LICENSE
This work is licensed under CC BY-NC-SA 

DESCRIPTION
This voice-sample data base has been annotated on the basis of the the GRB scale has been used (Dejonckere et al., 2001), which is a modified version of the original GRBAS (Hirano 1981). This is a perceptual rating system based on auditory judgements aimed at classifying the different qualities of dysphonic voices. 
By using this protocol, acoustic signals are mapped onto three GRB perceptual dimensions or parameters. Every dimension includes a specific type of characteristics: 
-	Grade (G) refers to the global amount of noise or perturbations that can be perceived in the signal. 
-	Roughness (R) , i.e. the hoarseness of noise. 
-	Breathiness (B), the �white noise�, aspiration noise or air component. 
Traditionally four values have been employed to measure each of these parameters. As a result of this choice, every GRB parameter can be classified as either 0 (absent), 1 (slight), 2 (moderate) or 3 (severe). 
Following the GRB protocol, 119 sustained /a/ have been perceptually annotated and classified in the following archives:
-	G archives. Signals labelled G0 are lacking noise, they are non-dysphonic voices. In G1, G2 and G3 archives recordings are perceived as slightly noisy, moderately noisy and severely noisy, respectively. The present noise can be rough, breathy or both. 
-	R archives contain signals that have been perceived as having different levels of roughness (R1, R2, R3). 
-	B archives contain signals that have been perceived as having different levels of breathiness (B1, B2, B3). 
It should be kept in mind that certain sustained vowels have both rough and breathy noises; therefore they are included in both R and B archives. Thus, when listening to those tracks, it is recommended to pay attention solely to the type of noise used to classify the archive. For instance, when listening to R1 archives, the listener should pay attention to the roughness component of the sound only, ignoring the possible breathy component which may also be present. 
Data in the dataset correspond to 28 individuals. Audio excerpts provided are sampled at 44100 Hz or 11025 Hz. 

Please, include a citation to this paper if you use this database. 
R. M. Bermudez de Alvear, J. Corral, L. J. Tardon, A. M. Barbancho, Elena Fernandez Contreras, Sara Rando Marquez, A. G. Martinez-Arquero, I. Barbancho, "A Database and Digital Signal Processing Framework for the Perceptual Analysis of Voice Quality", Pan European Voice Conference 2015, p. 35. 

REFERENCES
[Dejonckere et al., 2001] Dejonckere PH, Bradley P, Clemente P, Cornut G, Crevier-Buchman L, Friedrich G, Van De Heyning P, Remacle M, Woisard V. A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques. Guideline elaborated by the Committee on Phoniatrics of the European Laryngological Society (ELS). Eur Arch Otorhinolaryngol 2001;258:77�82. 
[Hirano 1981] Hirano M. Psyco-acoustic evaluation of voice: GRBAS Scale for evaluating the hoarse voice. Vienna: Springer; 1981. 
[Hirano 1981] Hirano M. Psyco-acoustic evaluation of voice: GRBAS Scale for evaluating the hoarse voice. Vienna: Springer; 1981.