INTERSPEECH 2004 - ICSLP
8th International Conference on Spoken Language Processing

Jeju Island, Korea
October 4-8, 2004

Intonation Recognition for Indonesian Speech Based on Fujisaki Model

Nazrul Effendy, Ekkarit Maneenoi, Patavee Charnvivit, Somchai Jitapunkul

Chulalongkorn University, Thailand

In this paper, we proposed to use the Fujisaki parameter to distinguish between declarative and interrogative intonation in Indonesian speech. Four combinations of Fujisaki parameter were selected as the features to distinguish between declarative and interrogative intonation. The first combination is only the amplitude of last accent command. The second combination consists of the amplitude of last accent command and the magnitude of last phrase command. The third combination consists of Fb, the amplitude of last accent command, and the magnitude of last phrase command. The fourth combination consists of Fb/100, the amplitude of last accent command, and the magnitude of last phrase command. The recognition rates using the neural network were 83.33 %, 90.00 %, 50.00 %, and 96.67 % for each combination. The highest recognition rate was achieved by using Fb/100, the last accent command amplitude and the last phrase command amplitude as its inputs.

Full Paper

Bibliographic reference.  Effendy, Nazrul / Maneenoi, Ekkarit / Charnvivit, Patavee / Jitapunkul, Somchai (2004): "Intonation recognition for indonesian speech based on fujisaki model", In INTERSPEECH-2004, 2973-2976.