Perceived Voice Quality Research Articles

Multidimensional assessment of vocal changes in benign vocal fold lesions after voice therapy

To evaluate through a multidimensional protocol voice changes after voice therapy in patients with benign vocal fold lesions. 65 consecutive patients affected by benign vocal fold lesions were enrolled. Depending on videolaryngostroboscopy the patients were divided into 3 groups: 23 patients with Reinke's oedema, 22 patients with vocal fold cysts and 20 patients with gelatinous polyp. Each subject received 10 voice therapy sessions and was evaluated, before and after voice therapy, through a multidimensional protocol including videolaryngostroboscopy, perception, acoustics, aerodynamics and self-rating by the patient. Data were compared using Wilcoxon signed-rank test. Kruskal-Wallis test was used to analyse the mean variation difference between the three groups of patients. Mann-Whitney test was used for post hoc analysis. Only in 11 cases videolaryngostroboscopy revealed an improvement of the initial pathology. However a significant improvement was observed in perceptual, acoustic and self-assessment ratings in the 3 groups of patients. In particular the parameters of G, R and A of the GIRBAS scale, and the noise to harmonic ratio, Jitter and shimmer scores improved after rehabilitation. A significant improvement of all the parameters of Voice Handicap Index after rehabilitation treatment was found. No significant difference among the three groups of patients was visible, except for self-assessment ratings. Voice therapy may provide a significant improvement in perceptual, acoustic and self-assessed voice quality in patients with benign glottal lesions. Utilization of voice therapy may allow some patients to avoid surgical intervention.

Evaluating the perceived voice quality on VoIP network using interpolated FIR filter algorithm

Voice over Internet Protocol (VoIP) is a popular communication service nowadays. VoIP reduces the cost of call transmission by passing voice and video packets through the available bandwidth for data packets through Internet protocol. The quality of the VoIP signal is degraded due to the various network impairments. The proposed scheme, interpolated finite impulse response, is implemented as post-processor after decoding the signal in VoIP system. The performance of the proposed scheme is evaluated for various network conditions. The results of the proposed scheme are measured with the objective measurement methods for signal quality evaluation. The performance of the proposed system is compared with the existing techniques for quality improvement in VoIP system. The results show much improvement in speech quality with the proposed scheme in comparison to other similar schemes.

The Acoustic Assessment of Voice in Continuous Speech

Acoustic measures are an essential component in the assessment of voice disorders, but the value of these measures is dependent on their relationship to perceptual voice quality and the degree to which these measures reflect the typical speaking patterns of the individual being assessed. Therefore, acoustic measures that can be accurately and reliably derived from continuous speech contexts, which are more representative of every day speaking patterns than sustained vowels, are fundamental to the assessment of voice. In this article, I review the current findings on acoustic measures that are applicable to continuous speech. I will identify spectral- and cepstral-based measures that show strong relationships to perceptual ratings of overall voice severity or relate to particular dimensions of voice quality. I also will discuss the prominence of the cepstral peak as a measure that consistently shows strong predictive capacity for perceptually rated voice severity and provides excellent discrimination of dysphonic and normal voices.

Investigating fuzzy-input fuzzy-output support vector machines for robust voice quality classification

The dynamic use of voice qualities in spoken language can reveal useful information on a speakers attitude, mood and affective states. This information may be very desirable for a range of, both input and output, speech technology applications. However, voice quality annotation of speech signals may frequently produce far from consistent labeling. Groups of annotators may disagree on the perceived voice quality, but whom should one trust or is the truth somewhere in between? The current study looks first to describe a voice quality feature set that is suitable for differentiating voice qualities on a tense to breathy dimension. Further, the study looks to include these features as inputs to a fuzzy-input fuzzy-output support vector machine (F2SVM) algorithm, which is in turn capable of softly categorizing voice quality recordings. The F2SVM is compared in a thorough analysis to standard crisp approaches and shows promising results, while outperforming for example standard support vector machines with the sole difference being that the F2SVM approach receives fuzzy label information during training. Overall, it is possible to achieve accuracies of around 90% for both speaker dependent (cross validation) and speaker independent (leave one speaker out validation) experiments. Additionally, the approach using F2SVM performs at an accuracy of 82% for a cross corpus experiment (i.e. training and testing on entirely different recording conditions) in a frame-wise analysis and of around 97% after temporally integrating over full sentences. Furthermore, the output of fuzzy measures gave performances close to that of human annotators.

