Cepstral Peak Prominence Analysis
The Surface: What is Cepstral Peak Prominence Analysis?
Cepstral Peak Prominence (CPP) Analysis and Smoothed Cepstral Peak Prominence Analysis (CPPs) measure the periodicity of the voice. As this analysis method relies on measuring the periodicity of peaks on a spectrogram (AKA how close to being equally distant between peaks) instead of the direct audio signal, it’s considered more reliable than other analysis methods, such as jitter or shimmer. Higher values mean higher degrees of periodicity (Sujitha & Pebbili, 2022).
What does periodicity mean?
Periodicity relates to how well the voice doesn’t deviate from a pattern. Fluctuations (i.e. tremor and vibrato), roughness, and breathiness in the voice will decrease the periodicity in the voice. A typically healthy voice will have a higher level of periodicity.
How do we measure it?
Before obtaining a CPP value, we first have to obtain a spectrum. A spectrum breaks an audio signal (AKA a complex sine wave) to its base harmonics on a frequency-intensity graph. A cepstrum is then created by measuring how equally distant those harmonics on the spectrum are. Here’s a visual example:
Spectrum
A frequency-intensity graph in the time domain. It shows the composite bands of frequency in Hertz of a complex sine wave at a specific moment in time. Simply, it breaks a real-world sound into the frequencies and their loudness that make up that sound. We can separate these bands of frequencies through a Fourier Transform
Some Definitions
Cepstrum: In basic terms, this is a “spectrum of a spectrum.” Cepstrums look at sound and find the periodicity between harmonics. This is a quefrency/intensity graph within the frequency domain.
Rahmonic: Uniformly spaced quefrencies that are multiples of the fundamental quefrency/cepstral peak.
Quefrency: The frequency of the frequency.
Smoothed Cepstral Peak Prominence: Adjusts sharp fluctuations in the graph to more easily identify trends.
Cepstral Peak Prominence: Distance from the top of the cepstral peak to the regression line.
How do we calculate it?
In PRAAT
With scripts
While I will not go in depth on how to use scripts in PRAAT for this project, there are other researchers who do. Elizabeth Murray, Andie Chao Lauren Colletti, and members of the Vocal Development Lab have created a PRAAT plug-in to measure CPP values. This is free to use, and includes a tutorial on how to install and use. (Murray et. al, 2025)
The article describing the plug-in: https://doi.org/10.1016/j.jvoice.2022.09.002
The video tutorial: https://osf.io/t5hrv/
The plug-in: https://osf.io/t5hrv/
Additionally, many voice labs and researchers will code their own scripts in PRAAT to automatically measure CPP values. If you need assistance, reaching out to voice researchers in a Communication Sciences and Disorders or Speech, Language, and Hearing Sciences department may prove beneficial.
Factors to consider when using PRAAT
Measuring CPP values is effective in both sustained vowels and speech.
Smoothed Cepstral Peaking Prominence analysis (CPPS) may be more reliable compared to regular CPP analysis, specifically when it comes to sustained vowels (Hillenbrand & Houde, 1996).
CPP values may be more reliable than jitter and shimmer for students with dysphonia as it doesn’t rely on accurate pitch tracking (Murton et. al, 2020).
It is best to use the same recording task and recording setup when analyzing CPP values over time (i.e. CAPE-V Passage). CPP analysis can vary between specific vowels, pitches and tasks, meaning consistency is key (Sampaio et. al, 2020; Murray et. al, 2025).
It’s important to use only one program while monitoring over time. (Murray et. al, 2025).
Singing generally has a higher CPP value compared to speaking as it demonstrates more periodicity.
Higher pitch, loudness, and nasality can impact the CPP value. It’s important to control these variables if monitoring over a duration of time. What does this mean? Have your vocalists try to match the same loudness and pitch everytime you record for CPP analysis (Murray et. al, 2025).
Vibrato can increase the CPP value compared to non-vibrato singing.
Citations
Heller Murray, E. S., Chao, A., & Colletti, L. (2022). A practical guide to calculating cepstral peak prominence in Praat. Journal of Voice, 39(2), 365–370. https://doi.org/10.1016/j.jvoice.2022.09.002
Hillenbrand, J., & Houde, R. A. (1996). Acoustic correlates of breathy vocal quality: Dysphonic voices and continuous speech. Journal of Speech, Language, and Hearing Research, 39(2), 311–321. https://doi.org/10.1044/jshr.3902.311
Murton, O., Hillman, R., & Mehta, D. (2020). Cepstral peak prominence values for clinical voice evaluation. American Journal of Speech-Language Pathology, 29(3), 1596–1607. https://doi.org/10.1044/2020_ajslp-20-00001
P S, S., & Pebbili, G. K. (2020). Cepstral analysis of voice in young adults. Journal of Voice, 36(1), 43–49. https://doi.org/10.1016/j.jvoice.2020.03.010
Randall, R. B. (2016). A history of Cepstrum analysis and its application to mechanical problems. Mechanical Systems and Signal Processing, 97, 3–19. https://doi.org/10.1016/j.ymssp.2016.12.026
Sampaio, M. C., Bohlender, J. E., Brockmann-Bauser, M. (2017). Fundamental Frequency and Intensity Effects on Cepstral Measures in Vowels from Connected Speech. Retrieved September 1, 2025, from YouTube website: https://www.youtube.com/watch?v=Z7BFMSlqmpg
