The Music of Dialect
The Music of Dialect
Composition based on Newfoundland discourse models
by Jason Noble
Published: April 15, 2025
This two-part lecture was originally delivered at the Université de Moncton in fall 2023. The original slide deck, containing images and audio, can be accessed here:
Slide 1
Human speech surrounds us with music almost constantly. Most of the time, we receive this speech in a functional way: we hear the words and focus on their meaning, our main interest being the exchange of information, not the aesthetics of the sound world. But all we have to do is make a choice, consciously change our listening mode, press a button in the brain and decide to listen to the music in speech, to suddenly find ourselves in a world of fascinating and beautiful melodies, rhythms and timbres. This is just a slightly more specific case of the lesson of John Cage and so many others, who encourage us to open our ears to the music that is there for us all the time, if only we allow ourselves to hear it. In this case, I'm talking specifically about dialect music.
For we Atlantic Canadians, well known for our distinct dialects, this is an important and personal topic. The colours of our language are a poignant part of our cultural identity, our sense of community. They contribute to our literature, our sense of humor, our folk music, our everyday interactions. Our speech patterns, both phonetically and acoustically, are incredibly rich and unique in all the world. So I was very motivated to celebrate the dialects of my home province of Newfoundland and Labrador in my work as a contemporary composer.
Slide 2
Today, I present the results of a long-term research-creation project, carried out in collaboration with my good friend, guitarist Steve Cowan. We've already presented this work in a number of contexts, most recently in concert at Memorial University of Newfoundland in St. John's, Newfoundland, and it is with great pleasure that I present it today in my new home, the Université de Moncton.
Slide 3
I'm by no means the first composer to note and find inspiration in the musical qualities of speech. There's a long history of musical practices and genres that deal with this, for example in operatic recitative and in performance indications such as parlando and declammando. And of course nowadays much of the world's most popular music is based in musical speech, in the genres of rap, hip-hop, etc. In the world of contemporary music, there's a wide repertoire of different approaches to speech-based music, some of which have given rise to astonishing works that have become contemporary classics. One such approach is transcribing or adapting vocal recordings for musical instruments; it has given rise to such remarkable works as Steve Reich's Different Trains, Rolf Wallin's Concerning King and Jonathan Harvey's Speakings.
Slide 4
A particularly influential example for this project is Quebec composer and performer René Lussier's Le trésor de la langue, which features interviews with people from the province of Quebec about the importance of language. Lussier says: "It's remarkable the melodies we say to each other every day! And no one is in the least disturbed by these phrases, but transposed into music, they can become surprising, even disturbing!" For the moment, I'll deal mainly with the musical transcription of speech, which is already a rich, multi-dimensional process that invites a highly analytical mode of listening and is at the heart of our methods in this project.
Slide 5
My interest in speech-based composition stems from my Newfoundland heritage, which I share with Steve. The province of Newfoundland and Labrador is home to "perhaps the greatest regional diversity [of spoken English] to be found in North America".
Slide 6
But like dialects the world over, the province's distinct regional modes of expression are in decline, as mass media and easier travel have removed the isolation that caused them to develop in the first place. As native Newfoundlanders, we are proud of our rich dialect heritage, and we wanted to celebrate it with a large-scale musical tribute while it's still a living reality. At the time this project was born, we were both living in Montreal, and we were impressed by the pride and vigor with which Quebecers protect their own speech patterns, which are rooted in their sense of cultural identity. René Lussier's speech-based album made an impression on us, and we wondered if we could adapt his methods to make a comparable tribute to Newfoundland dialects.
Slide 7
The result is One Foot in the Past, a 65-minute album that weaves together Newfoundland speech and traditional folk music with contemporary compositional techniques based on speech, timbre and texture. The album, eight years in the making, also draws on academic research in music perception, dialectology, timbre phonetics and several types of sound analysis and synthesis. It will be published on Centrediscs later this year. We're still finalizing the album design, but the painting you see here, by my brother Matthew Noble, will be part of the album cover.
