Voice Intensity Research Articles

Fear emotion classification in speech by acoustic and behavioral cues

Machine-based emotional speech classification has become a requirement for natural and familiar human-computer interactions. Because emotional speech recognition systems use a person’s voice to spontaneously detect their emotional status and take subsequent appropriate actions, they can be used widely for various reason in call centers and emotional based media services. Emotional speech recognition systems are primarily developed using emotional acoustic data. While there are several emotional acoustic databases available for emotion recognition systems in other countries, there is currently no real situational data related to the “fear emotion” available. Thus, in this study, we collected acoustic data recordings which represent real urgent and fearful situations from an emergency call center. To classify callers’ emotions more accurately, we also included the additional behavioral feature “interjection” which can be classified as a type of disfluency which arises due to cognitive dysfunction observed in spontaneous speech when a speaker gets hyperemotional. We used Support Vector Machines (SVM), with the interjections feature, as well as conventionally used acoustic features (i.e., F0 variability, voice intensity variability, and Mel-Frequency Cepstral Coefficients; MFCCs) to identify which emotional category acoustic data fell into. The results of our study revealed that the MFCC was the best acoustic feature for spontaneous fear speech classification. In addition, we demonstrated the validity of behavioral features as an important criteria for emotional classification improvement.

Speech training for neurological patients using a serious game

AbstractThe majority of patients with neurological impairment like Parkinson's Disease (PD) or stroke are affected by dysarthria. Dysarthria is a motor speech impairment which negatively affects speech dimensions such as articulation and loudness. This leads to reduced intelligibility, often hindering daily life communication. Intensive and prolonged speech training can increase patients’ speech intelligibility. Unfortunately, interventions by speech therapists are generally provided only for a short period of time, while continuing practice is needed to maintain or improve intelligibility. eHealth applications might provide a solution. In our research, we explored whether it is possible to develop a game that is suitable for providing speech training in elderly patients with dysarthria due to PD or stroke. In the game, we developed, called Treasure Hunters, two players interact verbally to find the way to the treasure, while receiving automatic feedback on voice loudness and pitch. Participants played with our game in several sessions and generally appreciated it, hinting at our game's potential for speech training in elderly patients. In a within‐subjects experiment with five dysarthric patients, our game was compared to a non‐game computer‐based speech training system: e‐learning‐based Speech Therapy (EST). We focussed on three variables: speech intelligibility, user satisfaction and user preference. Substantial variability between participants was observed, in the outcomes of these three variables and their relations. We conclude that ”one size that fits all” does not apply to computer‐based speech training, but a personalised approach is needed.

Objective and Subjective Assessment of Amplified Parkinsonian Speech Quality.

Hypophonia is a common speech impairment associated with Parkinson's disease (PD). Voice amplifiers are typically used to increase voice loudness, but little is known about their impact on perceived speech quality. In this paper, speech recordings were obtained from 11 PD subjects with and without the use of seven different amplification devices, and in the absence or presence of background noise. The recorded speech samples were rated for their sound quality by 10 naive listeners. The same speech recordings were analyzed objectively, where in linear prediction, mel-frequency cepstral coefficients (MFCCs), and gammatone cepstral coefficients (GFCCs) were extracted and mapped to predicted quality scores using linear regression and Support Vector Regression (SVR). Results showed that amplification devices differentially affect the perceived quality of PD speech, that objective and subjective quality scores correlated well, and that a reduced set of GFCC features mapped with SVR produced the best correlation with the subjective scores.

