libOpenMary

Summary

The blueprint is a quick outline of the initial steps in bringing the Java MARY TTS engine to the Ubuntu operating system via a simple C++ API.

Release Note

This API for libopenmary is an object-orientated version of the REST API found on the MARY TTS web API. This is designed to provide the same API that MARY does and add an additional that MARY doesn't; output.

Rationale

Such a proficient TTS should be packaged with Ubuntu and included perhaps in the repositories of Debian due to its ability to articulate English, German and Dutch clearly.

User stories

No stories applicable at the moment.

Assumptions

This library assumes that MARY is up and running.

Design

We're aiming for a pretty OOP-style application and work is going to be done to import this to Speech Dispatcher.

Implementation

  • Work To Be Done:
    • Complete API rendering (in openmary.cpp)
    • Perfect audio output
    • Permit conversion to different audio types (ie: OGG, MP3, etc.)

Effort

All work visible on the Launchpad page.

UI Changes

No UI is implemented in this application.

Migration

There's a bit of speculation about bringing D-Bus and audio playback into a forked version of MARY; this perhaps bringing a more native experience with MARY and thus eliminating the need to use this library.

Test/Demo Plan

There's a CLI included with this library, feeding text should trigger playback.

Unresolved issues

  • Audio playback needs to be perfected.
  • Perhaps implementing a daemon to handle this to be the D-Bus stand-in?

BoF agenda and discussion


CategorySpec

SpeechControl/Blueprints/libOpenMary (last edited 2011-03-13 22:59:19 by pool-96-224-24-113)