Slide 8
The tracks on the album fall into several categories. For the purposes of this lecture, I will concentrate on recitations and tracks with instruments and recorded speech, as these categories present the best opportunity to discuss dialects in their genealogical, phonetic and acoustic aspects, as well as the process of transcription into musical notation. I'll return to electronic aspects and other musical considerations in the second installment.
Slide 9
Our first step was to learn more about Newfoundland dialects. We had an intuitive sense of the colors and varieties of the language, but little technical knowledge of how, why and where they were different. To do this, we consulted linguists and folklorists at Memorial University of Newfoundland, as well as sources such as the Dialect Atlas of Newfoundland and Labrador, the Dictionary of Newfoundland English, and an academic book on Newfoundland and Labrador dialects by linguist Sandra Clarke. We learned that much of the province's dialect variety is linked to its settlement history. Clarke lists three main categories:
Standard Newfoundland and Labrador English (S-NLE), characterized as "the standard, Canadian-style speech of many of the province's young, urban middle-class residents" (p.1)
Newfoundland and Labrador Irish English (I-NLE)
British Newfoundland and Labrador English (B-NLE)
In addition, there are smaller pockets of French and Scottish influence. The communities where these dialects are most likely to be found are well documented in the Dialect Atlas.
Slide 10
Armed with this new knowledge, we set out to collect recordings of Newfoundland dialects as a source of musical creation. We recruited sixteen willing participants for our interviews through local media coverage and personal contacts. These interviews took place in people's homes or local establishments and were captured with a portable recorder. Our aim was to make the tone of these interviews as comfortable, friendly and informal as possible, to encourage people to speak naturally and freely. We obtained over 18 hours of recordings.
Slide 11
Our initial plan was to conduct interviews in French with Quebec residents, in parallel with our tour of Newfoundland. Unfortunately, this plan was cancelled by the COVID-19 pandemic. We would have had to collect interviews right at the time that the lockdown measures made it impossible to visit people in their homes as we had done in Newfoundland. We considered conducting interviews on Zoom, but the combination of the inauthentic interview setting and the difficulty of capturing quality sound prompted us to look for other options. Fortunately, we found an open-source public database of speech samples from French-speaking regions around the world, including Quebec: Projet floral, Phonologie de Français Contemporain. The interviews in this database cover a wide range of topics. We found extracts from speakers in Laval and Trois-Rivières that presented an interesting dialectal variety and addressed some of the themes we were interested in.
Slide 12
Another online archive also proved very useful: the MacEdward Leach Folk Song Archive, which contains recordings of hundreds of songs collected in Newfoundland in the 1950s, along with transcriptions of their lyrics. We chose five of these songs to use as recitations. This gave us recordings of the same words in different dialects, enabling us to compare them directly. We also used them to make mash-up recordings switching from one voice to another and from one dialect to another, giving listeners the opportunity to appreciate the range of speech patterns represented in Newfoundland dialects.
Slide 13
Here's an example of a recitation from a folk song called The Shabby Genteel.
Slide 14
The recitations also led to free conversation about the themes addressed in the songs. In the case of The Shabby Genteel, we asked people if they agreed with the statement that "a man can be happy in rags", which led to conversations on topics such as economic and social standing, making the most of what you've got, community identity and support, and technology and change. These conversations took many different directions and touched on a wide variety of topics, and it took us several months to analyze them and organize the clips by theme. The next step was to take the clips that addressed a common theme, select the ones that were compelling enough to include in the compositions, and order them to create narrative and musical arcs. It almost felt like journalism or a documentary: by choosing which passages to include and exclude, and deciding the order in which they would be presented to make the point I was trying to make, I felt in the end like I was the one telling the story, even if it was through other people's words.
To give you an idea of the end result of this process, I'm going to show you a complete piece. We'll come back later to the process of "musicalizing" the speech. For now, here's the score for No Barriers, about two and a half minutes long, about the role of the French language, and more generally of language, dialect and music, in the formation of identity.
Slide 15
Our next line of analysis was phonetic. We focused on understanding the differences in vowel colours between dialects, which is to say, differences in timbre. The use of timbre as a compositional parameter has great potential for speech-based music, opening the door to compelling relationships between speech and music that go beyond what is possible with melodic or rhythmic transcription alone; and you may have noticed relationships between guitar timbres and spoken vowels in the example you've just heard. While there are many precedents for melodic transcription in speech-based music, there are relatively fewer for timbral transcription, especially in a solo context. But as I hope to have shown in the pieces on our album, timbral variation on solo guitar can correspond impressively to spoken vowels. These analogies are shown scientifically in research by Caroline Traube, a professor at the Université de Montréal, which we'll be looking at soon.
Slide 16
But before examining the relationship between guitar sounds and vowel colors, let's take a brief look at how vowel colors differ from one another, which is strongly determined by formants. Formants are regions of concentrated energy in the spectrum of a sound, which play an important role in timbre perception. They influence the way we perceive differences between sound sources, for example between different instruments, or between an opera soloist and the rest of the orchestra. They also play an essential role in distinguishing between vowels. This diagram shows representations of the spectra of several vowels: formants appear as characteristic bumps. As you can see, the formants of the vowel æ are overall higher and more widely spaced than the formants of the vowel u. This is a determining factor in the color difference between these two vowels in our perception, especially in the way we perceive æ as being "brighter" than u.
Slide 17
This kind of difference in vowel colour or brightness is important for understanding how dialects differ from one another. If two speakers of different dialects read the same written words, their vowel realizations can differ considerably. For example, the vowel in the second syllable of the English word "consent" will be placed differently for speakers of Standard Newfoundland English, Irish-influenced Newfoundland English and British-influenced Newfoundland English. For Standard Newfoundland English, it is placed as an open-mid vowel. For Irish-influenced dialects, it's a close-mid vowel. And for British-influenced dialects, it's an open vowel. As this example shows, the vowels in the written word do not always correspond to the way they are realized in speech. Analyzing the differences between dialects therefore requires careful listening.
Slide 18
Knowledge of how vowel realizations differ from dialect to dialect in their formant structures opens up new avenues for understanding the relationships between speech sounds and the sounds of musical instruments. For example, the guitar, like all stringed instruments, can alter the timbres it produces by changing the position at which the string is excited, and since guitar timbres and vowels have characteristic energy distributions in similar regions of the sound spectrum, effective analogies can theoretically be perceived between them. The spectrographs in the bottom row show three different sound energy distributions resulting from the plucking of the guitar string in three different positions: near the bridge in a sul ponticello position, above the sound hole in an ordinario position, and above the fingerboard in a sul tasto position. The spectrographs in the top row show three different distributions of sound energy resulting from different vowels, moving from a front position to a back position. In both cases, the result is a progressive decrease in energy high up in the spectrum.
Slide 19
More precise discrimination of plucking positions can go beyond the three main categories of sul ponticello, ordinario, and sul tasto, and open up a substantial palette of colors on the guitar, which can theoretically be perceived by analogy with the vowels of speech. This was the subject of Caroline Traube's doctoral thesis. You can see on this diagram the plucking positions that correspond approximately with vowels.
Slide 20
In theory, timbre variation based on plucking position provides a way for guitar tones to map onto the vowel colors of different dialects. For example, the three contrasting vowels in the second vowel of the word "consent" we heard earlier correspond to three different plucking positions on the guitar. Therefore, if a composer wishes to capture not only the melodies and rhythms of dialects, but also something of their timbres, varying plucking positions by analogy with vowels may be one way of achieving this.
Slide 21
Phonetic analogies are known to be important in guitar pedagogy and performance practice, with many guitarists and guitar teachers using spoken vowels as models for different tone colors. In this project, we also used them for composition, noting the changes in tone color in the scores as a further means of imitating vocal patterns in classical guitar music. It wasn't always possible to establish a one-to-one correspondence between speech vowels and guitar timbres, so compromises were often necessary. We had to learn which vowels were important in the dialects we were trying to imitate, and which guitar timbres would provide convincing, if often approximate, analogies with those dialects. We believe that timbre modifications in analogy with speech vowels enhance the vocal quality of guitar melodies. For example, here are two examples, one played all ordinario and the other with frequent changes of plucking positions in analogy with vowels. Here also is the original recording, of which both are transcriptions:
Slide 22
The idea of notating the pitches of speech may seem simple, but we soon discovered that it's more complicated than we'd anticipated. On the one hand, musical notes are characterized by discrete, stable pitches: for example, a quarter note G maintains the pitch G for the full duration of the quarter note. This is rarely the case for speech pitches: speech pitches can fluctuate freely within and between syllables. Nevertheless, it would also be too simplistic to characterize our perception of speech as a continuous glissando: we often hear syllables as having defined pitches similar to musical notes, and we can hear spoken phrases as musical melodies. So it seems that in speech we find certain properties of continuous glissandi and certain properties of discrete musical notes.
Sound analysis software such as Sonic Visualizer can facilitate transcription by providing a precise analysis of the fundamental frequency of the speech clip, as in the example on this slide. Here's the speech clip analyzed in this image:
While useful, this type of computer analysis doesn't answer all of the relevant questions. Should we focus on the pitch at the beginning of a syllable, or the average pitch over time of the syllable as a whole? Which interpretation produces the most satisfying musical result? Which interpretation will be easiest for the performer to play? In transcribing the melodies for this album, we constantly rebalanced the equation to adapt to different contexts of speech, music and perception.
To demonstrate the multiplicity of possibilities, here are three different transcriptions of the speech excerpt shown above, with pitches giving priority to:
the appearance of syllables,
to the average pitch of syllables over time, and
to an implied tonality.
Note that none of these transcriptions is "correct": they are all approximations, for the simple reason that they have normalized the fluctuating pitch of speech to the static pitch representation of standard musical notation. The choice of one transcription or another depends on musical contexts and values. Note also that all of these transcriptions are in equal temperament: other solutions involving different tuning systems would multiply the possibilities still further.
Slide 23
Noting the rhythms of speech is perhaps even more complicated than noting its pitches. In preparing our rhythmic transcriptions, we felt the need to balance four competing values: precision of durations, implicit metres, performability, and legibility. These values are often at odds with each other. For example, seeking to produce transcriptions with the highest possible degree of accuracy in the notated durations (as a computer program might) can easily result in a notation so complex as to be impossible to read or perform. Conversely, oversimplifying the rhythm for the sake of intelligibility can produce a notation so imprecise that it deprives the music of the speech-like quality that is the ultimate goal of transcription.
Deciding which metres to use posed some interesting problems. In traditional music notation, metres may often be thought of as prescribing patterns of strong and weak beats. But from the listener's point of view, it's not the metre that prescribes the accent pattern, it's the accent pattern that creates the perception of meter. For example, it's not because the music is in 3/4 time that we experience every third beat as strong; rather, it's because we experience every third beat as strong that the music is perceived as being in 3/4 time. In a simple example like this, it may seem semantic to ask which way this relationship goes, but the importance of the question becomes clearer when it comes to noting complex rhythms such as those found in speech. The transcriber-composer must not simply ask themselves in what metre the rhythm can be notated: it can be notated in many different metres, perhaps in any metre. For example, here's the "same" rhythm notated in (a) 4/4, (b) 3/4, (c) 6/8, and (d) 7/16.
Although these rhythms are identical in terms of their duration patterns, it may be objected that they are not "the same" at all, because the different accent patterns mean that they will not be experienced in the same way. This is correct, and it underscores the risk of discrepancies between musical notation and musical experience when it comes to metres in speech rhythms. If the notated metres faithfully reflect the accentuat patterns of the rhythms, they will help the performers and accurately reflect the experience of the listeners. If the notated metres do not accurately reflect the felt accent patterns of the speech rhythms, they will hinder performers and contradict listeners' experiences. Here's the solution we've arrived at, as it appears in the score, which includes changes of metre and tempo. And here also is the original clip.