Effects of immediate feedback on learning auditory perceptual voice quality evaluation

The study investigated the effect of immediate feedback in training listeners to perceive subtle differences in voice quality, a perceptual skill that is important for speech-language pathologists. Sixty naive listeners were randomly assigned to a feedback group (Group F), a no feedback group (Group NF), and a no training group acting as a control group (Group C). The task was to evaluate the severity of a perceptual voice quality (breathiness) by using a reference-matching paradigm. All participants took part in three rating sessions (pre-training, 2 days after training and 1 week after training). Group F and Group NF participated in a training session immediately after the first rating session, where Group F practiced with immediate feedback given and Group NF practice with no immediate feedback given. The results showed that Group F and Group NF had significant improvement after training, but Group F did not retain the improvement in the third rating session. The use of a reference-matching training paradigm without giving frequent immediate feedback is suggested for auditory-perceptual voice evaluation training. The most effective frequency of immediate feedback is yet to be determined.

Voice Quality after Recurrent Laryngeal Nerve Resection and Immediate Reconstruction

To evaluate clinician perception of voice quality, patient self-reported psychosocial impact of dysphonia, and glottic closure in patients who have undergone resection of the recurrent laryngeal nerve (RLN) and immediate operative reconstruction during thyroid surgery. Prospective observational study. Single tertiary care hospital. Nine patients underwent immediate operative reconstruction of the RLN by the senior author from 2002 to present. Outcome measures included (1) perceptual voice ratings assessed using the GRBAS (Grade, Roughness, Breathiness, Asthenia, Strain) scale, (2) psychosocial impact of dysphonia assessed using the Voice Handicap Index (VHI), and (3) assessment of glottic closure from laryngeal videostroboscopy. The RLN was reconstructed with primary anastomosis (4), free nerve graft (3), or vagus-RLN anastomosis (2). Seven patients had voice samples and videostroboscopy examinations obtained at a minimum of 9 months from surgery. Six were judged to have slight disturbance of voice based on overall Grade scoring (G = 1). Five rated their voice as normal or mild on the VHI instrument (score range 8-29). Laryngeal analysis revealed the immobile vocal fold in the median, physiologic phonating position with preserved bulk, recovered tension, and glottic closure during phonation. Three patients less than 9 months from surgery had an expected severe self-reported rating of dysphonia. Primary anastomosis, free nerve grafting, and vagus-RLN anastomosis are viable options for RLN reconstruction. Patients who undergo resection and immediate reconstruction of the RLN are able to regain self-perceived functional voices.

Quasi-output-cost-ratio, perceived voice quality, and subjective evaluation in female kindergarten teachers

This study estimated vocal economy and investigated its relation to perceived voice quality and self-reported symptoms of vocal fatigue in kindergarten teachers. Quasi-output-cost ratio (QOCR) was calculated for sustained vowel [a:] at habitual loudness and at sound pressure level (SPL) of 90 dB6cm. QOCR was calculated as (SPL/CQEGG) × T/T0, where CQEGG is the contact quotient measured from the electroglottogram, T is the period length, and T0 is arbitrarily set mean period length (0.005 s in females). QOCR correlated negatively with perceived firmness at standard SPL and with the evaluation of too high pitch for the subject at both SPLs. QOCR did not predict self-reported vocal fatigue. Further studies with connected speech and objectively stated loading symptoms are warranted.

