3.1.1 Patch Release

Fixes a number of long standing issues:

• Introduces handling of direct speech attributes like aria-label, alt, exact-speech.
• Handling of elements like mglyph if they contain a direct speech attribute.
• Allows floats as part of mixed numbers.
• A number of fixes to the Clearspeak and Nemeth rule sets.
• Fixes regression of SVG highlighter.

German Localisation

German Localisation

This release includes the German localisation of MathSpeak and ClearSpeak rule sets, sponsored by ETH Zürich.

Other changes

• New formalism for speech rule definitions. This reduces the number of classes and object used for handling rule sets, leading to a size reduction of the library files. Rule sets are also split by locale in the code.
• Improved treatment of English units, in particular inclusion of more units from the US customary measurement system.

Bug fixes

• A number of additions/changes to the Nemeth rules.
• Improved semantics of singleton sets.

Patch for Nemeth numbers

Fixes a regression on Nemeth number output.

Patch for CLI

Patch release for breaking changes in the commander package. SRE relies on this for running on command line interface. This means that most older versions of SRE installed from npm will no longer run on the command line!

If you want to run older versions of SRE, please downgrade the commander package to v5.1.0 by running

npm install commander@5.1.0

Patch for CLI

Patch release for breaking changes in the commander package. SRE relies on this for running on command line interface. This means that most older versions of SRE installed from npm will no longer run on the command line!

If you want to run older versions of SRE, please downgrade the commander package to v5.1.0 by running

npm install commander@5.1.0

3.0.1

Some patches to work better with units:

• Introduces currencies as explicit units
  • Mathspeak follows regular unit rules, speaking currencies in the place where they occur.
  • Clearspeak introduces three currency preferences: Currency_Auto, Currency_Position, Currency_Prefix
Auto: All currencies are spoken after the value
    Position: All currencies are spoken in the position where they occur
    Prefix: All currencies are spoken before the value
• Adapts some of the rule set. In particular
  • cleans up the Mathspeak rules to follow grammar rules.
  • corrects some Clearspeak rules to fire on units.
• Adds many new imperial measures, mainly those provided in #364.
• Provides improved heuristic treatment of unit expressions.

Version 3

This is the next major SRE release. It contains a number of new features and changes.
Possible breaking changes over 2.X are indicated in bold.

Nemeth Braille Implementation

• A full implementation for linearized Nemeth Braille translation
• Includes full support for math alphabets on Unicode Plane 1.
• This was sponsored by the Big Ten Academic Alliance

Multiple Modalities

• SRE's constraint system and indexing structure now support different modalities
• Most important modalities are Speech and Braille, others include Prefixes and Summary speech
• Engine option modality, command line option --modality or -b

Unicode Characters

• JSON mappings for Unicode characters are now combined into single file. One per locale!
• Selective loading of mappings, only when necessary. This means:
  • Locale en is loaded by default.
  • All other locales are only loaded when the engine locale option is reconfigured and if they have not yet been loaded.
  • In asynchronous operation (i.e., modes async or http) one has to wait for the engine to be ready (either with engineReady API function or sre.Engine.isReady() method) not just once on load, but every time the engine is reconfigured with setupEngine.
• Automatic generation of speech rules for all alphabets (Latin and Greek) and digit sets, including those on Unicode Plane 1.
• Major cleanup of English character mappings.

Speech Rules

• Deprecates all non-semantic speech rules sets.
• Improved message handing in speech rule sets for easier localisation.
• Clean up of ChromeVox speech rules; there are now only two styles: default which corresponds to the old short style and alternative. The old default, which was very verbose, has been dropped.

Build System

• Support for the new unicode single file structure. They need to be cleaned (make clean) to be automatically rebuilt when changes to the mappings are made.
• Uses latest versions of development libraries.

Tests

• Speech rule tests are now grouped by locale.
• Tests can be called separately by providing FILE or LOCALE parameter to make.

Others

• Exposure of navigation structure is now possible in enriched MathML.
• A number of bug fixes.

Final Final Beta version

Absolutely last v3 beta due to issue #348 .

Final beta

Final beta before deprecating features for the 3.0 release.

Beta 10

• Major cleanup of Unicode mappings
• Single JSON file per locale
• Dynamic loading of locales on a need only basis

Note v3.0.0-beta.9 was published erroneously and should not be used!