Slide 24
These technical considerations on how to capture the timbre, pitches and rhythms of speech gave us the tools to make music from the speech we captured during our interviews. But as I said at the beginning, the music was already there. One of the main aims of this project is simply to draw attention to it. What we're doing is a bit like the painter Mary Pratt, who followed the opposite trajectory to me in beginning her life in New Brunswick and later settling in Newfoundland. Pratt chose everyday objects as her subjects: jam jars, a piece of aluminum foil, things you can look at in your own home without even noticing how beautiful they are. But if you place them in a painting, they're suddenly held up to be appreciated differently. In the same way, by bringing the speech patterns, stories, and ideas of our province's inhabitants to the concert stage, we invite people to listen differently to the music of dialect that surrounds them constantly. I'll conclude by playing another complete piece, It's Our Own, which lasts about three and a half minutes.
Slide 1
Thank you again for being here, and for the opportunity to present my work. Also, a big general thank you to the community of Université de Moncton for making me feel so welcome and at home here in this first couple of months.
Slide 2
In the first installment of this lecture, I said that human speech surrounds us with music almost constantly, and that all we have to do is open our ears and our minds to be suddenly surrounded by fascinating melodies, which are nearly omnipresent but which often go unrecognized as musical in our daily lives. That was the first layer of the music of dialect that I wished to discuss, but it is not the only one. My role as a composer of this kind of music does not end with drawing attention to the musical qualities of speech. As I described last time, the pitches and phonetic timbres of speech can be transcribed for musical instruments such as the guitar. But this is just an early stage in creating speech-based music, as learning the chord progressions of tonal harmony is an early stage in creating tonal music. In both cases, the next question is: what do you do with this foundation? The same harmonic progression can be used to create many different musical textures, and similarly, the same speech melody can be used as the basis for a wide variety of musical realizations. The most obvious of these is a note-for-syllable transcription, in which the instrument attempts to imitate the speech, the whole speech, and nothing but the speech.
Slide 3
But many other approaches may be of musical interest. In the many cases where speech melodies may be interpreted to imply tonalities, they may be harmonized by the addition of accompaniment notes
Slide 4
A kind of subtractive harmonization is also possible, in which instrumental notes are paired with only selected syllables, allowing them to resonate as a harmonic shadow of the speech.
Slide 5
The composer can choose to play with the negative space created by pauses in the speech, inserting notes between words rather than coincident with them to create a rhythmic counterpoint between speech melodies and instrumental melodies.
Slide 6
The motives and patterns of a speech melody can be extracted and developed in instrumental interludes between speech phrases.
Slide 7
And the possibilities continue to multiply with octave displacements, glissandi,
composite speech melodies using multiple excerpts or voices, and any number of other melodic approaches limited only by the imagination of the composer.
Slide 8
But even this range of possibilities represents only a fraction of the available options: all the possibilities mentioned so far are predicated on strict adherence to the melodic pitches of speech, perhaps enhanced with timbres imitating vowel colours. But there are a great many other ways that the natural music of dialect can interact with the intentional music of composition. Speech-based music can dispense with melodic pitch as a parameter entirely, using rhythm alone as the bridge between speech and music and entering whole other soundworlds into the composer’s palette. It can also delve more deeply into the sound of the voice itself, using spectral analysis to derive instrumental harmonies from what we normally perceive as the voice’s timbre. By stepping back from direct imitation and exploring other kinds of relations between the domains, speech-based music can produce fascinating hybrids that contain some qualities of speech and others of music, or create soundscapes that envelop speech like a soundtrack or a documentary, or dissect and recombine components of speech sounds and instrumental sounds in imaginary electroacoustic soundworlds. In this second installment I will begin to address some of these numerous possibilities by discussing creative pathways that I have found useful in speech-based musical composition.
Slide 9
I’d like to clarify that with a relatively small sample size of 16 interviews and a background in composition rather than sociolinguistics or dialectology, I do not consider myself qualified to draw general conclusions about differences between dialects, rhythmic or otherwise. I can tell you that studies by researchers in psychology and neuroscience such as Patel, Iversen and Rosenberg (2006) have found that some aspects of a culture’s speech can be reflected in its music. For example, comparing instrumental music from English speaking and French speaking cultures reveals some systematic differences in their rhythmic and intervallic patterns that are thought to originate in the languages. But the methods required to draw such conclusions are outside the scope of this project. I cannot infer scientifically from the speech samples we collected, for instance, which rhythmic figures are characteristic of particular dialects, or which dialects have overall faster tempos, since there is such a huge amount of variability between speech samples within a single dialect. My goal was rather to explore the artistic potential within the particular acts of speech we were able to collect.