Effect of nasal obstruction in the acoustic analysis of voice

<p class="abstract"><strong>Background:</strong> <span lang="EN-IN">The voice is the main form of communication between humans and the nasal and sinusal cavity participates in the resonance process. An alteration in the nasal diameter can produce changes in the voice. The objective of this study was to relate the degree of nasal obstruction measured by the SNOT 20 questionnaire to the acoustic analysis of the voice. </span></p><p class="abstract"><strong>Methods:</strong> <span lang="EN-IN">58 non-smoker volunteers without laryngeal pathology where included, 25 women (43%) and 33 men (57%). The average age was 22.1 years (±1.7). The laryngopharyngeal reflux and SNOT20 questionnaires were applied together with an acoustic analysis, we measured fundamental frequency, Jitter and Shimmer. The analysis was done separating in groups by sex and by SNOT score (less than 20 points and equal or higher than 20). </span></p><p class="abstract"><strong>Results:</strong> <span lang="EN-IN">The average of SNOT questionnaire was 15.06 (±11.8), we found no relation between the scores of the SNOT20 questionnaire with Jitter, fundamental frequency, Shimmer or voice intensity (p≥0.01). When separated by groups, the analysis by sex or by scores (SNOT > or < than 20 groups) did not show a statistically significant relationship. </span></p><p class="abstract"><strong>Conclusions:</strong> <span lang="EN-IN">In this study we did not find a relation between changes in acoustic analysis in Fundamental frequency, Jitter and Shimmer and the degree of nasal obstruction measured by SNOT 20. The groups of gender and severity of SNOT did not show relationship. More studies are necessary in order to evaluate the relation of nasal obstruction and the acoustic analysis of the voice.</span></p><p> </p>

Long-term effects of Lee Silverman Voice Treatment on daily voice use in Parkinson’s disease as measured with a portable voice accumulator

This study examines the effects of an intensive voice treatment focusing on increasing voice intensity, LSVT LOUD® Lee Silverman Voice Treatment, on voice use in daily life in a participant with Parkinson’s disease, using a portable voice accumulator, the VoxLog. A secondary aim was to compare voice use between the participant and a matched healthy control. Participants were an individual with Parkinson’s disease and his healthy monozygotic twin. Voice use was registered with the VoxLog during 9 weeks for the individual with Parkinson’s disease and 2 weeks for the control. This included baseline registrations for both participants, 4 weeks during LSVT LOUD for the individual with Parkinson’s disease and 1 week after treatment for both participants. For the participant with Parkinson’s disease, follow-up registrations at 3, 6, and 12 months post-treatment were made. The individual with Parkinson’s disease increased voice intensity during registrations in daily life with 4.1 dB post-treatment and 1.4 dB at 1-year follow-up compared to before treatment. When monitored during laboratory recordings an increase of 5.6 dB was seen post-treatment and 3.8 dB at 1-year follow-up. Changes in voice intensity were interpreted as a treatment effect as no significant correlations between changes in voice intensity and background noise were found for the individual with Parkinson’s disease. The increase in voice intensity in a laboratory setting was comparable to findings previously reported following LSVT LOUD. The increase registered using ambulatory monitoring in daily life was lower but still reflecting a clinically relevant change.

Dysarthrophonia in Association with Voice Analysis: A Case Report

Stroke is the second leading cause of death worldwide and the brain damage caused by it can affect communication in several aspects. Voice analysis in dysarthria is challenging because of the complexity of the disorder and its effects on the speech production system. In this study we are presenting a 56-years-old male who was visited to Medanta Hospital with history of hypertension and chief complaint of Right upper limb weakness and slurred speech to the Emergency and later Clinically and Radio logically Diagnosed as LT MCA Infarct. Later, on the day 3 the patient has undergone Speech and Language Evaluation and Diagnosed with Spastic Dysarthria based on Frenched Dysarthria Assessment scale and later a detail Voice Analysis was done with using PRAAT software and analysed voice features. Voice analysis basically deals with decomposition of voice signal into voice parameters for processing the resulted features in desirable application. The features that are extracted in this paper are: frequency, pitch, voice intensity, formant, speech rate and pulse functions like Jitter (local), Jitter (local, absolute), Jitter (rap), Jitter (ppq5), Jitter (ddp), Shimmer (local), Shimmer (local, dB), Shimmer (apq3), Shimmer (apq5), Shimmer (apq11), Shimmer (dda) and Harmonic coefficients. Over all, we conclude with the voice parameters in spastic dysarthria which reveals interesting data on the voice quality with features which helps the clinician for better management. However, large sample study is required.

SWIFT cognitive behavioral assessment model built on cognitive analytics of empirical mode internet of things

This paper introduces a study and analysis to predict the present human behaviour through his/her object interactions in the physical environment. The physical environment consists of a door, chair and telephone with accelerometer sensors attached to them and connected to computer using a raspberry pi IoT(Internet of Things) kit. Two other parameters used for assessment are human voice intensities and human motion analysis through a motion capture camera with inbuilt microphone and Wi-Fi module. The dataset is a collection of accelerometer data from chair and telephone, human interaction with door through camera and voice sample of a word ‘Hello’. These 4 parameter measurements are collected from 15 test subjects in the age group 19-21 without their knowledge. We used the dataset to train and test 3 predominant behaviours in the chosen age group namely, excitable, assertive and pleasant on an artificial neural network with backpropagation training algorithm. The overall recognition accuracy is 84.89% based on the physical assessment from a physiatrist of all the test subjects. This study can help individuals, doctors and machines to predict the current human emotional state and provide feedback to modify unpleasant current state of behaviour to a pleasant state to maximize human performance.

Intensity of the Reader’s Voice in the Reading Aloud of Fiction: Effects of the Character’s Gender

Abstract The speaker’s gender is a crucial factor affecting the acoustic features of the voice. One such feature is voice intensity, also known as sound pressure level (SPL). Previous studies have indicated that the female voice may involve lower values of SPL than the male voice. Moreover, there are suggestions that the variability of voice intensity tends to be lower for women than for men as well. The major aim of this paper is to examine the effects of literary character’s gender on the reader’s SPL, measured in decibels (dB), and the variability of voice intensity, measured as the standard deviation (SD) of SPL, while reading prose aloud. The secondary aims are to investigate the general shifts of SPL and SD of SPL in dialogues independently of other variables and to consider the possible effects of the reader’s gender and the reader’s dialect. In order to accomplish these tasks, a representative sample of dialogue excerpts with male and female characters was used. Each fragment was located in the corresponding audiobook and analysed in terms of the two acoustic features under discussion. Typical values of SPL and the SD of SPL for different readers were measured in the entire chapters from which fragments were selected and the results were compared with those obtained from the extracts. In this way, it was possible to establish the relative shifts of SPL and the SD of SPL for each of the analysed fragments. Contrary to what had been expected, a statistical analysis of the results revealed no effects of the character’s gender on any of the response variables. However, conclusions concerning secondary aims were more definitive. A general trend to decrease the SD of SPL in dialogues in comparison to the rest of the text in a novel was observed. This tendency is independent of any of the factors included in the study. It was also observed that male American readers tend to lower their voice intensity when reading dialogues. All these findings may be applied in developing text to speech software.

The Effects of Amplification on Vocal Dose in Teachers with Dysphonia

The purpose of this study was to determine if voice amplification influenced vocal dose in female teachers with dysphonia. This was an experimental study with comparative intrasubjects in which 15 individuals were compared in two different moments: condition 1 (C1) without voice amplification and condition 2 (C2) with voice amplification. All of them were female, kindergarten and elementary school teachers who presented organic or functional dysphonia. The search was carried out at the school where the teachers work. The professional voice use was considered the teachers' activity for a continuous period of two classes (average recording time of 96 minutes, with no difference in time between C1 and C2). To measure the dose we used the vocal dosimeter composed of a microphone, an accelerometer fixed to the neck, and a portable unit that stores the vocal data. The phonation data (intensity, fundamental frequency, phonation percentage, cycle dose, and distance dose) were analyzed by the equipment software (VoxLog). The use of vocal amplification in teachers promotes a reduction of the fundamental frequency (295.6-267.7 Hz), the voice intensity (96.2-93.3 dB sound pressure level), the cycle doses (489.4-345.2 thousand cycles per second), and distance doses (3,800-2,300m). The vocal amplification allows the teacher to maintain the same phonation time (phonation percentage) but decreases the number of vocal fold oscillations (cycle dose) and the total distance traveled by the vocal fold tissue during phonation (distance dose), reducing the exposure of the vocal folds to voice trauma.

Speaker responses to indirect evidence of acoustic signal transmission conditions.

Evidence has been accumulating that speakers adjust their speech production to account for conditions of signal transmission to an interlocutor, in addition to speaker- and listener-related influences. This study manipulated acoustic transmission conditions between speakers and a researcher while they completed dyadic cooperative puzzle tasks. Speakers heard evidence of four transmission perturbations (intensity increase/decrease, pitch increase/decrease) only indirectly, i.e., via changes in the researcher's voice. Analysis of speakers' conversational turns across conditions (baseline + four partner-voice manipulations) showed significant group-level imitation of transmission manipulations of partner voice intensity, with no general response to shifts in partner voice pitch. Significant voice alterations were also observed for individual speakers in intensity, with substantial inter-talker variability in responses. Thirteen of 29 speakers exclusively imitated the manipulation of partner intensity, 4/29 speakers exclusively opposed the transmission manipulation, and 4/29 speakers showed both response types. The presence of opposition responses coupled with a relatively low-magnitude group-level convergence suggests the possibility of a compensatory force at work in at least some speakers, adjusting for changes made during signal transmission despite their lack of direct experience with the transmission manipulation effects.

