Overview

TTS-based production of DAISY material is one of the most popular feature of the Pipeline 1. The Pipeline 2 is a good opportunity to redesign the TTS production system to better integrate with recent standards (SSML, PLS) and meet the users needs in terms of configurability and extensibility.

A general approach is to split the conversion process in smaller dedicated steps, rather than handling everything in a massive black-box step. This granular approach was also taken in the existing Pipeline 1 Narrator script, but with some limitation:

high coupling between all the steps: the steps in the narrator make strong assumption on the other steps and are hardly reusable in a different context
restriction to DTBook: most of the steps are dedicated to processing of DTBook documents and are not easily portable to another grammar (e.g. ZedAI)

For the Pipeline 2, the general objectives are:

support for multiple input document format, and especially next-generation DAISY XML (aka ZedAI)
native use/support of PLS and SSML standards
better interoperability (e.g. web services, distributed TTS engines)

TTS WG

See the TTS WG overview for more details.

Wiki pages

See:

a list of TTS-based production requirements
the notes on the TTS-based production workflow

See the full list of the pages from the TTS WG.

Meeting minutes

See the full list of meeting minutes.