Automatic alignment of spoken digital books
From HLT@INESC-ID
The IPSOM project had as its goal to improve access to digital spoken books. In our country, the main users of digital spoken books are visually-impaired citizens. This is not the typical situation in other countries where spoken books are much more widely used (for instance, when driving). The development of tools that allow the easy detection and indexation of units (words, sentences, topics) may help to spread even further the usage of digital spoken books.
This project resulted in several software modules:
- Automatic speech-text alignment systems with and without speaker adaptation (see demo);
- Multimedia interface for indexing digital spoken books;
- Plataform for the production of spoken books.
The Demos
- O Senhor Ventura, by Miguel Torga, read by Isabel Bahia [1]
- O Monge Desastrado, included in Histórias de Belém, Ana Maria Magalhães e Isabel Alçada, read by Jorge Moreira [2] (European Portuguese) and Juva Batella [3] (Brazilian Portuguese)
- O Monge Desastrado: [4] Parallel audio book (Audio in Portuguese / Text in English)