Robust Estimation of Fundamental Frequency Using Single Frequency Filtering Approach

Abstract

A new method for robust estimation of fundamental frequency (F0) from speech signal is proposed in this paper. The method exploits the high SNR regions of speech in time and frequency domains in the outputs of single frequency filtering (SFF) of speech signal. The high resolution in the frequency domain brings out the harmonic characteristics of speech clearly. The harmonic spacing in the high SNR regions of spectrum determine the F0. The concept of root cepstrum is used to reduce the effects of vocal tract resonances in the F0 estimation. The proposed method is evaluated for clean speech and noisy speech simulated for 15 different degradations at different noise levels. Performance of the proposed method is compared with four other standard methods of F0 extraction. From the results it is evident that the proposed method is robust for most types of degradations.

Estimation of Fundamental Frequency from Singing Voice Using Harmonics of Impulse-like Excitation Source

“…Performance of the proposed method is compared with eight standard methods. The eight standard methods are SWIPE [32], YIN [9], RAPT [7] and SHRP [16], YAAPT [8], SRH [15], PEFAC [19] and SFF-CEP [33]. For all the methods, F0 search range was set between 60 − 1500 Hz according the study on singing voice in [3].…”

Mel-Weighted Single Frequency Filtering Spectrogram for Dialect Identification

“…The SFF method is used to derive the amplitude envelope of the speech signal at every sample for a given frequency [32]. The SFF spectrum has been shown to be useful in finding burst-onset points [29] and glottal closure instants [30], and it has been demonstrated to exhibit high spectral resolution for important speech features such as harmonics and resonances [27].…”

“…This architecture was chosen in the current study because it was shown in [25] to be the best performing system in dialect classification compared to two reference techniques. The spectrum computed by single frequency filtering (SFF) has been shown to give good spectral resolution to indicate harmonics and resonances [27] and good temporal resolution to model speech excitation features such as impulse-like events [28]. The SFF spectrum has also shown promising performance in determining burstonset points related to voice-onset time (VOT) and glottal closure instances compared to the short-time Fourier transform (STFT) spectrum [28]- [30].…”

Detection of Replay Attacks Using Single Frequency Filtering Cepstral Coefficients

“…The instantaneous energy for a speech segment (Figure 2 (a)) is shown in Figure 2 (c). The equation for instantaneous energy E[n] is given below [28],…”