Logopedics Phoniatrics Vocology

ISSN: 1401-5439 (Print) 1651-2022 (Online) Journal homepage: http://www.tandfonline.com/loi/ilog20

Freddie Mercury—acoustic analysis of speaking fundamental , , and subharmonics

Christian T. Herbst, Stellan Hertegard, Daniel Zangger-Borch & Per-Åke Lindestad

To cite this article: Christian T. Herbst, Stellan Hertegard, Daniel Zangger-Borch & Per-Åke Lindestad (2016): —acoustic analysis of speaking fundamental frequency, vibrato, and subharmonics, Logopedics Phoniatrics Vocology, DOI: 10.3109/14015439.2016.1156737

To link to this article: http://dx.doi.org/10.3109/14015439.2016.1156737

Published online: 15 Apr 2016.

Submit your article to this journal

Article views: 34

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ilog20

Download by: [81.141.59.99] Date: 19 April 2016, At: 06:46 LOGOPEDICS PHONIATRICS VOCOLOGY, 2016 http://dx.doi.org/10.3109/14015439.2016.1156737

ORIGINAL ARTICLE Freddie Mercury—acoustic analysis of speaking fundamental frequency, vibrato, and subharmonics

Christian T. Herbsta, Stellan Hertegardb, Daniel Zangger-Borchc and Per-Åke Lindestadb aVoice Research Lab, Department of Biophysics, Faculty of Science, Palacky University Olomouc, 17. listopadu 12, 771 46 Olomouc, Czech Republic; bDepartment of Otolaryngology, Head and Neck Surgery B53, Clinical Sciences and Intervention, Karolinska Institutet, Karolinska University Hospital, 141 86 Stockholm, Sweden; cVoice Centre Stockholm AB, Zangger Vocal Art, Box 4342, 102 67 Stockholm, Sweden

ABSTRACT ARTICLE HISTORY Freddie Mercury was one of the twentieth century’s best-known singers of commercial contemporary Received 4 September 2015 . This study presents an acoustical analysis of his voice production and style, based on Revised 18 November 2015 perceptual and quantitative analysis of publicly available sound recordings. Analysis of six interviews Accepted 12 February 2016 revealed a median speaking fundamental frequency of 117.3 Hz, which is typically found for a Published online 14 April voice. Analysis of voice tracks isolated from full band recordings suggested that the singing voice range 2016 was 37 semitones within the pitch range of F#2 (about 92.2 Hz) to G5 (about 784 Hz). Evidence for KEY WORDS higher up to a fundamental frequency of 1,347 Hz was not deemed reliable. Analysis of 240 Freddie Mercury; Queen; sustained notes from 21 a-cappella recordings revealed a surprisingly high mean fundamental fre- singing voice range; quency modulation rate (vibrato) of 7.0 Hz, reaching the range of vocal tremor. Quantitative analysis uti- speaking fundamental lizing a newly introduced parameter to assess the regularity of vocal vibrato corroborated its frequency; subharmonics; perceptually irregular nature, suggesting that vibrato (ir)regularity is a distinctive feature of the singing ventricular folds; vibrato; voice. Imitation of subharmonic samples by a professional rock , documented by endo- voice quality scopic high-speed video at 4,132 frames per second, revealed a 3:1 frequency locked vibratory pattern of vocal folds and ventricular folds.

Introduction acoustical and physiological conditions is ambitious. Special care needs to be taken in selecting proper data material Freddie Mercury, born 5 September 1946 as Farrokh Bulsara, (ideally , no background sounds) and suitable ana- and died 24 November 1991, was one of the most influential lysis methods. In this study, a series of exemplary interpreta- singers of his . The magazine listed him tions of voice quality were conducted, and, where deemed eighteenth within the 100 best (commercial contemporary) appropriate, quantitative estimation of voice features like singers of all time (1). Having been the lead singer of the rock band Queen, he profoundly influenced this group’s speaking fundamental frequency or vibrato properties was musical style over more than two decades. performed. These analyses, in part inspired by a previous Downloaded by [81.141.59.99] at 06:46 19 April 2016 Freddie Mercury’s voice has been described as ‘a force of publication by the first author (4), are conducted here with nature with the velocity of a hurricane’ (2), which was ‘esca- improved methodology and larger data sets. Finally, endo- lating within a few bars from a deep, throaty rock-growl to scopic high-speed video examination of vocal fold vibration tender, vibrant , then on to a high-pitched, perfect col- of a contemporary rock singer mimicking typical sounds pro- oratura, pure and crystalline in the upper reaches’ (3, p. 2). duced by Freddie Mercury was performed, in an attempt to Such descriptions, while presumably adequate for a biog- understand the voice production mechanism of subharmonic raphy or a newspaper article, do not satisfy deeper scholarly sounds. No formant analysis was undertaken, however, since interest into the singer’s voice characteristics. The purpose of it cannot be taken for granted that the spectral contents of this study was therefore to conduct a viable analysis of pub- commercially available recordings have not been considerably licly available data material, in order to arrive at more changed by the editing sound engineers. empirically based insights into Freddie Mercury’s voice pro- duction and singing style. Speaking fundamental frequency (SFF) and singing voice range Materials and methods In order to assess Freddie Mercury’s speaking fundamental The task of objectively describing the voice of a singer who frequency (SFF), a total of six interviews, conducted by is not available for acquiring recordings under controlled Andrew Wigg (1984, 1985, 1986, 1987) and

CONTACT Christian T. Herbst [email protected] Voice Research Lab, Department of Biophysics, Faculty of Science, Palacky University Olomouc, 17. listopadu 12, 771 46 Olomouc, Czech Republic Supplemental data for this article can be accessed here ß 2016 Informa UK Limited, trading as Taylor & Francis Group 2 C. T. HERBST ET AL.

