Fundamental frequency is a fancy term for pitch in acoustic/phonetic field. This is one of the important parameter that is used to characterise the voice.
Fundamental frequency (F0) has been used in various fields like speaker detection, emotion recognition, sex determination and also in voice pathology.
Although there are various algorithms for speaker recognition such as GMM-based, i-vector based and now a days DNN (Deep Neural Network) based, using F0 to recognise a person is still active among phoneticians who work with court caseworks.
Earlier researches suggest that the range of fundamental frequency in male between 50-250 Hz and in female between 100-400 Hz. In children the frequency can even go higher.
There have been much research on emotion recognition using fundamental frequency (f0). Acoustic parameters such as Jitter and Shimmer can be used in determining emotion as well as few diseases like Parkinson’s.
Fundamental frequency can be calculated using few different techniques:
a. Zero crossing
b. Time domain
c. Frequency domain
d. Time and Frequency domain (Hybrid approaches)
The popular algorithms to calculate fundamental frequencies (f0) are
a. Auto correlation
d. SPINET etc.
If the audio is of poor quality and if there are lots of noise, estimation of lower frequency peaks will be very difficult as they might disappear. Also, due to the presence of harmonics pitch estimation is not easy. Therefore it is the researchers responsibility to select a good pitch tracking algorithms for accuracy and uniqueness of the audio signal.