Cepstral Peak Prominence (CPP) is an acoustic measure of dysphonia recommended by an ASHA expert panel. A lower CPP value is associated with a more dysphonic voice. Although CPP offers a more ecologically valid measure of dysphonia that can be completed on running speech, it is important to acknowledge that differences in algorithm used to calculate CPP, stimuli used for analysis, and other person-related factors (e.g., fundamental frequency, sound pressure level) may impact CPP values. See Watts and colleagues (2017) for discussing of two popular programs used, ADSV and Praat.
Murton and colleagues (2020) also compared CPP values from ADSV and Praat on both vowels and a portion of the rainbow passage (also see Murton and Colleagues for excellent discussion on previous CPP studies). They found the following cutoffs to distinguish between controls and individuals with voice disorders:
ADSV: (vowels) 11.46 dB ADSV: (rainbow) 6.11 dB
Praat: (vowels) 14.45 dB Praat: (rainbow) 9.33 dB
The current work in our lab replicated Murton and colleagues findings in Praat. We examined the the six CAPE-V sentences from 266 adults with and without voice disorders selected from the Perceptual Voice Qualities Database and found a cutoff CPP value of 9.52 dB. We also implemented a voice detection process to select only the voiced segments of speech for analysis, as some programs use voice detection and others do not. Using voice detection, our CPP cutoff value was 12.15 dB.
One of the most important factors to consider when using CPP is consistency!! Below we have made available our Praat scripts to calculate CPP with and without voice detection. As Praat is a free program and this script adapts to other open-source information, we hope this script will be helpful for both clinic and research. Furthermore, below you will find the poster on this topic from Voice Foundation 2022 as well a downloadable file with all the CPP values for the subjects used in the Perceptual Voice Qualities Database
Access Praat CPP scripts and data here!
Includes:
- Scripts – that provide both CPP-Full and CPP-Voice values
- Readme instructions
- Scripts for both single file and batch analysis
- CPP values for subjects in the PVQD database
CPP Voice Foundation Poster
Click on the file below to download the CPP values (CPP-Full, CPP-Voice) for subjects in the Perceptual Voice Qualities Database