I would like to extract pitch from a sound related to a instrument. I get sound from microscope and I convert it to frequency domain by FFT. Now I have a vector which represents the frequencies of the sound and I want to obtain its pitch.
Can you help me or introduce a good reference?
This document might help you get started.
http://www.ece.rochester.edu/projects/wcng/papers/theses/cai_MSthesis.pdf
It is more specific to noisy speech signals, but section 2.2 covers a general methodology and approach to pitch estimation (such as HPS). It also references documents (that you can trace back) that might contain more broad information on the subject. But on a general note, if you are dealing with a monophonic and relatively "clean" signal expressing a single sonic event, HPS should provide you decent results. Otherwise, you might have to look at other alternatives. Start with the wiki: http://en.wikipedia.org/wiki/Pitch_detection_algorithm
For polyphonic signals, this paper might be helpful as well:
http://mtg.upf.edu/system/files/publications/SalamonGomezMelodyTASLP2012.pdf
Hope this helps!
http://www.ece.rochester.edu/projects/wcng/papers/theses/cai_MSthesis.pdf
http://en.wikipedia.org/wiki/Pitch_detection_algorithm
Article Melody Extraction From Polyphonic Music Signals Using Pitch ...
As a first stab at it try to track the frequency with maximum power within the range of frequencies you consider acceptable. And read.
I compute Power Spectral Density (PSD) for each of 11 frequency ranges (1 kHz intervals). You can examine the values for each frequency to determine the greatest PSD.
The best algorithm I have found is SWIPE, by Camacho and Harris 2008. Its based on human auditory system (please remember that pitch is a matter of perception).
http://www.cise.ufl.edu/~acamacho/publications/dissertation.pdf
You can also read the sound file using tool PRAAT and view the pitch information using options already available with PRAAT. It is open source free downloadable software. It supports .wav file. I am not aware of the purpose of pitch extraction you have with the instrumental music but hope it will help you in same.
Link for your reference
http://praat.en.softonic.com/
In case of harmonic spectrum the pitch depends on the frequency of the lowest component of this spectrum. The distance between harmonic components is relevant for the pitch perception. This is the effect of "residuum pitch" : https://documentation.apple.com/en/logicstudio/instruments/index.html#chapter=7%26section=15%26tasks=true
Hello,
For human auditory perception, a famous reference is
"Terhardt, E., Stoll, G. and Seewann, M. (1982). Algorithm for extraction of pitch and pitch salience from complex tonal signals. Journal of the Acoustical Society of America, 71, 679-688"
This paper gives routines to calculate "spectral pitch" and "virtual pitch". The former is a representation of the actual 'pitch shifts' (perceived frequencies) derived from relevant spectral tonal peaks detection and accounting for spectral masking; the "virtual pitch" is derived from the spectral pitch content by harmonic coincidence; i.e. we can perceive the missing fundamentals from a complex tone because of a subset of harmonics present in its spectral pitch.
The paper also shows the dependence of real Pitch on sound pressure level. A typical example: Someone blowing an identical tone in a trumpet produces a louder and louder sound but also its pitch values will shift higher and higher (in the poor listener's perception).
The key point of these pitch detectors is that these are close to our perception!
However, I would not use it as a first choice for voice/speech analysis/detection - many other algorithms cited earlier were designed for.
But for a musical instrument - or any machine producing complex tonal signals - as in your case, this may be a good perceptual-based track to consider!
Best regards, Laurent.
Dear Professor Payam Ahmadvand ,
I would like suggest you my book, Auditory and Visual Sensations, Springer, NY
2009.
Thanking you and wishing you,
Yoichi Ando
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing