Sonification in practice/es

La sonificación con fines educativos, en la práctica, es un proceso de exploración de todas las posibilidades que responde esencialmente a la pregunta: "¿Cómo puedo utilizar el sonido para resaltar o demostrar una o más informaciones o conclusiones derivadas de un movimiento, una medición o un fenómeno que existe, ha ocurrido o se está desarrollando a lo largo del tiempo?". Los datos existentes a nuestra disposición, las condiciones y los métodos de su recopilación, así como el propósito educativo para el que se destina la sonificación, son los factores determinantes para su uso eficaz.

Los aspectos que se describen a continuación configuran la relación entre las necesidades educativas y el concepto de sonido, así como su organización estructurada a lo largo del tiempo, es decir, el concepto de música.

Aspectos de la enseñanza con la sonificación como práctica musical

La indiscutible conexión entre el sonido y los números —en concreto, el concepto de descomposición del sonido en frecuencias o armónicos— proporciona un marco suficientemente estructurado para la enseñanza interdisciplinaria mediante el sonido, dentro del cual se pueden abordar todos los aspectos de STEAM (Ciencia, Tecnología, Ingeniería, Arte y Matemáticas). Dado que el concepto de tiempo define el fenómeno sonoro, el acto representacional de un efecto de sonido no puede sino estar en el centro de cualquier enfoque pedagógico. En consecuencia, la disposición organizada de los elementos sonoros en el tiempo de forma armoniosa —tanto en términos de ritmo, intensidad, timbre, altura y su ubicación posicional en la escala musical, diatónica o no— constituye un resultado musical. Esta organización racional puede servir como campo de experimentación en la composición musical, mientras que la parametrización de todos los conceptos anteriores puede enriquecer cualquier objetivo educativo que dependa de la evolución de un fenómeno a lo largo del tiempo o de la conversión de datos en sonido.

Por lo tanto, podemos distinguir razonablemente el concepto de sonificación con fines educativos en tres enfoques básicos:

• El simbólico

• El matemático

• El adaptativo

Symbolic Sonification

The reproduction of sound characteristics, namely: pitch, intensity, timbre, repetition rate (if any), and duration—which are linked to scientific concepts, terms, and quantities without being logically mapped to a data-set (data-mapping)— constitutes the subject of symbolic sonification.

A simple example would be to “sound-paint” a gray cloud using low-frequency noise and a white cloud using high-frequency noise. Another example would be a class of students representing the sound of rain by randomly tapping their fingernails on their desks. Another example that relates composition with music representation is the leitmotif. A leitmotif is a short melodic theme consisting of a few specific notes which, as a unique motif (pattern), is associated with a character in an opera and played by the orchestra, particularly in Wagner’s operas. A character’s leitmotif brings the character to mind throughout the entire work, whether the character is on stage or not! Translating this to a data series, such a leitmotif could replace the expected sound of a prominent low or high value (or a specific value or even a range of values) without having any coherence or arising from the neighboring data.

Mathematical Sonification

When pitch, intensity, timbre, rhythm (if any), and duration as sound-characteristics, runs through a series of data-measurements connected to a physical term, or a scientific concept, they form a logical map of one or more parts of that series (direct data-mapping). The sounding result of this match is mathematical sonification.

An example that perfectly illustrates the above distinction, primarily by exploiting the characteristic of rhythm, is that of the mechanism for audibly indicating the distance between a car and the one next to it while parking, a feature found in many cars. The repetition frequency of this momentary acoustic signal forms a repeating pattern whose rhythm varies (slow-fast) depending on the proximity data to the obstacle, which is detected with high precision by a sensor.

To understand the difference between symbolic and mathematical representation, we can adapt the previous examples as “unplugged activities” in a classroom. Mathematical sonification in the cloud example would occur if we defined a color threshold for white or gray and represented the droplets that make up the clouds with millions of frequency particles of minimal duration (sound nebulae) the droplets of which clouds are composed. In the example of rain, we would have a mathematical sonification if students represented with absolute precision, one by one, every raindrop at a specific time and surface area. Finally, in the example of “parking,” we would have symbolic representation if the students’ eyes took on the role of the sensor, where data would be estimated visually without absolute mathematical measurement.