Vocal Outcome After Arytenoid Adduction and Ansa Cervicalis Transfer

To evaluate the long-term efficacy of arytenoid adduction (AA) combined with ansa cervicalis-recurrent laryngeal nerve anastomosis (ACN-RLN) in the treatment of unilateral vocal fold paralysis. Retrospective review of clinical records. Institutional practice. Nine patients with severe paralytic dysphonia with large glottal gap were included. Voice outcome was followed up over 24 months postoperatively. One patient did not attend the 24-month evaluation. All patients underwent AA + ACN-RLN. The ansa cervicalis nerve to the sternohyoid muscle was used as the donor nerve. Maximum phonation time (MPT), pitch range, harmonics-to-noise ratio (HNR), and perceptual voice quality were evaluated preoperatively and postoperatively at 1 to 3 months, 6 to 8 months, 12 to 14 months, and 24 months. All parameters improved significantly after surgery and continued to improve over the 24-month period. The MPT continued to improve over time (P = .01, P = .006, and P = .001 when comparing the 1- to 3-month evaluation with the 6- to 8-month, 12- to 14-month, and 24-month evaluations, respectively). Also, pitch range and HNR showed significant, steady improvement over the 24-month duration of the study. Perceptual voice quality markedly improved at 24 months compared with the 1- to 3-month, 6- to 8-month, and 12- to 14-month follow-ups (P = .004, P = .005, and P = .02, respectively, for grade overall, and P = .004, P = .008, and P = .02, respectively, for breathiness grade). Treatment with AA + ACN-RLN provides near-normal vocal function in the 24-month follow-up. Therefore, this method could be a successful surgical treatment for severe paralytic dysphonia.

Performance Analysis of VoIP by Communicating Two Systems

VoIP is a technology that digitizes voice packets and enables telephony communication by using internet as a backbone. In this paper we introduce and explore the problems that occur in our day to day life in sending voice and data with different windows. The paper evolves in the VoIP environment. Sending of text and voice along single window is possible. In earlier days VoIP merges over devices like IP Telephony, Palms, PDAs, etc over IP network for communication. But less security is provided with packet loss, retransmission, end to end delay and overhead. VoIP is composed of Signalling and Media. The use of Signalling is for controlling communication and it does the call set up, locate users, and tear down sessions. Media is used for transporting the voice packets. Ad hoc wireless networks are formed by a set of mobile nodes that communicate with each other over a wireless channel. Perceived Voice Quality can be achieved in terms of NS-2 Simulator. The work focuses with the combined forward error correction, layered coding and multiple. VoIP is a term used for a set of facilities for managing the delivery of voice information using the Internet Protocol. With VoIP there is no dedicated connection between the communicating devices. In addition to IP VoIP uses parameters such as residual packet loss rate and average burst length for secured delivery of voice and data. The protocols used are SIP and RTP. Our task in this work is to configure and create the test bed to simulate the working of Voice over IP (VoIP) Network. Also detailed and scrutinized study is made to test the performance of the proposed configuration. Two systems are installed with Client software and the third one with Server and the settings are made in such a way that Call from Client A will pass to the server and then server transfers to client B. NS-2 is used to analyse the network communication among the three machines

VoIP Quality Enhancement with Wideband Extension Method in Broadband Networks

Voice over Internet Protocol (VoIP) is one of the most important technologies in the world of communication. In this work, end-users connect via ADSL modem to broadband IP network and a variety of multimedia services utilized in this network. VoIP is the most important services in the broadband network and attempts focused on improving quality of voice in VoIP toward the quality of voice in PSTN. Perceived voice quality is an important function in VoIP applications. The main aim is to find an efficient speech enhancement method for perceptual optimization of voice quality. In this work, use band-width extension (BWE) method to estimate wideband envelope and improve quality and intelligibility. This method applies to some common codecs (e.g. G.711/ G.726/ G.729/ iLBC, speex) and then evaluates quality with ITU P.862.2 (wideband PESQ). Simulation takes under both low packet loss conditions and non packet loss conditions. The results illustrate that the BWE algorithm can achieve reliable perceived voice quality in comparison with the previous condition.