(1984 and 1987) were analyzed. The interviews by Andres authors of this manuscript (two graduated voice pedagogues Wigg were taken from the commercial compilation Freddie and two phoniatricians), and all assessments are based on Mercury: The Solo Collection (2000, EMI/). The unanimous consensus of the four authors. interview conducted by Andrew Wigg in 1979 was excluded from the analysis due to excessive background noise. The Analysis of vocal vibrato interviews by Rudi Dolezal (‘A musical prostitute’, 1984, and ‘The last interview’, 1987) were downloaded from the inter- Vibrato is constituted by a (predominantly regular) modula- net (https://www.youtube.com/watch?v¼8wk9hPubD1Q and tion of the singing fundamental frequency, typically in the https://www.youtube.com/watch?v¼PnsiSjsSNYQ). range of 4–7 Hz (10). For the assessment of Freddie The acoustic tracks of all interviews were normalized in Mercury’s vibrato characteristics, the compilation The amplitude and then annotated and analyzed with the soft- Acapella Collection (http://queenspain.blogspot.co.at/2013/06/ ware Praat (5). Only those segments were considered where freddie-mercury-acapella-collection.html), compiled from without doubt Freddie Mercury’s voice was exclusively aud- multichannel recordings by the Queen Poland fan site ible. The fundamental frequency of all extracted sequences (https://queenpoland.wordpress.com/), was analyzed. This was estimated using Praat’s autocorrelation algorithm (6), assortment of studio recordings contains the vocal tracks of using a rather conservative parameter setting in order to 28 songs by Queen and Freddie Mercury. Sustained notes avoid analysis artifacts stemming from background noise or from these recordings have been considered for analysis unvoiced consonants: fundamental frequency range: when the following conditions were met: 1) exactly one sing- 70–400 Hz; silence threshold: 0.2; voicing threshold: 0.6; oct- ing voice was audible, and this voice could be clearly identi- ave cost: 0.01; octave jump cost: 0.8; voiced/unvoiced cost: fied as Freddie Mercury’s; 2) no background noise, 0.14; read progress: 10 ms. The analysis resulted in a total of background vocals, or orchestration was audible; 3) no note- 134,418 data points. worthy delay or reverberation effects have been applied; and For the perceptual assessment of the singing voice range, 4) the sustained note had a duration of more than 500 ms Freddie Mercury: The Solo Collection as well as 23 commer- and contained audible cyclic frequency modulations. cially available recordings of Queen were considered. The Data annotation and f0 extraction was conducted with the assessment was conducted by author C.T.H. by matching the software Praat. For f0 extraction, Praat’s autocorrelation perceived pitch of candidate sound samples, sung at algorithm was used with these parameters: timeStep: extremely high and low , to the perceived musical 10 ms; frequency range 70–600 Hz; voicingThreshold: 0.3; pitch of a tuned to A4 ¼ 440 Hz (equal temperament). octaveJumpCost: 0.35; silenceThreshold: 0.03; octaveCost: Only those sound samples were considered where Freddie 0.07; voicedUnvoicedCost: 0.14. After the automated analysis, Mercury could be clearly identified as the exclusively audible all extracted fundamental frequency contours were superim- singer, and no background vocals were included in the posed upon the respective spectrograms and then visually analysis. assessed for correct f0 extraction. Nine segments were excluded from the analysis since they contained obvious errors (mostly octave jump errors during the occurrence of Descriptive analysis of functional singing voice subharmonics). parameters A total of 128 segments of sustained notes (from 15 of the

Downloaded by [81.141.59.99] at 06:46 19 April 2016 In speech and singing, the spectral composition of the emit- 28 songs in The Acappella Collection) with vibrato (11,475 ted sound is causally determined by the configuration of the data points total, 114.75 seconds total duration), having reli- sound production organ, constituted by the respiratory sys- ably estimated f0 contours of over 500 ms duration each, were tem, the larynx, and the vocal tract. At the laryngeal level, considered for further analysis. For each segment, three quan- two main determinants are constituted by the vocal register tities were computed: 1) the average fundamental frequency and the degree of posterior vocal fold adduction, controlled f0 ; 2) the average absolute deviation from the mean musical along the dimension of ‘breathy’ versus ‘pressed’ (7,8). pitch Dc as an estimator of vibrato extent; and 3) the domin- Without any physiological or biomechanical data of vocal ant modulation frequency fmod (see further below). fold vibration the assessment of the laryngeal configuration Vibrato extent (i.e. the modulation amplitude) is typically during singing is difficult, particularly when based on com- expressed in semitones or cents, where one octave equals 12 mercial recordings whose spectral composition could have semitones or 1,200 cents (11). For the purpose of this ana- been changed by sound engineers and by influences of the lysis, the frequency values of the fundamental frequency con- recording equipment (microphone and amplifier transfer tours were converted to cents relative to middle C as functions, data conversion, recording device), room acoustics f0½i (reverberation), and the mouth-to-microphone distance logðf Þ c½i¼1200 0 C4 (1) (high-frequency selective directionality) (9). Nevertheless, a logð2Þ small number of exemplary sound samples are discussed here, based on perceptual and (partly) spectral analysis, and where f0[i] is the fundamental frequency estimate at time hypothetical physiological interpretations as regards vocal index i, and f0_C4 is the fundamental frequency of middle C register and posterior vocal fold adduction are suggested. (C4), calculated as 440 2ˆ(–9/12) 261.63 Hz. The average The perceptual assessments were conducted by the four musical pitch c of the f0 contour, expressed in cents from LOGOPEDICS PHONIATRICS VOCOLOGY 3

