Long-term Average Spectrum Analysis Research Articles

Long-Term Average Spectrum as an Outcome Measure of Vocal Projection After Resonant Voice Therapy in Carnatic Classical Singers

The current study evaluated the effects of resonant voice therapy (RVT) on the long-term average spectrum (LTAS) parameters of vocally healthy trained Carnatic classical singers. The objectives of the study were i) to understand the perceptual differences and ii) to understand the change in LTAS contour parameters and the alpha ratio of the sung voice samples of the trained Carnatic classical singers before and after RVT. The study used a quasi-experimental, prepost design and purposive-convenient sampling method. Nineteen vocally healthy singers (4 males and 15 females) aged 18-25 years (M=23; standard deviation [SD]=2.09) with minimum 10 years of formal Carnatic classical vocal music training volunteered to participate in the study. The audio recordings of the participants' sung aakaara (/a/ vowel singing) were obtained before and after the training. All participants were assigned to a 21-day-long RVTtraining program scheduled as 15-20 minutes sessions. The participants' pre-training and post-training singing samples were subjected to LTAS analysis. The spectral mean, SD, skewness, kurtosis, and alpha ratio were extracted from the sung samples. The mean and SD of the parameters were obtained using descriptive statistical methods. Wilcoxon signed-rank test was performed to evaluate significant differences in the obtained parameters across pre-training and post-training sung samples. Though there was an increase in perceptual vocal resonance in the post-training sung samples, the difference was observed to be statistically insignificant. The results indicated a significant difference only in the alpha ratio values, which were decreased, indicating an increase in the high-frequency energies. The descriptive statistical evaluations showed a reduction in the parameters of spectral mean, SD, skewness, and kurtosis. The observations from the current study indicated that the RVT training increased the energies of higher frequencies in the singers' voices (1-5kHz). This high-frequency energy favored improved voice projection, imparting perceptual resonanceor ringing quality of the voice.

Overdrive and Edge as Refiners of “Belting”?: An Empirical Study Qualifying and Categorizing “Belting” Based on Audio Perception, Laryngostroboscopic Imaging, Acoustics, LTAS, and EGG

We aimed to study the categorizations "Overdrive" and "Edge" from the pedagogical method Complete Vocal Technique as refiners of the often ill-defined concept of "belting" by means of audio perception, laryngostroboscopic imaging, acoustics, long-term average spectrum (LTAS), and electroglottography (EGG). This is a case-control study. Twenty singers were recorded singing sustained vowels in a "belting" quality refined by audio perception as "Overdrive" and "Edge." Two studies were performed: (1) a laryngostroboscopic examination using a videonasoendoscopic camera system (Olympus) and the Laryngostrobe program (Laryngograph); (2) a simultaneous recording of the EGG and acoustic signals using Speech Studio (Laryngograph). The images were analyzed based on consensus agreement. Statistical analysis of the acoustic, LTAS, and EGG parameters was undertaken using the Student paired t test. The two modes of singing determined by audio perception have visibly different laryngeal gestures: Edge has a more constricted setting than that of Overdrive, where the ventricular folds seem to cover more of the vocal folds, the aryepiglottic folds show a sharper edge in Edge, and the cuneiform cartilages are rolled in anteromedially. LTAS analysis shows a statistical difference, particularly after the ninth harmonic, with a coinciding first formant. The combined group showed statistical differences in shimmer, harmonics-to-noise ratio, normalized noise energy, and mean sound pressure level (P ≤ 0.05). "Belting" sounds can be categorized using audio perception into two modes of singing: "Overdrive" and "Edge." This study demonstrates consistent visibly different laryngeal gestures between these modes and with some correspondingly significant differences in LTAS, EGG, and acoustic measures.

