Sounds of the Human Vocal Tract

Sounds of the Human Vocal Tract

INTERSPEECH 2017 August 20–24, 2017, Stockholm, Sweden Sounds of the Human Vocal Tract Reed Blaylock, Nimisha Patil, Timothy Greer, Shrikanth Narayanan Signal Analysis and Interpretation Laboratory University of Southern California, USA [email protected], {nimishhp, timothdg}@usc.edu, [email protected] Do speech and beatboxing articulations share the same Abstract mental representations? Previous research suggests that beatboxers only use sounds that Is beatboxing grammatical in the same way that exist in the world’s languages. This paper provides evidence to phonology is grammatical? the contrary, showing that beatboxers use non-linguistic How is the musical component of beatboxing represented articulations and airstream mechanisms to produce many sound in the mind? effects that have not been attested in any language. An analysis of real-time magnetic resonance videos of beatboxing reveals To address these questions, a comprehensive inventory of that beatboxers produce non-linguistic articulations such as beatboxing sounds and their articulations must first be ingressive retroflex trills and ingressive lateral bilabial trills. In compiled. addition, beatboxers can use both lingual egressive and Some steps toward this goal have already been taken. pulmonic ingressive airstreams, neither of which have been Splinter and TyTe [5] proposed a Standard Beatbox Notation reported in any language. (SBN) to encode contrasts between different beatboxing The results of this study affect our understanding of the sounds, and Stowell [6] uses a combination of new characters limits of the human vocal tract, and address questions about the and characters from the International Phonetic Alphabet (IPA) mental units that encode music and phonological grammar. to represent some sounds of beatboxing. Stowell and Plumbley [7] and Stowell [8] impressionistically describe the acoustic Index Terms: beatboxing, MRI, singing, speech production properties of certain beatboxing sounds compared to speech and singing sounds. Lederer [9] provides a detailed acoustical 1. Introduction analysis of a few common beatboxing sounds, and explains how Although the human vocal tract is most well-known and widely to produce those sounds based on descriptions provided by the studied for its speech movements, it also produces skilled study’s subject. All of these studies, however, are based on movements in domains other than language. Many of those impressionistic and acoustic data; none examine the articulatory skilled movements are unattested in languages used today. For strategies used in beatboxing. instance, throat singers and metal singers use epilaryngeal In an articulatory study using real-time magnetic resonance vibration to create a growl-like vocalization [1]. This growl imaging (rtMRI), Proctor et al. [10] reported the productions of appears in speech as a marker of emotion, but is not contrastive one beatboxer whose sounds were all produced with pulmonic in any known phonologies. Similarly, some operatic singers egressive, glottalic egressive (“ejective”), or lingual ingressive produce a singer’s formant (a clustering of the third, fourth, and (“click”) airstreams, and had articulations amenable to fifth formants) to ensure that the sound of their voice can be transcription using IPA—an orthographic system that was heard over an orchestra [2]. The production of the singer’s designed to describe speech sounds that are contrastive in formant is the result of precise articulatory positioning that has languages of the world. no known role in speech. This paper describes the articulatory As Proctor et al. point out, their study is limited by having patterns used in beatboxing, a type of musical expression that only one beatboxer; this beatboxer apparently did not have any requires skilled movements of the vocal tract—many of which non-linguistic sounds in their repertoire. Further observation of are not found in known languages. skilled beatboxers, including the acoustic research cited above, In beatboxing, the human vocal tract is used to emulate clearly indicates that beatboxers deploy articulator strategies percussion and other physical or synthesized sounds. Vocal that are not represented in the sounds of the world’s languages. percussion is a component of many musical styles around the What those strategies are, however, is currently unknown. world, including Southern Indian Carnatic musical traditions The aim of this paper is to show that beatboxers use non- and North American scat singing [3]. In the past few decades, linguistic articulations in addition to linguistic (or pseudo- beatboxing has been used in hip hop and a cappella music, and linguistic) productions. In particular, this paper highlights the as a stand-alone musical style [4]. articulatory