middle C, was then computed as Freddie Mercury’s vibrato, the dominant modulation fre- quency (fmod) was computed with Fourier analysis. For each 1 Xn1 c ¼ c½i (2) segment, the DC offset (i.e. the mean value) of each f0 con- n i¼0 tour was removed. The normalized f0 contour was scaled by a Hann window, and then zero-padded to eight times its ori- Finally, the average absolute deviation from the mean ginal length. The resulting time series vectors were subjected musical pitch Dc, representing vibrato extent, was computed to a forward FFT, and the amplitude spectrum was calcu- as lated. The frequency of the spectral maximum in the ampli- 1 Xn1 Dc ¼ jjc c½i (3) tude spectrum was considered as the dominant modulation n i¼0 frequency, fmod. In order to rule out artifacts, four segments As a reference, for completely regular vibrato with per- with fmod < 3 Hz were excluded from the analyzed data set. fectly sinusoidal frequency modulation, the vibrato extent As a by-product of the FFT analysis of the f0 contours, would equal twice the (maximum) modulation amplitude the amplitude difference of the two strongest frequency com- divided by p, i.e. the amplitude divided by about 1.57. ponents of the f0 modulation contour was calculated as Consequently, a perfectly sinusoidal vibrato with a modula- DAFmod ¼ (1 – A2)/A1, where A1 and A2 are the two stron- tion amplitude of 61 semitone would correspond to about gest components found in the normalized amplitude spec- Dc 63.7 cents. In order to rule out analysis artifacts, four trum of the f0 contour (dark triangles and light diamonds, segments with Dc < 127.3 (representing a perfectly sinus- respectively, in Figure 1A and B). The DAFmod parameter is oidal vibrato with an extent of more than two semitones) an estimator of the vibrato regularity. It can theoretically were excluded from the analyzed data set. assume values in the range of 0 to 1, where 0 would repre- A completely regular vibrato naturally contains only one sent f0 modulation with the two strongest frequency compo- modulation frequency (see e.g. Figure 1A for an example of nents in the f0 contour being equally strong, and 1 would Luciano Pavarotti’s vocal vibrato with an almost sinusoidal represent purely sinusoidal frequency modulation with one modulation frequency of about 5.7 Hz—example 1.1 from single modulation frequency component. When considering Miller (12)). In contrast, preliminary inspection of Freddie the preliminary analysis done in Figure 1, the respective Mercury’s vocal vibrato suggested more irregular frequency DAFmod values found in Luciano Pavarotti’s and Freddie modulation patterns, caused by the superposition of more Mercury’s vibrato are about 0.89 and 0.11, suggesting a quite than one modulation frequency component (see Figure 1B regular vibrato for Luciano Pavarotti and an irregular vibrato for an example, taken from ‘’, t ¼ 71 s with two almost equally strong modulation frequency com- in The Acapella Collection release (t ¼ 132 s in the A Night ponents for Freddie Mercury. at the album), ‘go’ within the phrase ‘Good-bye every- body, I’ve got to go’). Imitation and analysis of rough voice production When considering the f0 contour in Figure 1B it became apparent that time domain methods (like peak-picking) for Preliminary perceptual analysis suggested that a portion of determining the number of vibrato cycles per second are the available data material was sung with a pronounced problematic: it is unclear whether the small ripples around t degree of roughness. The phenomenon was most clearly aud- ¼ 1 should be counted as individual cycles or not. Therefore, ible in sustained notes found at t ¼ 31 s, 39 s, 120 s (‘on’

Downloaded by [81.141.59.99] at 06:46 19 April 2016 in order to obtain a robust numerical parameter for assessing within the phrase ‘turn it on’) in the a-cappella track ‘Let’s

(A)

(B)

Figure 1. Illustration of vibrato characteristics analysis. Preliminary data for two sustained notes sung by Luciano Pavarotti (A) and Freddie Mercury (B), see text. Left panels: fundamental frequency contour; right panels: normalized amplitude spectrum of fundamental frequency contour computed by Fourier analysis (see Methods). The two strongest spectral components within the range of 2–13 Hz (light gray vertical lines) are marked by a dark triangle and a light diamond, respectively, and their relative amplitude difference DAFmod is indicated by the dashed line. 4 C. T. HERBST ET AL.

(A)

(B)

Figure 2. A: Interquartile range distributions of speaking fundamental frequency estimates for the six analyzed interviews. The whiskers indicate the 5th and the 95th percentile, respectively, and the superimposed stars indicate the group means. B: Histogram of all speaking fundamental frequency data, grouped by interviewer (Dolezal: dark gray; Wigg: light gray; Cumulative: white).

Table 1. Average and median speaking fundamental frequency values (including 5th and 95th percentile) for the six analyzed interviews. Year Interviewer Number of data points Average f0 (Hz) Median f0 (Hz) 5th percentile f0 (Hz) 95th percentile f0 (Hz) 1984 Dolezal 26,124 119.2 111.1 87.1 175.9 1984 Wigg 39,487 142.2 125.6 96.5 253.4 1985 Wigg 11,795 130.4 121.5 93.7 187.3 1986 Wigg 22,772 120.9 107.2 86.2 218.7 1987 Dolezal 13,816 119.6 111.4 87.6 173.2 1987 Wigg 20,424 137.7 124.8 88.3 240.5

Turn It On’ from Freddie Mercury: The Solo Collection a sampling frequency of 16,000 Hz. The digitized sound was Downloaded by [81.141.59.99] at 06:46 19 April 2016 (2000, EMI/Parlophone). The respective sound examples stored in the smp file format using the software Soundswell were extracted and played back to a professional rock singer and Phog (Electronix Hitech, T€aby, Sweden), and was later (co-author D.Z.-B.) who served as the subject in previous converted to and stored as uncompressed wav. studies (13,14). The singer was asked to imitate these sounds, and phonation was recorded with a 70 rigid endoscope (Karl Storz, Tuttlingen, Germany) attached to a Hispec 1 Results b/w high-speed camera (Fastec Imaging, San Diego, CA, Speaking fundamental frequency USA) which was operated at 4,132 frames per second, with a 300 W light source ( GMBH, Fundamental frequency data extracted from the analyzed Knittlingen, Germany). The video recordings were stored on interviews was not normally distributed. Overall, an average a computer as uncompressed AVI files. Digital kymograms speaking fundamental frequency of 130.1 Hz was computed, (DKGs) (15,16) were extracted with a Jython plugin (http:// with a median at 117.3 Hz and the 5th and 95th percentiles homepage.univie.ac.at/christian.herbst/index.php?page¼fiji) at 89.1 Hz and 231.4 Hz, respectively. Boxplots of the funda- written by author C.T.H. for the Fiji image analysis frame- mental frequency extracted from the six interviews are shown work (17). in Figure 2. The mean and median values are provided in The acoustic signal was recorded with a Sennheiser MKE- Table 1. In order to test whether the interviewer had an 2 microphone (Sennheiser, Wennebostel, Germany) mounted effect on the speaking fundamental frequency, all data points on a headset at 15 cm distance from the singers’ mouth, were pooled by interviewer. Interviews conducted by Rudi using a custom-built microphone amplifier. The sound was Dolezal had a mean speaking fundamental frequency of digitized with a LSI LSI PC/C32 sound card (Loughborough 119.3 Hz, with the median at 111.2 Hz and the 5th and 95th Sound Images plc, Loughborough, Leicestershire, England) at percentiles at 87.3 and 174.5 Hz, respectively. The interviews LOGOPEDICS PHONIATRICS VOCOLOGY 5

