Cognitive Signal Processing : Feature Estimation

1955 Words8 Pages
Behavioural Signal Processing: Feature Estimation in Speech Abstract Human behaviour interrelates closely with the human beings’ mental state. The behavioural information reflects communication, social interaction and even personality. To build a bridge to the human mind over engineering advances, an operational method, Behavioural Signal Processing (BSP) technology, has been introduced, which aims to analyse speech-based human behaviour. The main task of this project is the feature estimation based on BSP. In feature extraction, there are numerous analysis modes, of which the Mel-frequency cepstral (MFC) represents the short-term power spectrum of a speech signal. As MFC is about the power spectrum and requires Fourier Transform, this…show more content…
[1] Moreover, behavioural research for psychological estimations deeply depends on cautious evaluation of numerous observational indications. [5] Both the observation theory and current experimental results state that significant other’s behaviour is relevant with plentiful factors including age, gender, education background, cultural background, life experiences as well as recent mental and physical health conditions such as stress, illness, or disability. [3] Also, the affection extent between intimate relationships matters much about how people act or communicate/speak. In this project, the audio data is extracted from a dialogue in drama to simulate a marital discussion. Behavioural Signal Processing According to the main reference, “Toward automating a human behavioural coding system for married couples’ interactions using speech acoustic features”, in this project, the self-reports of observable behaviours can be extremely untrustworthy to some extent. Additionally, many of speaking recognition algorithms centres more on demonstrating objective human behaviour, for instance, the exact words, that is to say, what words or sentences people exactly say to their partners. [1] Consequently, the paper states that one of effective methods to analyse the audio data is Behavioural Signal Processing
Open Document