this 17 Jul 14:28

Eloquence64RS 19.1-RS Beta 1.1

This beta fixes the Enhanced 22 kHz audio mode after Eloquence is reselected
or selected again after switching synthesizers. A newly launched Rust host now
restores the retained audio-quality setting before NVDA opens its matching
audio player, preventing 11 kHz PCM from being played at 22 kHz with an
extremely high pitch.

The Enhanced 22 kHz mode remains otherwise unchanged from Beta 1. Standard
11 kHz audio remains the default, and the production updater remains pinned to
the stable release.

Eloquence64RS 19.1-RS Beta 1

Pre-release

Choose a tag to compare

@Nick6489 Nick6489 released this 17 Jul 13:55

Eloquence64RS 19.1-RS Beta 1

This beta introduces the optional Enhanced 22 kHz audio mode. The new mode is
an original Rust DSP implementation with speech-presence shaping, stateful
polyphase interpolation, and suppression of mirrored high-frequency energy.

Standard 11 kHz audio remains the default and preserves the engine's native
PCM. Enhanced mode is intentionally being released as a beta so its sound can
be evaluated across voices, rates, languages, audio devices, and extended NVDA
sessions before 19.1-RS is promoted toward a release candidate.

The production updater remains pinned to the stable branch and does not offer
this beta automatically.

Eloquence64RS 19.0-RS-RC3

Pre-release

Choose a tag to compare

@Nick6489 Nick6489 released this 14 Jul 04:20

Eloquence64RS 19.0-RS-RC3

This release candidate improves Say All reliability, modernizes the text-to-engine
boundary, and makes native-host updates safer to test on secure screens.

Changes since RC2

  • Moves Eloquence Text construction into a pure, independently tested builder
    that produces voice-appropriate engine bytes before the Rust transport layer.
  • Removes mutable transport-layer language encoding state so language changes
    cannot select the wrong encoding for queued text.
  • Corrects BreakCommand pause durations at speech rates exactly 43, 60, and 75.
  • Preserves NVDA's chunk separator before full month names, avoiding speech
    concatenation introduced by the phrase-final em-dash workaround.
  • Statically links the native host C runtime for more reliable deployment.
  • Warns when the helper copied into NVDA's systemConfig differs from the helper
    shipped with the installed add-on.
  • Adds regression coverage for text encoding, pause policy, language scoping,
    native-host protocol behavior, Say All progress, and secure-screen helper
    detection.

Installation

Remove or disable the IBMTTS add-on before installing. Eloquence64RS replaces
Eloquence64 and should not be installed alongside another Eloquence add-on.

After updating, users who need Eloquence on logon or other secure screens must
open NVDA Settings > Eloquence and run Copy Helper to System Config (for
Logon Screen)
again. RC3 will warn after startup if the copied helper is out
of date.

Release-candidate status

Please test normal reading, Say All in browsers and documents, automatic
language switching, punctuation pauses, synthesizer switching, cancellation,
secure screens, and different audio devices. Report stalls, missing or repeated
speech, clicks, timing problems, or crashes with an NVDA debug log and clear
reproduction steps.

Eloquence64RS 19.0-RS-RC2

Pre-release

Choose a tag to compare

@Nick6489 Nick6489 released this 05 Jul 23:51

Eloquence64RS 19.0-RS-RC2

This release candidate fixes text compatibility and update delivery issues
found while testing RC1.

Changes since RC1

  • Converts capital German sharp S () to lowercase ß before legacy
    code-page conversion so Eloquence speaks it instead of a question mark.
  • Moves update checks to the Eloquence64RS release repository.
  • Allows release-candidate installations to discover newer prereleases while
    keeping stable installations on stable releases.
  • Documents that the secure-screen helper must be copied again after a host
    upgrade.

Installation

Remove or disable the IBMTTS add-on before installing. Eloquence64RS replaces
Eloquence64 and should not be installed alongside another Eloquence add-on.

After updating, users who need Eloquence on logon or other secure screens must
open NVDA Settings > Eloquence and run Copy Helper to System Config (for
Logon Screen)
again.

Release-candidate status

Please report speech stalls, synthesizer switching failures, clicks or other
audio artifacts, audio-device issues, and secure-screen failures with an NVDA
debug log and reproduction steps.