conducted by Andrew Wigg had generally higher speaking within the phrase ‘let’s turn it on’), which was most probably fundamental frequency values, with an average of 134.6 Hz, produced in register. the median being at 120.2 Hz and the 5th and 95th percen- tiles at 90.0 Hz and 245.7 Hz, respectively. A Kruskal–Wallis Descriptive analysis of functional singing voice rank sum test performed on the non-normally distributed parameters data with the software R (18) revealed a highly significant difference between the two groups (Dolezal versus Wigg, Overall, based on perceptual assessment, Freddie Mercury P < 0.001). seemed to have ample control over vocal registration and ‘blending the registers’, i.e. mixing the chest and falsetto registers). More often than not, both these registers tended Singing voice range to have comparable timbral characteristics. Sounds produced in the falsetto register, where unambiguously identifiable (e.g. The lowest note found in the analyzed data corpus was a ‘Let’s Turn It On’ at t ¼ 23.2–25.2 s), could contain har- short F#2 (about 92.2 Hz) in ‘Don’t Try Suicide’ (The Game), monic content up to 10 kHz and sometimes even beyond. found at t ¼ 52.5 s (‘so’ within the phrase ‘So you think it’s (Note, though, that any existing harmonic content could have the easy way out?’). A series of sustained notes at musical been amplified by the sound engineer during the post-pro- pitch G2 (about 98 Hz) were found in ‘Get Down Make duction stage of the recording.) A few examples where the Love’ at t ¼ 58.7 s, 61.8 s, 116.7 s, 120 s, 146.7 s, and 149.9 s choice of vocal registration was more evident due to abrupt in the version from The Acapella Collection (‘down’ within transitions between the registers include: 1) ‘Nevermore’ the phrase ‘Ev’ry time I get hot you wanna cool down’)— (Queen II), t ¼ 15–28 s (‘Can’t you see? Listen to the breeze, note that in relation to the version on the News of the World whisper to me please. Don’t send me to the path of never- album, these time offsets are shifted by about 12.5 seconds. more’); 2) ‘Seaside Rendezvous’ (A Night at the Opera), t ¼ For the highest notes a couple of candidates were found: 1) a 34–45.2 s (‘Can we do it again? Can we do it again some- sustained E5 (about 659.3 Hz) in ‘Hang on in There’ (The time? (ooh I like that) Fantastic, c’est la vie Madames et Miracle)att¼ 134.5 s (‘wait’ within the phrase ‘wait for that Monsieurs’); 3) ‘Love of My Life’ (A Night at the Opera), t ¼ moment’); 2) a sustained F5 (about 698.5 Hz) in ‘All God’s 131–134.6 s (‘How I still love you’), where there was a per- People’ (Innuendo)att¼ 42 s (background voice ‘yeah’, ceivable and abrupt timbral shift between ‘love’ (sung in chest right after ‘majesty around the world’), which was masked by register) and ‘you’ (falsetto register); or 4) ‘Keep Passing the other sung sounds in the recording; 3) a very short but Open Windows’ (The Works)att¼ 18–23 s, where ‘need’ of clearly perceivable G5 (about 784 Hz) in ‘Let’s Turn It On’ the phase ‘love is all you need’ again bore perceptual clues of falsetto voice production. (Freddie Mercury: The Solo Collection)att¼ 25 s (‘on’ Downloaded by [81.141.59.99] at 06:46 19 April 2016

Figure 3. Examples of ‘breathy’ and ‘pressed’ sound production. A and B: narrow-band spectrograms (FFT bin width 10.77 Hz, dynamic range 55 dB); C and D: corre- sponding amplitude spectra, extracted at t ¼ 20.5 s and t ¼ 2 s, respectively. The light gray (orange color online) line indicates the noise floor at about –40 dB and –55 dB, respectively. 6 C. T. HERBST ET AL.

(A) (B) (C) (D)

Figure 4. Interquartile range distributions of computed vibrato parameters. The whiskers indicate the 5th and the 95th percentile, respectively. A: Fundamental fre- quency of analyzed notes sung with vibrato; B: dominant modulation frequency fmod; C: vibrato extent; D: vibrato regularity measure DAFmod.

