Objective: To analyze the diagnostic accuracy of Cepstral Peak Prominence (CPP) and Smoothed Cepstral Peak Prominence (CPPS) in identifying dysphonic voices, considering sustained vowel (SV) and connected speech (CS) tasks.

Method: This is a systematic review (SR) of diagnostic accuracy, registered in PROSPERO, following the methodology of the Joanna Briggs Institute (JBI), and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) as a reporting tool. Our research question was: "Are CPP and CPPS measures sensitive and specific for identifying dysphonic voices through connected speech and sustained vowel tasks?". We searched different databases. Studies of diagnostic accuracy involving individuals with suspected voice disorders were selected; This study included studies that used the CPP or CPPS extracted from sustained vowels (CPPSV and CPPSV) or connected speech (CPPSCS and CPPSCS) as an index test; that employed the overall severity of vocal deviation as a reference standard; that presented sensitivity and specificity measures; and that had a cross-sectional observational design. Risk of bias was assessed using the JBI Critical Appraisal Checklist for Diagnostic Test Accuracy Studies. Certainty of evidence was assessed using GRADE. A meta-analysis was performed using a random-effects model, and estimates of the Summary Receiver Operating Characteristic (SROC) and the diagnostic odds ratio (DOR) were calculated in RStudio, along with the positive (LR+) and negative (LR-) likelihood ratios.

Results: 14 studies were included in the synthesis for CPPS evaluation, and 16 for CPP. Most of the included studies demonstrated a low risk of bias (93.75% for CPP and 92.86% for CPPS). CPPSV showed 77% sensitivity and 82% specificity, while CPPCS showed 82% in both parameters. DOR levels indicated better performance for CPPCS (20.82) compared to CPPSV (15.26). Positive reliability (VR+) ranged from 4.15 to 4.34 and negative reliability (VR-) from 0.22 to 0.29 for CPP. The certainty of the evidence, according to the GRADE approach, was high for all outcomes. In CPPSSV, the sensitivity was 0.81 and the specificity was 0.89, with a DOR level of 35.11, a positive reliability of 6.84, and a negative reliability of 0.22. The CPPSCS demonstrated a sensitivity of 0.85 and a specificity of 0.88, with a DOR of 38.64, RV+ of 6.24, and RV– of 0.17. The evidence was low in certainty for all outcomes (CPPSSV and CPPSCS). We mainly attributed this downgrade to a slight inconsistency between studies, which requires caution in interpretation and clinical application.Conclusion: Both CPP and CPPS demonstrate good diagnostic accuracy in identifying dysphonic voices across sustained-vowel and connected-speech tasks. CPP, backed by high-certainty evidence, consolidates its position as a reliable, valid, and objective measure for clinical practice, with CPPcs showing slightly superior performance, underscoring the value of natural speech context. Conversely, despite CPPS's sometimes higher point estimates for sensitivity and specificity, the low certainty of its supporting evidence necessitates cautious clinical interpretation.