Many composers have been similarly excited about the musical potential of speech and have given us great examples of how this potential can be exploited. Some fun and exciting examples of speech-based music have dispensed with pitch entirely and focused only on rhythm, such as this example by Irish drummer David Dockery:
I love this example not just because it is so impressive and funny, but also because it shows how speech-based music is not limited to contemporary music but has also entered popular culture. One thing that Dockery did not do, so far as I know, is create a score for this transcription: in the YouTube comments below this video, he said that he listened to the speech repeatedly and figuring out the drum pattern bit by bit. That works fine if the transcriber and the performer are the same person, but a score is necessary if you want to write a piece for somebody else to play, which was the case in my work with guitarist Steve Cowan. The next few examples I’ll discuss are from one of the speech-based pieces I wrote for him, we never told nobody, whose theme is an important one in Newfoundland folklore: ghost and faerie stories. We were amazed and delighted with the supernatural stories people shared with us. Given the fantastical subject matter, we delved into a rich and strange soundworld for this piece, using a variety of unconventional instrumental and electroacoustic techniques. In some cases we dropped pitch out of the equation and used the guitar as a percussion instrument following speech rhythms. We immediately ran into two notational challenges: how to notate speech rhythms, and how to notate guitar percussion.
Slide 10
First, the question of notating speech rhythms. Imagine that you had to notate the rhythm of this excerpt in your dictation class. What would this rhythm look like in musical notation? Listen again. In transcribing complex rhythms such as this, I realized that the same rhythm can be written down in many different ways. We have several different devices at our disposal including metre changes, tempo changes, and subdivisions, and different combinations of these things can be used to represent the same rhythm. How you choose to do it depends on which musical values you are trying to prioritize. For example, if maintaining a single metre is the priority, you might choose to write the whole thing down in 4/4, with the tempo staying constant or a close to it as possible, in which case it might look something like this. Listen again and see if this notation coheres with your perception.
From a performer’s point of view, although the metre is simple and constant in this notation, the rhythms themselves are fairly complex. More problematically, the longer the excerpt goes on, the less relation the notated metre bears to the accent patterns we actually hear in the speech, requiring significant and counter-intuitive syncopations which I’ve notated here with accents. To avoid these shortcomings, we might prefer to prioritize simpler rhythms that are easier to read and changing metres and tempos that reflect the accents and pacing of the speech, in which case it might look like this. Here is the speech clip again.
In some ways this is easier to read, but it can be very difficult for performers to change metres and tempos so often without getting lost. In the end, we settled on a compromise with this realization. By opting for longer but asymmetrical metres—the 8/8 measures with beats of different lengths—we maintained something closer to the pulse as felt in the speech without as many metre and tempo changes, and without requiring inordinately complex rhythmic notation. I don’t know if other, better solutions may be possible, but this one worked well enough for the performer to synchronize with:
Slide 11
Slide 12
The next challenge was notating percussive effects on the guitar. It is quite common to use an ‘x’ notehead to indicate striking the body of the instrument, but without further specifications, this leaves the resulting timbre wide open, and I wanted greater control of the soundworld we were creating. The timbre of percussive effects on the guitar can vary quite a lot depending on what part of the instrument is struck, and what it is struck with. Guitarists are often enthusiastic to show off this variety, and yet I know of no standardized notation system that accommodates it. My solution was to develop a two-staff system for the guitar, with one standard five-line treble staff for the strings, and another staff with each line representing a part of the body of the guitar. The number of lines in this second staff is variable depending on the needs of the composition: in this case there are four, representing the neck, upper top of body, lower top of body, and side of body. In addition, the parts of the hand used to strike the instrument are specified with icons: fingernails, fingertips, flat fingers, thumb, knuckles, and palm. Using the staff to indicate the parts of the guitar and icons to indicate the parts of the hand opens up a rich palette of percussion sounds and an intuitive way to notate them. Each combination will result in a slightly different timbre.