Freddie Mercury appeared to have had good control over the phonation type along the dimension ‘breathy’ to ‘pressed’. Two such examples illustrating the range of timbral possibilities are illustrated in Figure 3. The example for breathy phonation (Figure 3A and C), taken from ‘Teo Toriatte’ (A Day at the Races) contained clearly audible broadband noise above 1.5 kHz, which could hypothetically have been created by turbulent noise (19), presumably caused by incomplete vocal fold adduction. This broadband noise was strong enough clearly to highlight several vocal tract res- onances, identified as the second, third, and fourth formant in the spectrogram (see arrow annotation in Figure 3A and C). An example for the ‘pressed’ phonation type is illus- trated in Figure 3B and D. Apart from the perceptually strained quality of the sound and the apparent lack of vocal vibrato, the example was characterized by a clearly audible release of subglottal pressure at the end of the phrase (t ¼ 5.2 s). When the normalized spectra of the two antagonistic examples are compared, clear differences of spectral compos- ition become apparent. The ‘breathy’ example, although pol- luted by keyboard accompaniment, contained a far smaller Figure 5. Narrow-band spectrograms of subharmonic voice production. A: number of noteworthy harmonics, and the normalized spec- Freddie Mercury: ‘Let’s Turn It On’ (The Acapella Collection), t ¼ 67 s; B: Imitated Downloaded by [81.141.59.99] at 06:46 19 April 2016 trum had a higher noise floor (about –40 dB overall) as com- sound sung by professional rock singer D.Z.-B. pared to the ‘pressed’ example (about –55 dB)—see light gray (orange color online) lines in Figure 3C and D. of fundamental frequency were analyzed with linear regres- sion analysis. While there was only a slight tendency for Vibrato decreased dominant modulation frequency as f0 increases 2 (fmod ¼ –0.0033 f0 þ 8.19, R ¼ 0.09), there existed a slightly Boxplots of the calculated vibrato parameters are shown in stronger (but still weak) inverse correlation between the f0 Figure 4. The average fundamental frequency of the analyzed 2 and modulation extent (Dc ¼ –0.0674 f0 þ 58.98, R ¼ vibrato notes was 350.2 Hz, with a median of 347.0 Hz and 0.21). In other words, the vibrato extent tended to be smaller 5th and 95th percentiles at 199.6 Hz and 494.9 Hz, respect- at higher pitches. ively. The mean dominant modulation frequency fmod was 7.0 Hz (median 7.2 Hz, 5th and 95th percentiles at 5.2 Hz and 8.2 Hz). The average vibrato extent Dc was calculated to Subharmonic voice production be 35.0 cents (corresponding to a perfectly sinusoidal vibrato with an amplitude of about 55 cents or 0.55 semitones—see An example of Freddie Mercury’s phonation with a perceived Methods), with a median of 32.1 cents and the 5th and 95th degree of roughness is illustrated in Figure 5A. In the spec- percentiles at 19.6 cents and 64.4 cents, respectively. The trogram, the harmonics appear as integer multiples of the vibrato regularity estimator DAFmod had a mean value of fundamental frequency, and so-called subharmonics emerge 0.57, with a median at 0.60 and 5th and 95th percentiles at at integer multiples of a third of the fundamental frequency, 0.09 and 0.97, respectively. The variation of dominant modu- constituting a period-3 subharmonic pattern (period tri- lation frequency fmod and modulation extent rC as a function pling). The same phenomenon is seen in the vocal imitation LOGOPEDICS PHONIATRICS VOCOLOGY 7

vibration of right vocal fold ventricular fold vibration

right (A) (C) (D)

left

right (B)

left Time (82 ms total) ventricular fold lateral surface waves Figure 6. Endoscopic evidence of subharmonic vocal fold vibration. A and B: digital kymograms (DKGs); C and D: depiction of virtual kymographic scan lines for the DKGs shown in A and B, respectively.

produced by professional rock singer D.Z.-B. (Figure 5B). (21); and Chernobel’skii reported a SFF of 123.2 Hz Analysis of the laryngeal endoscopic HSV data from D.Z.- (610.2 Hz) for (22). Based on these data it can be B.’s phonation revealed frequency-locked vibration of the hypothesized that Freddie Mercury had a baritone’s speaking ventricular folds at a third of the frequency of vibration of voice, since the respective mean SFF values for trained the vocal folds (see Supplementary Movie M1, available were about 20 Hz higher (20–22). This assumption is further online). This was also seen in the corresponding DKGs motivated by the notion that the SFF is located at a roughly (Figure 6), extracted at two positions along the anterior–pos- predictable offset within a speaker’s total f0 range (23). terior vocal fold dimension. Anteriorly, the ventricular fold Normative values given for that offset in various studies are vibration exhibited clear lateral surface waves (similar to in the range of 3–8 semitones above the lowest achievable mucosal waves within the true vocal folds), possibly induced fundamental frequency (20,24,25), and this variability might by high levels of glottal air flow and subglottal pressure. be caused by effects of vocal training (23). Anecdotally, the Overall, the epilaryngeal tube was constricted, thus limiting suggestion that Freddie Mercury had a baritone’s voice range the visibility of the true vocal folds. The arytenoid cartilages is corroborated by a quote from his duet partner, classical (and thus also the vocal processes) were firmly adducted. Montserrat Caballe, who maintained that ‘He had a baritone voice. I told him one day, ‘‘Let’s do a small duet of baritone and soprano’’, and he said, ‘‘No, no, my fans only Discussion know me as a rock singer and they will not recognise my The main difficulty of this retrospective study was posed by voice if I sing in baritone’’’ (2). the fact that no controlled data acquisition was possible. The There was a noteworthy and statistically significant differ-

Downloaded by [81.141.59.99] at 06:46 19 April 2016 results of the acoustic and perceptual analyses conducted here ence between SFF data from interviews by Andrew Wigg ver- need thus be treated with care: not all sound samples were a- sus Rudi Dolezal. Interviews conducted by Andrew Wigg cappella recordings, variable recording tape wheel speeds contained several spoken phrases where the SFF exceeded could have slightly changed the measured fundamental fre- 200 Hz, with maximum values up to 320 Hz (which occurred quencies, and any spectral component could have been during connected speech as a means of putting emphasis on changed by the intervention of sound engineers at the post- certain words) as indicated by manual inspection of the raw production stage (therefore no formant data were analyzed). data material. Other confounding factors notwithstanding, Except for the vocal imitation by rock singer D.Z.-B. no the SFF difference might be attributed to Andrew Wigg physiological data were available to back up the physiology- being a native speaker of British English (as was Freddie related conclusions drawn from the analyzed acoustic data. In Mercury), thus creating a more familiar and personal atmos- spite of these complications, this study was successful in deliv- phere during the interviews that allowed for a somewhat ering a few insights into Freddie Mercury’s voice production higher SFF. In contrast, the Austrian Rudi Dolezal was a and singing style. These are summarized and discussed below. native speaker of German.