Can Genre Be “Heard” in Scale as Well as Song Tasks? An Exploratory Study of Female Singing in Western Lyric and Musical Theater Styles

Using an empirical design, this study investigated perceptual and acoustic differences between the recorded vocal products of songs and scales of professional female singers of classical Western Lyric (WL) and non-legit Musical Theater (MT) styles. A total of 54 audio-recorded samples of songs and scales from professional female singers were rated in a blind randomized testing process by seven expert listeners as being performed by either a WL or MT singer. Songs and scales that were accurately perceived by genre were then analyzed intra- and inter-genre using long-term average spectrum analysis. A high level of agreement was found between judges in ratings for both songs and scales according to genre (P < 0.0001). Judges were more successful in locating WL than MT, but accuracy was always >50%. For the long-term average spectrum analysis intra-genre, song and scale matched better than chance. The highest spectral peak for the WL singers was at the mean fundamental frequency, whereas this spectral area was weaker for the MT singers, who showed a marked peak at 1 kHz. The other main inter-genre difference appeared in the higher frequency region, with a peak in the MT spectrum between 4 and 5 kHz-the region of the "speaker's formant." In comparing female singers of WL and MT styles, scales as well as song tasks appear to be indicative of singer genre behavior. This implied difference in vocal production may be useful to teachers and clinicians dealing with multiple genres. The addition of a scale-in-genre task may be useful in future research seeking to identify genre-distinctive behaviors.

Lower Vocal Tract Morphologic Adjustments Are Relevant for Voice Timbre in Singing.

The vocal tract shape is crucial to voice production. Its lower part seems particularly relevant for voice timbre. This study analyzes the detailed morphology of parts of the epilaryngeal tube and the hypopharynx for the sustained German vowels /a/, /e/, /i/, /o/, and /u/ by thirteen male singer subjects who were at the beginning of their academic singing studies. Analysis was based on two different phonatory conditions: a natural, speech-like phonation and a singing phonation, like in classical singing. 3D models of the vocal tract were derived from magnetic resonance imaging and compared with long-term average spectrum analysis of audio recordings from the same subjects. Comparison of singing to the speech-like phonation, which served as reference, showed significant adjustments of the lower vocal tract: an average lowering of the larynx by 8 mm and an increase of the hypopharyngeal cross-sectional area (+ 21.9%) and volume (+ 16.8%). Changes in the analyzed epilaryngeal portion of the vocal tract were not significant. Consequently, lower larynx-to-hypopharynx area and volume ratios were found in singing compared to the speech-like phonation. All evaluated measures of the lower vocal tract varied significantly with vowel quality. Acoustically, an increase of high frequency energy in singing correlated with a wider hypopharyngeal area. The findings offer an explanation how classical male singers might succeed in producing a voice timbre with increased high frequency energy, creating a singer‘s formant cluster.

Examining the extent of anticipatory coronal coarticulation: A long-term average spectrum analysis

Phonetic studies of English liquids /r/ and /l/ have shown these consonants can exert strong coarticulatory effects on both adjacent and non-adjacent vowels. The studies of long-range coarticulation, however, have so far been limited to British English and have not explicitly compared liquids to other coronal consonants. The current study investigated local and long-range effects of coronals /l/, /r/, and /d/ in Canadian English. 14 speakers were recorded reading the sentences “We thought it might be(a) ram/lamb/dam/ham” repeatedly. Long-term average spectra of five vowels preceding the target consonants were calculated and compared to baseline values. The results revealed significant differences between the coronal consonants and the control (/h/) in up to five preceding syllables. Significant differences among the three coronals, however, were limited to the immediately preceding vowel. The other vowels showed some differences between /r/ and /d/, but not between /l/ and the other coronals. The results show that long-range coarticulation can spread for up to five syllables and involve both liquid and non-liquid coronals. The spectral differences between the two liquids in Canadian English, however, may not be as robust as have been reported for British English.