configurations and airstream mechanisms that are Beatboxing is of particular interest to speech scientists unattested in language, which until now have not been because it uses speech sounds or speech-like sounds to create described for beatboxing. The findings of this paper point to music that often has no linguistic content. In beatboxing, further avenues of investigation for understanding skilled vocal percussion sounds like kick drums, hi-hats, and snares can be tract movement in different domains, including the possibility emulated by the vocal tract with bilabial ejectives, alveolar of a phonotactic grammar for beatboxing. affricates, and trills—all of which appear in language. MR videos of the sounds in this paper can be found at Given the apparent similarities between beatboxing and http://sail.usc.edu/span/beatboxing2017/index.html. speech, questions like the following emerge: Copyright © 2017 ISCA 2287 http://dx.doi.org/10.21437/Interspeech.2017-1631 2. Method style of beatboxing and are comparable to sounds made in language, they are among the first for new beatboxers to learn, Five beatboxers were asked to produce beatboxing sounds in and they are used frequently at all levels of beatboxing [4, 5]. isolation and in musical rhythms (“beats”), and to speak several passages, in an MRI machine. Of those five participants, the Table 1: Sample of beatboxing sounds that appear productions of just two—participants IF (second author of this linguistic paper) and AF, both females—are reported in the present study. Participant IF was an intermediate-level beatboxer, and Name IPA Description participant AF was an advanced-level beatboxer. Skill level was Kick drum, [p͡ ɸ’] Voiceless glottalic egressive defined based on the participant’s level of artistic control and “B” (ejective) bilabial affricate the apparent difficulty of the sounds in their repertoire. Both Closed hi- [ts’] Voiceless glottalic egressive participants reported English as their native language, and IF hat (ejective) alveolar affricate also reported fluency in Marathi and Hindi. Of the sounds PF snare, [pf’:] Voiceless glottalic egressive reported, only some were produced by both IF and AF. “PF” (ejective) labial affricate (with long The participants were asked to produce all of the percussion frication) effects in their repertoire and demonstrate some beatboxing Voiced [r̻ ] Voiced pulmonic egressive laminal sequences by performing in short intervals as they lay supine in tongue bass alveolar trill an MRI scanner bore. For each sound the elicitation was repeated at least three times in a single MRI recording and subsequently used in a sample beat pattern. In addition, some 3.2. Non-linguistic sounds speech passages were recorded, and a full set of the subject's American English vowels was elicited using the [h_d] corpus. 3.2.1. Non-linguistic airflow The subjects were paid for participation in the experiment. The Though some of the sounds reported here are similar to sounds study presented in this paper draws from a subset of this data. used in language, other beatboxing sounds were also observed Data were collected using an rtMRI protocol developed for that are not attested in any language known to the authors. the dynamic study of vocal tract movements, especially during One source of novelty is that beatboxers use a wider range speech production [11, 12]. The subjects’ upper airways were of airstream mechanisms. Speech relies almost exclusively on imaged in the midsagittal plane using a gradient echo pulse four airstreams: pulmonic egressive (e.g. [k], glottalic egressive sequence (TR = 6.004 ms) on a conventional GE Signa 1.5 T (e.g. [p’]), glottalic ingressive (e.g. [ɓ]), and lingual ingressive scanner (Gmax = 40 mT/m; Smax = 150 mT/m/ms), using an 8- (e.g. [ʘ]). Previous studies in beatboxing have differentiated channel upper-airway custom coil. sounds as using “inhalation” or “exhalation,” but have not The slice thickness for the scan was 6 mm, located specified whether the airflow was due to glottal, lingual, or midsagittally over a 200 mm × 200 mm field-of-view; image pulmonic activity [5, 7, 8, 9]. It has been reported that resolution in the sagittal plane was 84 × 84 pixels beatboxers use all airstreams found in language, but does not (2.4 × 2.4 mm). The scan plane was manually aligned with the report on whether they use non-linguistic airstreams [9]. This midsagittal plane of the subject’s head. The frames were study finds that beatboxers can—and frequently do—use non- retrospectively reconstructed to a temporal resolution of 12ms linguistic pulmonic ingressive and lingual egressive airstreams. (2 spirals per frame, 83 frames per second) using a temporal Pulmonic ingressive airflow is inhalation. Inhalation is

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    5 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us