Speaking voice fundamental frequency Singing voice range Analysis of interviews revealed a mean SFF of 130.1 Hz Measurement of the singing voice range constituted the most (median at 117.3 Hz). Morris et al. found an average SFF of challenging task in this study, since it was solely based on 127.4 Hz (610.1 Hz) for trained middle-aged baritones and perceptual assessment of commercially available recordings. basses (20); Brown et al. documented an average SFF of The determined extreme pitches of F#2 (about 92.5 Hz) and 119 Hz (no standard deviation given) for the same group G5 (about 784 Hz) constitute a singing voice range of 37 8 C. T. HERBST ET AL.

semitones. Such values are well in line with observations by 1 to 2 Hz higher than average vibrato rates reported for non- Morris et al., who report an average voice range of 38.7 classical singers, thus being more in line with modulation semitones for middle-aged baritones (20). rates reported for vocal tremor (36,37). As such, the ten- Online sources like Wikipedia (26) report singing voice dency for higher fundamental frequency modulation rates in ranges of four octaves from F2 (about 87.3 Hz) to F6 (about Freddie Mercury’s voice may well contribute to his unique 1,396.9 Hz) for Freddie Mercury. However, in the examples vocal style. listed by these sources the vocalist is not clearly identifiable The average vibrato extent Dc of 35.38 cents (correspond- (mostly because the putative examples for extreme notes ing to a perfectly sinusoidal vibrato with an amplitude of stem from background vocals, which could have been sung about 55 cents or about 0.55 semitones) is by and large com- by other members of Queen), or the respective data sup- parable to the average values previously reported for non-clas- posedly came from vocal improvisations that were not avail- sical singing, which range from 30.9 cents (reported in percent able as data material in this study. A sustained note with a and converted to cents by C.T.H.) in singers of ‘contemporary maximum fundamental frequency of 1,347 Hz (slightly above music, pop, gospel and ’ (33) to 66.78 cents for male jazz musical pitch E6) was found in ‘It’s Late’ (News of the singers (34). These values are generally lower than the vibrato World)att¼ 370–372 s. On perceptual grounds it is unclear extent reported for classical singing, ranging from 71 cents whether this note was sung by Freddie Mercury or any other (30) to 112 in ‘Lied’ and 138 cents in operatic singing (31). member of Queen, or whether it was played by a . In a perfectly sinusoidal vibrato, the DAFmod parameter, as While such high fundamental frequencies of notes were defined in this study, assumes a value of 1. A value very close reported to be possible for a ‘male soprano’ (27) and for to 1 has been found for operatic tenor Luciano Pavarotti in males using the ‘whistle register’ (28), it can, given the ana- the preliminary analysis (see Methods and Figure 1). In con- lyzed data material, not be reliably concluded (nor can it be trast, the analysis of Freddie Mercury’s notes sung with ruled out) that such a phenomenon existed in Freddie vibrato resulted in an average DAFmod value of 0.57, and half Mercury’s voice. the analyzed samples had values of 0.6 or smaller, indicating that largely more than one modulation frequency component was found in the spectral analysis of the f0 contours. Descriptive analysis of functional singing voice The principal vibrato agent is assumed to be a muscle or parameters a group of muscles that regulates fundamental frequency. Evidence for describing the functional aspects of Freddie Typically, the cricothyroid is hypothesized to form an Mercury’s singing habits was derived from perceptual assess- agonist-antagonist pair with the thyroarytenoid and/or the ment and spectral analysis of selected sound samples. The lateral cricothyroid muscles (10,38). In the ‘standard’ case of presented analysis (see Results and Figure 3) suggested the regular vibrato (see e.g. the Pavarotti example in Figure 1), use of different vocal registers (termed ‘chest’ and ‘falsetto’ in such an agonist–antagonist system would exhibit simple this manuscript), as well as different phonation types along sinusoidal oscillation. Such a simple model could, however, the dimension of ‘breathy’ versus ‘pressed’. However, as no not explain the complex f0 modulation seen in some of physiological data (i.e. endoscopic laryngeal recordings, etc.) Freddie Mercury’s sustained notes. It is thus hypothesized were available, no rigorous proof can be given. The assump- that either a single agonist–antagonist pair exhibited complex tions made in this study vis-a-vis laryngeal configurations irregular oscillation, or more than one agonist–antagonist pairs for fundamental frequency modulation were present. Downloaded by [81.141.59.99] at 06:46 19 April 2016 and physiological quantities, such as subglottal pressure (see Figure 3 and corresponding text), remain therefore specula- Naturally, as no physiological data were available, neither position can be maintained. The proposed notion might, tive. This weakness is mainly constituted by the most obvi- however, be relevant for further physiological investigation ous fact that this work is a post hoc analysis of a singer who involving contemporary singers. would not be available for further data acquisition. Perceptually, Freddie Mercury’s irregular (and typically Nonetheless, based on the perceptual impression of the faster) vibrato is clearly audible in the sustained notes of assessed recordings it is not too far-fetched to conjecture that famous songs such as ‘Bohemian Rhapsody’ (A Night at the Freddie Mercury was rather skillful in adapting his laryngeal Opera) or ‘’ (News of the World), and configuration to musical needs, thus exhibiting a great vari- it appears to be one of the hallmarks of his vocal style. On a ability of sound for enhanced musical expression. more general level, these findings suggest that the regularity of vibrato (indicated here by the DAFmod parameter) could be Vibrato an important feature for any singing voice, which may deserve to receive further scientific attention. For Western classical singing, vibrato typically has a rate of about 5.4 Hz to 6.9 Hz (29–31). This rate may depend on age Subharmonic voice production and emotional involvement (29), and on esthetical preferen- ces which may have changed over the last 100 years (32). The analysis of rock singer D.Z.-B.’s imitation of Freddie For non-classical singing, values of 5.45 Hz (33), 6.09 Hz Mercury’s ‘rough’ singing style revealed subharmonic vibration (34), and 5.04 Hz (35) were reported. In contrast, the average (39,40) created by 3:1 frequency locking of the vocal folds and dominant frequency of Freddie Mercury’s fundamental fre- the ventricular folds: during each vibration of the ventricular quency modulation was found to be 7.04 Hz, which is about folds the vocal folds completed three oscillations. Similar LOGOPEDICS PHONIATRICS VOCOLOGY 9

