Compression of a Slovak speech database using harmonic, noise and transient model

In this article, we propose an improved analysis/synthesis model for speech that effectively parameterizes and compresses the speech signal. The method is based on HNM (harmonic plus noise) model, which is extended by transient model. However, the method for noise modeling used in our HNM system is different than in the classical HNM model. Moreover, as mentioned, to process sounds like plosives, the transient model was added. The whole system allows us to compress the parameterized speech into a format, in which it is easy to take prosodic modifications of the speech needed for speech synthesis. This approach of speech representation and compression allows us to reduce significantly the database size of speech segments for a concatenative or diphone speech synthesis, too.