Intensity of vocal responses to spider and snake pictures in fearful individuals

ObjectiveStrong bodily responses have repeatedly been shown in participants fearful of spiders and snakes when they see pictures of the feared animal. In this study, we investigate if these fear responses affect voice intensity, require awareness of the pictorial stimuli, and whether the responses run their course once initiated.MethodAnimal fearful participants responded to arrowhead‐shaped probes superimposed on animal pictures (snake, spider, or rabbit), presented either backwardly masked or with no masking. Their task was to say ‘up’ or ‘down’ as quickly as possible depending on the orientation of the arrowhead. Arrowhead probes were presented at two different stimulus onset asynchronies (SOA), 261 or 561-ms after picture onset. In addition to vocal responses, electrocardiogram, and skin conductance (SC) were recorded.ResultsNo fear‐specific effects emerged to masked stimuli, thereby providing no support for the notion that fear responses can be triggered by stimuli presented outside awareness. For the unmasked pictures, voice intensity was stronger and SC response amplitude was larger to probes superimposed on the feared animal than other animals, at both SOAs. Heart rate changes were greater during exposure to feared animals when probed at 561-ms, but not at 261-ms, which indicates that a fear response can change its course after initiation.ConclusionExposure to pictures of the feared animal increased voice intensity. No support was found for responses without awareness. Observed effects on heart rate may be due to change in parasympathetic activation during fear response.

Voice in different phases of menstrual cycle among naturally cycling women and users of hormonal contraceptives.

Previous studies have shown changes in women's behavior and physical appearance between the non-fertile and fertile phases of the menstrual cycle. It is assumed that these changes are regulated by fluctuations in sex hormone levels across the cycle. Receptors for sex hormones have been found on the vocal folds, suggesting a link between hormone levels and vocal fold function, which might cause changes in voice production. However, attempts to identify changes in voice production across the menstrual cycle have produced mixed results. Therefore, the purpose of this study was to investigate changes in sexually dimorphic vocal characteristics and quality of women’s voices in different phases of the cycle and to compare these with users of monophasic hormonal contraception. Voice samples (vowel phonation) of 44 naturally cycling women were obtained in the menstrual, late follicular (confirmed by LH surge) and luteal phases, and in 20 hormonal contraceptive users across equivalent stages of the monthly cycle. Results showed that voices of naturally cycling women had higher minimum pitch in the late follicular phase compared with the other phases. In addition, voice intensity was at its lowest in the luteal phase. In contrast, there were no voice changes across the cycle in hormonal contraceptive users. Comparison between the two groups of women revealed that the naturally cycling group had higher minimum pitch in the fertile phase and higher harmonics to noise ratio in the menstrual phase. In general, present results support the assumption that sex hormones might have an effect on voice function. These results, coupled with mixed findings in previous studies, suggest that vocal changes in relation to hormonal fluctuation are subtle, at least during vowel production. Future studies should explore voice changes in a defined social context and with more free-flowing speech.

Employee Voice and Team Innovation: Examining the Role of Team Voice Intensity in R&D Teams

Drawing on the dialectical perspective on innovation (Bledow, Frese, Anderson, Erez, & Farr, 2009), we propose that both promotive and prohibitive voices are necessary for team innovation through d...

Júri (dis)simulado: proposta de atividade para o estudo do gênero textual/discursivo “depoimento pessoal” sob o viés da oralidade e da variação linguística