vibratory patterns of the vocal and the ventricular folds as Funding information synchronized oscillators were observed in previous studies This study was partially financed by the institutional fund of the involving the same (14) and other singers (41,42). The Palacky University Olomouc, Czech Republic (C.T.H.). observed lateral surface waves along the ventricular folds are comparable to mucosal waves in true vocal folds (43,44). This type of voice production, involving vibrating References ventricular folds at various degrees, has been termed 1. 100 Greatest Singers: Freddie Mercury. New York, NY: Rolling ‘growl’ (45), ‘dist’ (14), ‘throat-singing’ (46), or ‘vocal-ven- Stone; 2011 [cited 2011 Nov 14]; Available from: http://www. roll- tricular mode’ (VVM) (42). The umbrella term ‘distortion’ ingstone.com/music/lists/100-greatest-singers-of-all-time-19691231/. (47) or ‘intentional distortion’ (ID) (48) has been suggested 2. Sullivan C. Freddie Mercury: the great enigma. London, UK: The for singing voice effects where, in addition to the ventricular Guardian; 2012. Available from: http://www.theguardian.com/ folds, also other supraglottal structures are engaged in music/2012/sep/27/freddie-mercury-great-enigma. 3. Bret D. Living on the edge: the Freddie Mercury Story. London, vibration. UK: Robson Books, 1996. Spectrograms of sound production by both D.Z.-B. and 4. Herbst CT. Freddie Mercury—Akustische Stimmanalyse. LOGOS Freddie Mercury show comparable subharmonic patterns Interdisziplin€ar. 2012;20:174–83. (period tripling). It is therefore likely that frequency-locked 5. Boersma P, Weenink D. Praat: doing phonetics by computer. ventricular fold vibration also occurred in the original sample Amsterdam, The Netherlands: Institute of Phonetic Sciences, University of Amsterdam, 2014. of Freddie Mercury’s ‘rough’ singing which was imitated in this 6. Boersma P, editor. Accurate short-term analysis of the fundamen- study. An alternative, but less likely explanation would suggest tal frequency and the harmonics-to-noise ratio of a sampled subharmonic oscillation of the vocal folds alone (49), or a fre- sound. Proceedings of the Institute of Phonetic Sciences, 1993. quency-locked vibration of the vocal folds and the aryepiglottic Amsterdam; 1993. folds (45). 7. Herbst CT, Howard DM, Svec JG. The sound source in singing – basic principles and muscular adjustments for fine-tuning vocal . In: Welch G, Howard DM, Nix J, editors. The Voice production and stage persona handbook of singing. Oxford, UK: Oxford University Press, in press. Subharmonics, as described in the previous paragraph, are a 8. Herbst CT, Hess M, Muller€ F, Svec JG, Sundberg J. Glottal adduc- possible route of a system ‘on its way to chaos’ (50), with tion and subglottal pressure in singing. J Voice 2015;29:391–402. 9. Svec JG, Granqvist S. Guidelines for selecting microphones for potential communicative relevance (51). Their occurrence human voice production research. Am J Speech Lang Pathol aids in creating the impression of a sound production system 2010;19:356–68. driven to its limits, even while used with great finesse. These 10. Hirano M, Hibi S, Hagino S. Physiological aspects of vibrato. In: traits, in combination with the fast and irregular vibrato, Dejonckere PH, Hirano M, Sundberg J, editors. Vibrato San might have helped create Freddie Mercury’s eccentric Diego: Singular Publishing Group, 1995. p. 9–33. 11. Young RW. Terminology for logarithmic frequency units. and flamboyant stage persona. This stage persona and its J Acoust Soc Am 1939;11:134–9. apparent constant concourse with Freddie Mercury’s more 12. Miller DG. Resonance in singing Princeton, NJ: Inside View private side is best characterized by some quotes from his Press, 2008. interviews: ‘I’m just a musical prostitute’ (Dolezal, 1984); ‘I 13. Zangger Borch D, Sundberg J. Some phonatory and resonatory am the nicest person you could meet, my dear. It’s just characteristics of the rock, pop, soul, and Swedish dance band that I am handicapped, and therefore I have to sort of virtu- styles of singing. J Voice 2010;25:532–7. Downloaded by [81.141.59.99] at 06:46 19 April 2016 14. Zangger Borch D, Sundberg J, Lindestad PA, Thalen M. Vocal ally fight my so-called stage persona. Most of it fold vibration and voice source aperiodicity in ‘dist’ tones: a study sort of works against me’ (Wigg, 1984); and, concerning the of a timbral ornament in rock singing. Logoped Phoniatr Vocol audience’s expectations and losing control: ‘Because of my 2004;29:147–53. persona on stage they think that I carry on this way off- 15. Svec JG, Schutte HK, Sram F. Videokymography: high-speed line stage. If I did, I would have been dead a long time ago’ scanning of vocal fold vibration. In: McCafferty G, Coman W, (Wigg, 1984). Carroll R, editors. XVI World Congress of Otorhinolaryngology Head and Neck Surgery, Sydney ‘97 Proceedings, vol 2. Bologna: Monduzzi Editore, 1997. p. 1685–8. 16. Wittenberg T, Tigges M, Mergell P, Eysholdt U. Functional imag- Acknowledgements ing of vocal fold vibration: digital multislice high-speed kymogra- This study is a work of ‘fan-science’. Author contributions: C.T.H. con- phy. J Voice 2000;14:422–42. ceived the study, collected the data material, conducted the perceptual 17. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, and quantitative analysis, and wrote the manuscript. S.H., D.Z.-B. and Pietzsch T, et al. Fiji: an open-source platform for biological- P.A.L. collected the vocal imitation data and contributed to the final image analysis. Nat Methods 2012;9:676–82. version of the manuscript. We are very grateful to Dr Hans Larsson 18. R: A language and environment for statistical computing. In: (Karolinska University Hospital) for his help in acquiring the respective Team RDC, editor. Vienna, Austria: R Foundation for Statistical endoscopic data. Parts of this manuscript were previously published in: Computing, Vienna, Austria, 2011. Christian T. Herbst (2012). Freddie Mercury—Akustische Stimm- 19. Childers D, Lee CK. Vocal quality factors: analysis, synthesis, and Analyse. L.O.G.O.S. Interdisziplin€ar, 20 (3), 174–83. perception. J Acoust Soc Am 1991;90:2394–410. 20. Morris RJ, Brown WS Jr, Hicks DM, Howell E. Phonational pro- files of male trained singers and nonsingers. J Voice 1995;9:142–8. Declaration of interest 21. Brown W, Morris RJ, Hollien H, Howell E. Speaking fundamental frequency characteristics as a function of age and professional The authors report no conflicts of interest. singing. J Voice 1991;5:310–15. 10 C. T. HERBST ET AL.

