Formant Frequency Estimation for Speaker

The normal resonant modes of vibration of the vocal tract are called formants. The formant is one of the most important features in speech signals, and is used for many applications, such as speech recognition, speech characterization, and synthesis. Formants frequencies represent the most immediate source of articulatory information and are critical in speech perception. Previous formant extraction methods can largely be classified into spectral peak picking, root extraction, and analysis by synthesis. The spectral peak picking methods and their variants have been widely used for a long time because of low computational complexity, but they often seriously suffer from the peak merger problems, where two adjoining formants are identified into a single one. In this paper, we propose a new formant extraction algorithm that the short term spectral. In the proposed algorithm, the formant candidates are found by using the short term spectral method. It requires a data segment with suitable length such that the harmonics can be resolved and the data within that length should be approximately stationary. Windowing can be done to extract the required length of the data segment.