This article shows that the work adapted to the school jury, here referred to as "Jury Covert" is configured as a practice of literacy performed in a meaningful context for learning. Theoretically we support us in Bakhtin (2003), Bortoni-Ricardo and Dettoni (2001), Hilgert (1996), Marcuschi (2008), among others. Through this article, we present a proposal of didactic material that stimulates the students, especially the reading and analysis of the genre/discourse "personal testimony", seeking to overcome the naive consciousness of limits for critical awareness of this kind, recognizing the explicit intentions and implicit transmitted it. The didactic activity proposed here allows students oriented time of study, focusing on critical and analytical way the orality continuum and writing of this genre, as well as the observation of the importance of issues such as respect for conversational turn round to round, orality marks, intonation of sentences and voice intensity, among other factors. Permeating these issues also aim to enable reflection activities about language and language, deconstructing false beliefs and linguistic prejudices that there is a better linguistic variety or superior to others. Through the "Jury Covert", we intend to promote the work not only with the varieties most prestigious, but also with the varieties sometimes stigmatized by the majority of society and are frequently used by students. Thus, the activities aim also appreciate the variety of non-prestige that characterize the identity of the students and that are present in textual/discursive genre "personal testimony" to then develop use of the skills cultivated variety of the language itself.

Acoustic Perturbation Measures Improve with Increasing Vocal Intensity in Individuals With and Without Voice Disorders

In vocally healthy children and adults, speaking voice loudness differences can significantly confound acoustic perturbation measurements. This study examines the effects of voice sound pressure level (SPL) on jitter, shimmer, and harmonics-to-noise ratio (HNR) in adults with voice disorders and a control group with normal vocal status. This is a matched case-control study. We assessed 58 adult female voice patients matched according to approximate age and occupation with 58 vocally healthy women. Diagnoses included vocal fold nodules (n = 39, 67.2%), polyps (n = 5, 8.6%), and muscle tension dysphonia (n = 14, 24.1%). All participants sustained the vowel /a/ at soft, comfortable, and loud phonation levels. Acoustic voice SPL, jitter, shimmer, and HNR were computed using Praat. The effects of loudness condition, voice SPL, pathology, differential diagnosis, age, and professional voice use level on acoustic perturbation measures were assessed using linear mixed models and Wilcoxon signed rank tests. In both patient and normative control groups, increasing voice SPL correlated significantly (P < 0.001) with decreased jitter and shimmer, and increased HNR. Voice pathology and differential diagnosis were not linked to systematically higher jitter and shimmer. HNR levels, however, were statistically higher in the patient group than in the control group at comfortable phonation levels. Professional voice use level had a significant effect (P < 0.05) on jitter, shimmer, and HNR. The clinical value of acoustic jitter, shimmer, and HNR may be limited if speaking voice SPL and professional voice use level effects are not controlled for. Future studies are warranted to investigate whether perturbation measures are useful clinical outcome metrics when controlling for these effects.

Tratamiento de voz Lee Silverman en la disartria hipocinética: revisión de la efectividad del tratamiento en los enfermos de Parkinson

Dysarthria is a neurological disorder of the central and/or peripheral nervous system that affects muscle control mechanisms of speech. Parkinson's disease (PD) is characterised by hypokinetic dysarthria, which produces a weak, slow and monotonous speech. It is affected by breathing, resonance, phonation, prosody, and articulation. There is a voice treatment called “Lee Silverman” (LSVT), which stimulates and educates the patient to produce a higher voice through the increase of vocal effort.The purpose of this review is to determine the effectiveness of LSVT as a method of speech therapy rehabilitation of hypokinetic dysarthria in patients with PD.A literature review was conducted in the PubMed and Cochrane databases. A total of 65 articles selected articles, but they were reduced to 7 after applying the inclusion and exclusion criteria. A comparative analysis was performed on these.The results of the study and comparison with other reviews have shown significant improvements after LSVT. Significant changes have been observed in intelligibility, voice intensity with an increase in vocal loudness and greater tone of the buccal-phonation organs, as well as an improvement in the articulation and coordination movements. This has led to a more functional speech.To conclude, the results show that LSVT could be effective for the rehabilitation of the hypokinetic dysarthria in Parkinson's patients. However, the results cannot be considered universal, as only 7 articles have been compared, which is not sufficient to support or refute the effectiveness of the treatment.

Telerehabilitation in Parkinson's disease: Influence of cognitive status.