22. Chernobel’skii SI. [The speaking fundamental frequency and the 39. Titze IR, Baken RJ, Herzel H. Evidence of chaos in vocal fold singing voice type]. Vestn Otorinolaringol 2010(6):36–7. [Russian] vibration. In: Titze IR, editor. Vocal fold physiology: frontiers in 23. Baken RJ, Orlikoff RF. Clinical measurement of speech and voice basic science San Diego, CA: Singular Publishing Group, 1993. p. 2nd ed. Singular Publishing, Thompson Learning; 2000. 143–88. 24. Awan SN. Superimposition of speaking voice characteristics and 40. Herzel H, Berry D, Titze IR, Saleh M. Analysis of vocal disorders phonetograms in untrained and trained vocal groups. J Voice with methods from nonlinear dynamics. J Speech Hear Res 1993;7:30–7. 1994;37:1008–19. 25. Friedrich G, Dejonckere PH. Das Stimmdiagnostik Protokoll 41. Bailly L, Henrich N, Pelorson X. Vocal fold and ventricular fold der European Laryngological Society (ELS)—erste Erfahrungen im vibration in period-doubling phonation: physiological description Rahmen einer Multizenterstudie. Laryngo-, Rhino-, Otologie and aerodynamic modeling. J Acoust Soc Am 2010;127:3212–22. 2005;84:744–52. 42. Fuks L, Hammarberg B, Sundberg J. A self-sustained vocal-ven- 26. Wikipedia. Freddie Mercury. Wikimedia Foundation; 2015 [cited tricular phonation mode: acoustical, aerodynamic and glotto- 1 June 2015]; Available from: http://en.wikipedia.org/wiki/ graphic evidences. Royal Institute of Technology - Speech, Music Freddie_Mercury. and Hearing Quarterly Progress and Status Report (Stockholm). 27. Directed by Francesca Kemp. London: BBC Four, 2006. 1998;3/1998:49–59. 28. Edgerton M. Pitch profile of the glottal whistle. Malaysian 43. Hirano M. Clinical examination of voice New York: Springer- Journal of Science 2013;32:78–85. Verlag, 1981. 29. Sundberg J. Acoustic and psychoacoustic aspects of vocal vibrato. 44. Berke G, Gerratt BR. Laryngeal biomechanics: an overview of In: Dejonckere PH, Hirano M, Sundberg J, editors. Vibrato San mucosal wave mechanics. J Voice 1993;7:123–8. Diego: Singular Publishing Group, 1995. p. 35–62. 45. Sakakibara K-I, Fuks L, Imagawa H, Tayama N, editors. Growl 30. Prame E. Vibrato extent and intonation in professional Western voice in ethic and pop styles. Proceedings of the International lyric singing. J Acoust Soc Am 1997;102:616–21. Symposium on Musical Acoustics (ISMA2004); 2004. Nara, 31. Johnson-Read L, Chmiel A, Schubert E, Wolfe J. Performing Japan, 2004. Lieder: expert perspectives and comparison of vibrato and singer’s 46. Lindestad PA, Sodersten M, Merker B, Granqvist S. Voice source formant with opera singers. J Voice 2015;29:645.e15–32. characteristics in Mongolian ‘‘throat singing’’ studied with high- 32. Titze IR. Principles of voice production. National Center for speed imaging technique, acoustic spectra, and inverse filtering. J Voice and Speech, 2000. Voice 2001;15:78–85. 33. Reddy AA, Subramanian U. Do rate and extent of vibrato vary 47. Titze I. Deliberate use of distortion in singing. In: Kuhl P, Crum with gender in untrained singers? International Journal of L, editors. 16th International Congress on Acoustics and 135th Interdisciplinary Research and Innovations 2014;2:1–5. Meeting Acoustical Society of America: Vol I The Sound of the 34. Manfredi C, Barbagallo D, Baracca G, Orlandi S, Bandini A, Future: A Global View of Acoustics in the 21st Century. Seattle: Dejonckere PH. Automatic assessment of acoustic parameters of University of Washington, 1998. p. 435–6. the singing voice: application to professional Western operatic 48. McGlashan J, Sayles M, Kjelin H, Sadolin C, editors. Vocal effects and jazz singers. J Voice 2015;29:517.e1–9. in singing: a study of intentional distortion using laryngostrobo- 35. Pecoraro G, Curcio DF, Behlau M, editors. Vibrato rate variability scopy and electrolaryngography. AQL2013, 10th international in three professional singing styles: opera, rock and Brazilian conference ‘‘Advances in quantitative laryngology, voice and country. Proceedings of Meetings on Acoustics; 2013. Montreal, speech research’’; 2013. Cincinnati, OH: AQL Press, 2013. CA: Acoustical Society of America, 2013. 49. Herzel H, Berry D, Titze I, Steinecke I. Nonlinear dynamics of 36. Ramig L, Shipp T. Comparative measures of vocal tremor and the voice: signal analysis and biomechanical modelling. Chaos vocal vibrato. J Voice 1987;1:163–7. 1995;5:30–4. 37. Hertegard S, Granqvist S, Lindestad PA. Botulinum toxin injec- 50. Lauterborn W, Cramer E. Subharmonic routes to chaos observed tions for essential voice tremor. Ann Otol Rhinol Laryngol in acoustics. PhysRevLett 1981;47:1445–8. 2000;109:204–9. 51. Fitch WT, Neubauer J, Herzel H. Calls out of chaos: the adaptive 38. Titze IR, Story B, Smith M, Long R. A reflex resonance model of significance of nonlinear phenomena in mammalian vocal pro- vocal vibrato. J Acoust Soc Am 2002;111(5 Pt 1):2272–82. duction. Animal Behaviour 2002;63:407–18. Downloaded by [81.141.59.99] at 06:46 19 April 2016