Objective: To describe Phase Space Reconstruction (PSR) in voice signals according to the type of voice signals. Method: The study was approved by the Institutional Review Board (IRB) of the institution under protocol number: 2,332,833. The study analyzed 106 voice signals from patients of both sexes, aged 18 to 50 years: 53 type I voices, 45 type II voices, and 8 type III voices. The non-linear analysis was performed using the Phase Space Reconstruction (PSR) method through a two-dimensional graph obtained by the Voice Analysis Program. The analysis was conducted using the 4-point crescent scale, called curves-irregularity and spacing (C-IS). Statistical analysis was performed with the Mann-Whitney test, with a significance level of 5%. Results: In the PSR analysis, voice signals of type I and II showed graphs with both open and closed trajectories, with a predominance of 4 curves, while type III voice signals predominantly showed 3 curves. Irregularity grade 1 (mild) was predominant for type I voice signals, grades 1 and 2 (mild and moderate) for type II, and grades 1 to 3 (normal to severe) for type III. Spacing grade 1 (small) was predominant in type I and II signals, while grade 3 (large) was predominant in type III. There was a significant difference (p < 0.05) between the types of voice signals and the C-IS scale in irregularity and spacing (types I and II), and in curves, irregularity, and spacing (types I and III; types II and III). Conclusion: Voice types I and II presented PSR graphs with open and closed trajectories, with a predominance of 4 curves, while type III showed a predominance of 3 curves, ranging from mild to severe irregularity and small to large spacing depending on the type of voice signal. All signals showed significant differences when compared using the PSR graph.