BackgroundThe need for efficacy in voice rehabilitation in patients with Parkinson's disease is well established. Given difficulties traveling from home to treatment centers, the use of telerehabilitation may represent an invaluable tool for many patients.ObjectiveTo analyze the influence of cognitive performance on acceptance of telerehabilitation.MethodsFifty patients at stages 2-4 on the Hoehn-Yahr scale, aged 45-87 years old, with cognitive scores of19-30 on the Mini-Mental State Examination, and 4-17 years of education were enrolled. All patients were submitted to evaluation of voice intensity pre and post in-person treatment with the Lee Silverman Voice Treatment (LSVT) and were asked to fill out a questionnaire regarding their preferences between two options of treatment and evaluating basic technological competence.ResultsComparisons between pre and post-treatment values showed a mean increase of 14dBSPL in vocal intensity. When asked about potential acceptance to participate in future telerehabilitation, 38 subjects agreed to take part and 12 did not. For these two groups, 26% and 17% self-reported technological competence, respectively. Agreement to engage in remote therapy was positively associated with years of education and cognitive status.ConclusionResponses to the questionnaire submitted after completion of traditional in-person LSVT showed that the majority of patients (76%) were willing to participate in future telerehabilitation. Age, gender, disease stage and self-reported basic technological skills appeared to have no influence on the decision, whereas other factors such as cognitive status and higher school education were positively associated with acceptance of the new therapy approach.

The Speaking Voice in the General Population: Normative Data and Associations to Sociodemographic and Lifestyle Factors

Normative data concerning the speaking voice in the general population were gathered with the aim to establish standard values for clinical diagnostics. Associations between the speaking voice and sociodemographic factors were examined. This is a prospective cross-sectional population-based study. Speaking voice profiles were measured for 2472 (1154 male and 1318 female) participants between the ages of 40 and 79 years, using four speaking voice intensity levels: softest speaking voice (I), conversational voice (II), classroom voice (III), and shouting voice (IV). Smoking status and socioeconomic status were assessed. Data were analyzed using multivariate regression. The mean voice frequencies were 111.8 Hz for male and 161.3 Hz for female participants (I), 111.9 Hz for male and 168.5 Hz for female participants (II), 130.2 Hz for male and 198.0 Hz for female participants (III), and 175.5 Hz for male and 246.2 Hz for female participants (IV). Frequencies increased significantly with age for male but not for female participants. Sound pressure levels rose significantly with age at intensity levels I-III for both sexes, but decreased at intensity level IV. Frequencies and sound pressure levels were similar between nonsmokers and former smokers. Current smokers showed significantly lower frequencies as opposed to non- and former smokers. Speaking voice range and dynamics increased with higher socioeconomic status. The data are suitable as age-adjusted normative values for clinical measurement of the speaking voice. The mean fundamental speaking voice frequency of female participants was six to seven semitones lower than previously described.

Effect of the tracheoesophageal voice resistance test in total laryngectomees.

To analyze tracheoesophageal voice resistance in laryngectomized patients by comparing the acoustic, perceptive-auditory and self-evaluation measures before and after the voice resistance test. An observational, experimental study with a quantitative and descriptive approach. Nineteen total laryngectomees were submitted for capture of vocal intensity, self-assessment of signs and symptoms of voice fatigue using a visual analogue scale, and perceptive-auditory and acoustic analysis of the tracheoesophageal voice before and after the voice resistance test, which consisted of continuous repeated counting from one to 100. The mean time of execution of the test was 33.68 minutes. Type II sound signal, voice intensity, voice quality, degree of severity of dysphonia, and the pitch and loudness parameters were maintained after the test. There were no changes in signs and symptoms of vocal fatigue. Most of the total laryngectomees performed the voice resistance test for less than 30 minutes. However, the voice resistance test did not induce significant changes in the quality of tracheoesophageal voice of total laryngectomees or in the signs and symptoms of vocal fatigue. 4 Laryngoscope, 2016 127:405-410, 2017.

Stable Voice Clusters Identified When Using the Maximum versus Minimum Intensity Curve in the Phonetogram

Objective: To examine whether the minimum as well as the maximum voice intensity (i.e. sound pressure level, SPL) curves of a voice range profile (VRP) are required when discovering different voice groups based on a clustering analysis. In this approach, no a priori labeling of voice types is used. Patients and Methods: VRPs of 194 (84 male and 110 female) professional singers were registered and processed. Cluster analysis was performed with the use of features related to (1) both the maximum and minimum SPL curves and (2) the maximum SPL curve only. Results: Features related to the maximum as well as the minimum SPL curves showed three clusters in both male and female voices. These clusters, or voice groups, are based on voice types with similar VRP features. However, when using features related only to the maximum SPL curve, the clusters became less obvious. Conclusion: Features related to the maximum and minimum SPL curves of a VRP are both needed in order to identify the three voice clusters.