demo

Inputted Vietnamese text will be convert into speech using our speech synthesizer.
The system is based on unit selection synthesis. Syllables are automatically segmented by using our speech recognizer and then used as basic units in the baseline system. Moreover, the units have been expanded to phrases to produce more natural voices.
Use web demo here

Input keyworks and search for news in which the keyworks present.
This is a spoken document retrieval (SDR) system for Vietnamese broadcast news. Our system is based on a speech recognizer, which explores the use of multiple-pass decoding, for better performance on transcribing the Vietnamese speech into text.
Use web demo here

Sing or hum a melody and retrieve the songs which have similar tunes...
Our system extracts the melodic feature in users' humming (or singing) tunes and retrieves the songs which have similar melodies. There are 200 sample tunes in the database covering 158 songs of Trinh Cong Son.
Use web demo here