Pre- and Posttreatment Voice and Speech Outcomes in Patients With Advanced Head and Neck Cancer Treated With Chemoradiotherapy: Expert Listeners’ and Patient’s Perception

Perceptual judgments and patients' perception of voice and speech after concurrent chemoradiotherapy (CCRT) for advanced head and neck cancer. Prospective clinical trial. A standard Dutch text and a diadochokinetic task were recorded. Expert listeners rated voice and speech quality (based on Grade, Roughness, Breathiness, Asthenia, and Strain), articulation (overall, [p], [t], [k]), and comparative mean opinion scores of voice and speech at three assessment points calculated. A structured study-specific questionnaire evaluated patients' perception pretreatment (N=55), at 10-week (N=49) and 1-year posttreatment (N=37). At 10 weeks, perceptual voice quality is significantly affected. The parameters overall voice quality (mean, -0.24; P=0.008), strain (mean, -0.12; P=0.012), nasality (mean, -0.08; P=0.009), roughness (mean, -0.22; P=0.001), and pitch (mean, -0.03; P=0.041) improved over time but not beyond baseline levels, except for asthenia at 1-year posttreatment (voice is less asthenic than at baseline; mean, +0.20; P=0.03). Perceptual analyses of articulation showed no significant differences. Patients judge their voice quality as good (score, 18/20) at all assessment points, but at 1-year posttreatment, most of them (70%) judge their "voice not as it used to be." In the 1-year versus 10-week posttreatment comparison, the larynx-hypopharynx tumor group was more strained, whereas nonlarynx tumor voices were judged less strained (mean, -0.33 and +0.07, respectively; P=0.031). Patients' perceived changes in voice and speech quality at 10-week post- versus pretreatment correlate weakly with expert judgments. Overall, perceptual CCRT effects on voice and speech seem to peak at 10-week posttreatment but level off at 1-year posttreatment. However, at that assessment point, most patients still perceive their voice as different from baseline.

From high-speed imaging to perception: In search of a perceptually relevant voice source model

The goal of this project is to develop a new source model from high-speed recordings of vocal fold vibrations and simultaneous audio recordings. By analyzing area waveforms and acoustics and performing perception experiments, we will better parameterize the new source model and also uncover those aspects of the model that are perceptually salient, leading to improved TTS systems. We recorded acoustic and high-speed video signals from eight speakers producing the vowel /i/. Each speaker produced four phonation types (breathy, modal, pressed, and creaky) and three pitches (low, normal, and high). Glottal area waveforms were extracted from the high-speed recordings, semi-automatically, on a frame-by-frame basis. The acoustic results indicated that phonation types are differentiated by a number of spectral and noise measures, including H1*-H2* and the harmonics-to-noise ratio. In addition, the different pitch levels were responsible for changes in quality within each phonation type, but such changes were subject to inter-speaker variability. Perceptual results highlight the importance of lower-frequency harmonics in voice quality perception. These studies are critical to our computational modeling efforts, because they help us understand which aspects of the source model are perceptually important. [Work supported in part by NSF and NIH.]

Voice Quality after Recurrent Laryngeal Nerve Resection and Primary Reconstruction

Objective: Evaluate clinician and patient perception of voice quality in patients who have undergone resection of the recurrent laryngeal nerve (RLN) and primary operative reconstruction with anastomosis or nerve graft during thyroid surgery. Method: Prospective observational study of 9 patients who underwent primary operative reconstruction of the RLN by the senior author from 2002 to present. Outcome measures included: 1) speech pathologist perception of voice scored using the GRBAS scale, 2) patient perception of voice assessed by the Voice Handicap Index (VHI) questionnaire, and 3) visual laryngeal analysis from videostroboscopy examinations. Results: The RLN was reconstructed with primary anastomosis (4), free nerve graft (3), or vagus-RLN anastomosis (2). Seven patients had voice samples and videostroboscopy examinations obtained at a minimum of nine months from surgery. Six were judged to have slight disturbance of voice based on overall Grade scoring (G = 1). Five rated their voice normal or with a mild degree of severity on the VHI questionnaire (score range, 8-29). Laryngeal analysis revealed the immobile vocal fold in the median, physiologic phonating position with preserved bulk, recovered tension, and glottic closure during phonation. Two patients less than 9 months from surgery had expected severe self perceived voice quality. Conclusion: Primary anastomosis, free nerve grafting, and vagus-RLN anastomosis are viable options for RLN reconstruction. VHI scores are comparable to VHI scores of patients who have undergone silastic medialization laryngoplasty. Patients who undergo resection and immediate reconstruction of the RLN are able to regain self-perceived functional and pleasing voices.