Slide 13
A number of other timbral techniques were useful in this piece. In tambora, the strings are struck with a flat part of the hand. In some contexts the strings are allowed to resonate, but here they are deadened. Other muted-string effects are used as well. Elsewhere, the nail is scraped laterally along the string, producing a scratchy, granular sound, and the strings can be strummed behind the sillet for high indeterminate pitches.
Slide 14
One guitar technique that is particularly close to my heart is the natural harmonic, in which the string is not depressed at a fret but is touched lightly at a harmonic node, producing a thinner, higher-pitched, and arguably purer tone. By retuning the open strings, you retune the entire series of natural harmonics, producing a whole different range of possible combinations. But it is also possible to call for specific retunings, and precisely calculate the harmonic fields that result. In we never told nobody, this is achieved by embedding a tone in the electronics track and asking the performer to retune one of the strings to it in mid-performance. The precise order of the harmonics is sometimes specified, and sometimes left to the free choice of the performer. If left free, the effect is something like windchimes: the overall harmonic effect is always basically the same, but the timing and order in which the notes are produced is always slightly different. I notate this using a flowchart-like notation that shows the performer sets of harmonics and possible pathways between them. The performer moves freely through the networks until the next text cue prompts them to move to the next section, producing a mysterious soundscape around the stories being told with an element of unpredictability built in. The music here functions like a soundtrack, to some extent blurring the lines between musical composition and documentary. We’re going to listen to a longer excerpt this time, just under four and a half minutes.
Slide 15
Slide 16
That example featured some electronic sounds in addition to the guitar harmonics, so that makes a nice segue to talk about some of the electroacoustic procedures I have used in this project. There were many opportunities to explore digital manipulations in a piece about the supernatural. Two of the techniques, familiar in the electroacoustic world but not necessarily outside of it, are time stretch and spectral freeze.
Slide 17
Most people will be familiar with the fact that if you play a record or a tape at a slower than normal speed the sound slows down and drops in pitch, and if you play it at a faster speed the sound speeds up and rises in pitch. Time stretch is a digital procedure that achieves something similar, but with a much greater range of manipulation and more control of speed and pitch as separate parameters. For example, you can slow down the speed of a sound but keep its pitch constant. The reverse is also true through the electroacoustic technique of transposition, which is related to but not identical with tonal transposition. You can keep the speed of a sound constant while varying its pitch. In both the time and the pitch domain, you can go quite far with these digital techniques. For example, here is an unmodified recording sample. And now here is just one word from that sample, the word “here”, time stretched to one hundred times its original duration. You might hear that some digital artefacts or distortions are introduced into the sound by this extreme time stretch. In some cases that would be undesirable, but in a deliberately supernatural soundworld, I found they added to the disorienting effect and was happy to keep them. Now here is the same time-stretched sound transposed down two octaves. At this point, to my ears, the connection to the human voice is no longer audible, but the resulting sound has direct musical value, and I take some delight in knowing that an eerie and mysterious soundworld that some listeners may find entirely unfamiliar originates in the most familiar sound source of all. You will hear this same sound, along with several other electroacoustic procedures, in this except from we never told nobody that withdraws from the normal world of speech and rhythm into a suspended, ghostly world.
Slide 18
The technique of spectral freeze also suspends time, but in a different way. With time stretch, even if the speed of a sound sample is greatly slowed down, it continues to progress linearly. With spectral freeze, a very narrow window of the sound’s spectrum is used to create a micro-loop, with the goal being that you cannot hear the loop itself but instead hear the indefinite continuation of the sound, frozen at that moment in its evolution. It is like hitting pause on a sound. For example, here is an unaltered speech recording. And now here is the same example with a spectral freeze added. It can be extra fun to create harmonies with the frozen spectra. In this example, the freeze is heard at three different levels: the original, transposed down a major seventh, and transposed down a perfect fourth.
Slide 19
A more recent innovation in electroacoustic processing is concatenative synthesis, which is conceptually similar to the visual art techniques of pointillism and photo mosaic,
in which composite images are made from a large number of smaller elements. To give you a sense of how this works, here is an excerpt from a famous contemporary piece by Austrian composer Peter Ablinger called Deus Cantando:
What you just heard was produced on a normal piano: we could do it on any piano. What is special about this set-up is not the piano itself but the mechanism playing it, a custom-designed apparatus of 88 robotic fingers. The notes played by this mechanism are modelled after a digital analysis of a human voice, mapped onto piano notes and played with superhuman speed and accuracy.