Electroglottographic Analysis of Actresses and Nonactresses' Voices in Different Levels of Intensity

Previous studies with long-term average spectrum (LTAS) showed the importance of the glottal source for understanding the projected voices of actresses. In this study, electroglottographic (EGG) analysis was used to investigate the contribution of the glottal source to the projected voice, comparing actresses and nonactresses' voices, in different levels of intensity. Thirty actresses and 30 nonactresses sustained vowels in habitual, moderate, and loud intensity levels. The EGG variables were contact quotient (CQ), closing quotient (QCQ), and opening quotient (QOQ). Other variables were sound pressure level (SPL) and fundamental frequency (F0). A KayPENTAX EGG was used. Variables were inputted in a general linear model. Actresses showed significantly higher values for SPL, in all levels, and both groups increased SPL significantly while changing from habitual to moderate and further to loud. There were no significant differences between groups for EGG quotients. There were significant differences between the levels only for F0 and CQ for both groups. SPL was significantly higher among actresses in all intensity levels, but in the EGG analysis, no differences were found. This apparently weak contribution of the glottal source in the supposedly projected voices of actresses, contrary to previous LTAS studies, might be because of a higher subglottal pressure or perhaps greater vocal tract contribution in SPL. Results from the present study suggest that trained subjects did not produce a significant higher SPL than untrained individuals by increasing the cost in terms of higher vocal fold collision and hence more impact stress. Future researches should explore the difference between trained and nontrained voices by aerodynamic measurements to evaluate the relationship between physiologic findings and the acoustic and EGG data. Moreover, further studies should consider both types of vocal tasks, sustained vowel and running speech, for both EGG and LTAS analysis.

Long-Term Average Spectrum in Screening of Voice Quality in Speech: Untrained Male University Students

Voice quality has mainly been studied in trained speakers, singers, and dysphonic patients. Few studies have concerned ordinary untrained university students' voices. In light of earlier studies of professional voice users, it was hypothesized that good, poor, and intermediate voices would be distinguishable on the basis of long-term average spectrum characteristics. In the present study, voice quality of 50 Finnish vocally untrained male university students was studied perceptually and using long-term average spectrum analysis of text reading samples of one minute duration. Equivalent sound level (Leq) of text reading was also measured. According to the results, the good and ordinary voices differed from the poor ones in their relatively higher sound level in the frequency range of 1-3 kHz and a prominent peak at 3-4 kHz. Good voices, however, did not differ from the ordinary voices in terms of the characteristics of the long-term average spectrum (LTAS). The strength of the peak at 3-4 kHz and the voice-quality scores correlated weakly but significantly. Voice quality and alpha ratio (level difference above and below 1 kHz) correlated likewise. Leq was significantly higher in the students with good and ordinary voices than in those with poor voices. The connections between Leq, voice quality, and the formation of the peak at 3-4 kHz warrant further studies.

The Singer's Formant and Speaker's Ring Resonance: A Long-Term Average Spectrum Analysis

ObjectivesWe previously showed that a trained tenor's voice has the conventional singer's formant at the region of 3 kHz and another energy peak at 8-9 kHz. Singers in other operatic voice ranges are assumed to have the same peak in their singing and speaking voice. However, to date, no specific measurement of this has been made.MethodsTenors, baritones, sopranos and mezzo sopranos were chosen to participate in this study of the singer's formant and the speaker's ring resonance. Untrained males (n=15) and females (n=15) were included in the control group. Each subject was asked to produce successive /a/ vowel sounds in their singing and speaking voice. For singing, the low pitch was produced in the chest register and the high notes in the head register. We collected the data on the long-term average spectra of the speaking and singing voices of the trained singers and the control groups.ResultsFor the sounds produced from the head register, a significant energy concentration was seen in both 2.2-3.4 kHz and 7.5-8.4 kHz regions (except for the voices of the mezzo sopranos) in the trained singer group when compared to the control groups. Also, the chest register had a significant energy concentration in the 4 trained singer groups at the 2.2-3.1 kHz and 7.8-8.4 kHz. For speaking sound, all trained singers had a significant energy concentration at 2.2-5.3 kHz and sopranos had another energy concentration at 9-10 kHz.ConclusionThe results of this study suggest that opera singers have more energy concentration in the singer's formant/speaker's ring region, in both singing and speaking voices. Furthermore, another region of energy concentration was identified in opera singer's singing sound and in sopranos' speaking sound at 8-9 kHz. The authors believe that these energy concentrations may contribute to the rich voice of trained singers.