Septorhinoplasty With Spreader Grafts Enhances Perceived Voice Quality Without Affecting Acoustic Characteristics

To identify the effect of septorhinoplasty with spreader grafts on patients' perception of voice and to measure formant frequencies that may be responsible for perceived changes in voice quality. A total of 20 patients who underwent septorhinoplasty and had spreader grafts placed during the operations were included. All subjects were tested within the week before surgery and 1-3 months postoperatively by means of perceptual assessment (Voice Handicap Index-10 [VHI-10] and self-assessment of hypo/hypernasality), acoustic analysis, and formant frequency analysis. The mean of VHI-10 score was decreased from 9.44±6.1 to 5.1±3.94 postoperatively (P=0.03). Fifteen patients (75%) perceived their voices to be hyponasal before surgery, but only three perceived the hyponasality to persist after surgery (P<0.001). No patient perceived the voice to be hypernasal either before or after surgery. Fifteen patients (75%) perceived their overall voice quality to be improved, whereas five patients perceived no change. None of the patients perceived their voice to be worse after surgery. There were no significant differences between pre- and postoperative acoustic analysis and formant frequency analysis (P>0.05). Septorhinoplasty with spreader grafts significantly improved patients' perception of voice; however, acoustic analysis and formant frequency analysis of nasalized vowels did not reveal any significant differences after the operation.

Relative fundamental frequency as an acoustic correlate of vocal effort in spasmodic dysphonia.

Individuals with vocal hyperfunction have lowered relative fundamental frequency (RFF) in vowels immediately preceding and following a voiceless consonant, but the relationship between RFF and perceived voice quality has yet to be studied in neurogenic voice disorders. The purpose of this study was to determine the relationship between RFF with listeners’ perceptions of 19 individuals with adductor spasmodic dysphonia. Speech recordings were presented to 20 inexperienced listeners who evaluated effort and overall severity using visual analog scales. The correlation coefficient (R2) between average effort and severity measures and an RFF measure was calculated as a function of the number of acoustic instances used for the RFF estimate (1–9, of a total of nine voiced-voiceless-voiced instances). Increases in the number of acoustic instances used for the RFF average led to increases in the correlation with perceptual measures, with use of six or more instances resulting in a stable estimate. Correlation coefficient values were highest between effort and RFF (range R2=0.06–0.43), with slightly lower correlation coefficients between severity and RFF (range R2=0.06–0.35). This study indicates that RFF measures are correlated with perceived voice quality and that guidelines should be developed for calculating reliable estimates of RFF in future research.

A Comparative Survey of Adaptive Codec Solutions for VoIP over Multirate WLANs: A Capacity versus Quality Performance Trade-Off

In multi-rate WLANs, users can suffer transmission rate changes due to the link adaptation mechanism. This results in a variable capacity channel, which is very hostile for VoIP and can cause serious quality of service (QoS) degradation in all active calls. Various codec adaptation mechanisms have been proposed as a solution to this, as well as to solve congestion problems on WLAN environments. Here, these solutions are presented, categorized according to the adaptation policy and scenario they implement, and evaluated at call-level in terms of the resulting blocking and dropping probabilities, as well as the perceived voice quality. To define a common performance metric, a new index named VGoS-factor is presented, which, by combining these capacity and quality indicators, can provide an overall view of the capacity versus quality trade-off of the proposed mechanisms and consequently help in choosing the adequate policy for each scenario.

Evaluating perceived voice quality on packet networks using different random neural network architectures

Voice over Internet Protocol (VoIP) is one of the fastest growing technologies in the world. In VoIP speech signals are transmitted over the same network used for data communications. The internet is not a robust network and is subjected to delay, jitter, and packet loss. It is very important to measure and monitor the quality of service (QoS) the users experience in VoIP networks; this is not an easy task and usually requires subjective tests. In this paper we have analyzed three non-intrusive models to measure and monitor voice quality using Random Neural Networks (RNN). A RNN is an open queuing network with positive and negative signals. We have assessed the voice quality based on various parameters i.e. delay, jitter, packet loss, and codec. In our approach we have used the Mean Opinion Score (MOS) calculated using a Perceptual Evaluation of Speech Quality (PESQ) algorithm to generate data for training the RNN model. We have studied two feed-forward models and a recurrent architecture. We have found that the simple feed-forward architecture has produced the most accurate results compared to the other two architectures.