Slide 20
We attempted to do something similar with concatenative synthesis, not involving robots unfortunately, but with digital manipulations of pre-recorded sounds intended to create a ghostly aura around recordings of the voices. We recorded each fretted note of the guitar one at a time: these notes, taken all together, make up our sound library. We then instructed the computer to chops this sound library up into thousands of tiny slices of sound called grains, and reorder them to best match a target sound, one of our speech samples. The result is a hybrid guitar-voice. Here is the original sound alone and now its concatenative synthesis alone, and now the two together.
Slide 21
The last subject I’ll address today is the more hidden parameter of the harmony of speech. We do not usually think of speech sounds in terms of harmony, since harmony implies more than one sound simultaneously, and we usually think of speech as monophonic, or one sound at a time. But the idea that speech is monophonic, like many of our auditory experiences, is a marvelous illusion. In every word or phoneme that you hear as “one” sound, there are in fact many sound components that can be analyzed harmonically. Our auditory system performs the everyday miracle of fusing them together into timbre, but digital analysis can help us decompose this perceptual illusion and appreciate the underlying harmony.
I mentioned in the first installment that the changing distribution of energy in a voice’s sound spectrum determines the phonemes we hear. The voice might contain a great many frequencies at a time, but it turns out that you can simplify it greatly while still preserving its vocal character and intelligibility. Researcher Chris Darwin produced what he called “sine wave speech” using only sine waves, which are the simplest possible sounds consisting of only a single regular vibration. When sine waves appear as component parts of a complex sound, they are referred to as partials. Darwin was able to produce intelligible speech using only three partials at a time. Here is an example. Most people do not perceive this as speech when it is presented directly like that, but here is the original speech sample. Now that you have been prompted, see if you can hear the voice in the sine tone speech this time.
Slide 22
We are very fortunate to live in an age when the technological tools to explore and create with this kind of phenomenon are widely available. A software I have found very helpful is SPEAR, which stands for Sinusoidal Partial Editing, Analysis, and Resynthesis. SPEAR analyzes sound signals in a particular way: all components of the sound, including noise elements, are interpreted as sine waves of different amplitudes. This is not a perfect representation of the acoustical reality, so an imported sound in SPEAR plays back with a slightly artificial quality.
Slide 23
The closing section of my piece Take Me Back, for electric guitar and electronics, is based on the speech clip we have been working with. The piece features speech samples reflecting on technology and change, and the final section sinks into a meditative nostalgia with this clip coming back as a refrain. I transcribed each of the four main chords that show up in the SPEAR analysis we just saw: you see the chords in the diamond-shaped network in the bottom left of the screen. Before this, selected pitches from the G dominant seventh chord are played, bent freely to imitate the way partials in speech are constantly fluctuating. The soundtrack contains several different sound sources, some of which are natural sounds recorded on site in Newfoundland, and some of which are partials extracted from SPEAR. Here are the final three minutes of the piece.
Slide 24
If my goal in the first instalment of this lecture was to underscore the musicality that is inherent in speech patterns we hear every day, my goal in this second instalment has been to underscore the immense potential for speech to serve as source material for musical composition. I showed you some of the ways I have tried to tap into this potential, including treating speech melodies in various ways, mirroring speech with nonpitched percussive sounds, creating soundscapes around speech with natural harmonics, using electroacoustic effects such as time stretch, spectral freeze, transposition, and concatenative synthesis to produce imaginary soundscapes out of vocal recordings, and analyzing the spectral structure of the voice to generate musical harmony. These techniques are never ends in themselves, but are always means to musical ends. Their value is only ever measured by the musical effects they are able to produce, and the stories they are able to tell. It has given me great pride to draw on the dialects and stories of my fellow Newfoundlanders as sources of musical creation, and great joy to share this project with you in the two instalments of this lecture. Thank you.