Throaty Voice Quality: Subglottal Pressure, Voice Source, and Formant Characteristics

"Throaty" voice quality has been regarded by voice pedagogues as undesired and even harmful. This study attempts to identify acoustic and physiological correlates of this quality. One male and one female subject read a text habitually and with a throaty voice quality. Oral pressure during p-occlusion was measured as an estimate of subglottal pressure. Long-term average spectrum analysis described the average spectrum characteristics. Sixteen syllables, perceptually evaluated with regard to throaty quality by five experts, were selected for analysis. Formant frequencies and voice source characteristics were measured by means of inverse filtering, and the vocal tract shape of the throaty and normal versions of the vowels [a,u,i,ae] of the male subject were recorded by magnetic resonance imaging. From this material, area functions were derived and their resonance frequencies were determined. The throaty versions of these four vowels all showed a pharynx that was narrower than in the habitually produced versions. To test the relevance of formant frequencies to perceived throaty quality, experts rated degree of throatiness in synthetic vowel samples, in which the measured formant frequency values of the subject were used. The main acoustic correlates of throatiness seemed to be an increase of F1, a decrease of F4, and in front vowels a decrease of F2, which presumably results from a narrowing of the pharynx. In the male subject, voice source parameters suggested a more hyperfunctional voice in throaty samples.

Long-term average spectrum (LTAS) analysis of sex- and gender-related differences in children's voices

Long-term average spectrum (LTAS) analysis offers representative information on voice timbre providing spectral information averaged over time. It is particularly useful when persistent spectral features are under investigation. The aim of this study was to compare perceived sex of children to the LTAS analysis of their audio signals. A total of 320 children, aged between 3 and 12 years, were recorded singing a song. In an earlier analysis, the recorded voices were evaluated with respect to perceived and actual sex by experienced listeners. From this group, a subgroup of 59 children (30 boys and 29 girls) was selected. The mean LTAS revealed a peak at 5 kHz for children perceived with confidence as boys, and a flat spectrum at 5 kHz for children perceived confidently as girls (whether male or female in actuality).

A study of the effects of vocal intensity variation on children's voices using long-term average spectrum (LTAS) analysis

Studies of adult voices have shown that, as vocal intensity is increased, the partials at higher frequencies gain more than those at lower frequencies. Investigations involving children's normal productions are uncommon, however, and there is, consequently, little knowledge of how children's vocal function differs from that of adults. Using LTAS analysis, this study investigates the effects of vocal intensity variation on the voices of 10-year-old schoolchildren singing in soft, mid and loud voice. A frequency-dependent gain factor was calculated which showed the increase in level to have been greater for partials at higher than at lower frequencies for these children. Also, gain within frequency bands was often different between boys and girls, although this was not demonstrated statistically.

Long-term average spectrum analysis of phonatory effects of noise and filtered auditory feedback

Using long-term average spectrum (LTAS) analysis of fluent speech, the effects of speaking in noise and of speaking with differently filtered auditory feedback as well as of voluntarily changing vocal intensity is analyzed in nine male adult singers and non-singers. Three values in the LTAS are analyzed: the level of the main peak near 500 Hz, of a secondary peak near 2000 Hz and at the average F 0 . The level at 500 Hz was highly correlated with the equivalent sound level, with the average F 0 , and with the level of the peak near 2000 Hz. The singers were found to produce a higher sound level than the non-singers under all conditions, and their voices contained stronger high-frequency components. Also, unlike the non-singers they reduced vocal intensity when reading without added noise with enhanced high-frequency components in the auditory feedback. The LTAS effects of noise and of auditory feedback filtering was found to be similar to those of the changes of vocal intensity according to instruction for both singers and non-singers. When the high-frequency components of the auditory feedback was enhanced during reading in noise, both subject groups reduced vocal intensity.