Integrated software for analysis and synthesis of voice quality

Voice quality is an important perceptual cue in many disciplines, but knowledge of its nature is limited by a poor understanding of the relevant psychoacoustics. This article (aimed at researchers studying voice, speech, and vocal behavior) describes the UCLA voice synthesizer, software for voice analysis and synthesis designed to test hypotheses about the relationship between acoustic parameters and voice quality perception. The synthesizer provides experimenters with a useful tool for creating and modeling voice signals. In particular, it offers an integrated approach to voice analysis and synthesis and allows easy, precise, spectral-domain manipulations of the harmonic voice source. The synthesizer operates in near real time, using a parsimonious set of acoustic parameters for the voice source and vocal tract that a user can modify to accurately copy the quality of most normal and pathological voices. The software, user's manual, and audio files may be downloaded from http://brm.psychonomic-journals.org/content/supplemental. Future updates may be downloaded from www.surgery.medsch.ucla.edu/glottalaffairs/.

Spectral- and Cepstral-Based Measures During Continuous Speech: Capacity to Distinguish Dysphonia and Consistency Within a Speaker

Spectral- and cepstral-based acoustic measures are preferable to time-based measures for accurately representing dysphonic voices during continuous speech. Although these measures show promising relationships to perceptual voice quality ratings, less is known regarding their ability to differentiate normal from dysphonic voice during continuous speech and the consistency of these measures across multiple utterances by the same speaker. The purpose of this study was to determine whether spectral moments of the long-term average spectrum (LTAS) (spectral mean, standard deviation, skewness, and kurtosis) and cepstral peak prominence measures were significantly different for speakers with and without voice disorders when assessed during continuous speech. The consistency of these measures within a speaker across utterances was also addressed. Continuous speech samples from 27 subjects without voice disorders and 27 subjects with mixed voice disorders were acoustically analyzed. In addition, voice samples were perceptually rated for overall severity. Acoustic analyses were performed on three continuous speech stimuli from a reading passage: two full sentences and one constituent phrase. Significant between-group differences were found for both cepstral measures and three LTAS measures (P<0.001): spectral mean, skewness, and kurtosis. These five measures also showed moderate to strong correlations to overall voice severity. Furthermore, high degrees of within-speaker consistency (correlation coefficients ≥0.89) across utterances with varying length and phonemic content were evidenced for both subject groups.

Intervenção fonoaudiológica para consultores em um serviço de teleatendimento: bem-estar vocal

OBJETIVO: verificar efeitos de um programa de bem-estar vocal aos consultores de um serviço de tele-atendimento para a saúde. MÉTODOS: participaram do estudo 27 consultores do Call Center VivaVoz que foram avaliados pré e pós-treinamento de voz por meio de análise perceptivo-auditiva da qualidade vocal, articulação, velocidade de fala, ressonância, intensidade e freqüência, por meio da emissão de vogais sustentadas e de fala encadeada com contagem de números de 1 à 20 e a emissão dos dias da semana. O registro de voz foi realizado em sistema digital e a escala japonesa GRBAS-I como instrumento de análise. Também foram realizadas a auto-avaliação vocal e o levantamento dos sintomas vocais. Atividades de voz foram divididas em cinco oficinas sobre saúde vocal, técnicas de aquecimento e desaquecimento, articulação e respiração. Análises descritivas e bivariadas foram realizadas, utilizando Teste T para amostras pareadas e Teste de McNemar. RESULTADOS: os resultados pré-intervenção são de que 50% das mulheres e 33% dos homens consultores apresentaram alterações em relação à respiração, 50% das consultoras com alterações na qualidade vocal e 33% dos homens dificuldades em relação à articulação. Após as oficinas, os resultados mostraram melhora da qualidade vocal, do padrão articulatório e da fluência de fala, bem como da satisfação vocal nos relatos dos consultores. CONCLUSÃO: observaram-se mudanças positivas na qualidade do atendimento à população, além da ampliação do conhecimento dos teleatendentes em relação ao uso correto da voz e manutenção do bem-estar vocal.