Adaptive Sonification

It is a sound design or musical composition (by expanding this notion), resulting from mathematical sonification in which, however, methods of aesthetic sound-rendering are creatively utilized to meet teaching objectives in describing learning concepts.

Furthermore, the analysis of data-mapping methods in conjunction with the diatonic scale opens up a fruitful field for exploring teaching tools that allow sound to be processed in terms of musical composition. The use of MIDI for sound processing or the highlighting of musical motifs—which can serve as a starting point for creating musical compositions—perfectly extends adaptive sonification. In fact, the graphical representation of data (graphical display) can be creatively transformed into sound by treating the display as a two-dimensional scheme, or even a photograph as a three-dimensional image. The result is referred to as “schematic sonification”.

This adaptable approach broadens access to the auditory outcome of data sonification across a wide range of age groups and grade levels, inviting educators from other disciplines—such as Art, Theater, and Music, to actively participate in interdisciplinary teaching. An example of this approach has been implemented in the "Sounds of the Stars" scenario in collaboration with the National Observatory of Athens (community: Ήχοι των Άστρων). The scenario is part of the SoundScapes learning scenarios repository .

THE MIDI protocol. Why is it useful for sonification in school?

MIDI Protocol stands for Musical Instrument Digital Interface and was introduced in early ’80s as machine language allowing analog and later digital instruments interconnection. This language interprets several aspects of music performance and notation in an electronic format.

MIDI enables the user to receive, transmit, store and edit electronically produced signals that correspond to several aspects of music. Main parameters of these aspects include note-on, note-off, velocity, timbre and pitch. All these parameters can be stored as code in timeline fashion within a MIDI file. A MIDI file resembles the “program” in the form of a revolving cylinder or perforated paper used in late 18th c. music boxes or early 20th c. “pianolas”, which are musical automata. It is this characteristic that can be proved enormously useful for educational purposes, as numerous midi applications, sensors and programs are widely spread throughout the internet. However, it is the ability to edit the output as a musical score or as a part of a polyphonic composition that makes MIDI an exceptionally powerful educational tool. Within the present WIKI pages sensors using MIDI in particular, are widely displayed.

Sonification components

A sonification activity consists in the design and building of a sonification system. A sonification system can be accomplished in many different ways but 3 components must always be considered:

1) INPUT DATA;

2) MAPPING PROTOCOL;

3) AUDIO OUTPUT;

Input Data

In a sonification system, which is our final product, the data is the source of the sound engine, and some particular sounds will be the output. The inputs and outputs are mapped onto each other following a protocol that establishes which sounds are played according to which data. So first we need to know and understand the data we want to sonify. We must know what we want to say with our system - what we will talk about. We must know how the data change (usually we have time based data but there can also be spatially referenced data, like maps) and what characteristics of its behavior we want to represent. For example if you have a single value (like luminosity of a star, linear position of a car, amount of likes in a youtube channel, number of new posts on wikipedia, etc) you can choose to play a sound when this value is more than a certain threshold value, or play a sound that gets louder when the values becomes higher, or a sound when the values are raising or decreasing in time. In some cases it is useful to determine the highest and the lowest value within the whole range of values available. In terms of outputs this can help define a “container” of initial values that can define the range of deviations in the output. We can highlight certain features of data. There are many types of data. The most common are: 

Single data: indicating a state ON-OFF (boolean data).

A single data value covering a range of values: usually mapped to a single sound or sound feature like the pitch, or bpm (beats per minute), or an effect, but it can control more than one feature or sound at once.

Multiple data: more than one data of the previous type. Usually there are many types of data collected at the same time so these data sets consist of several layers of synchronized data.

Sound has the advantage over visual perception that more layers of data can be perceived at the same time. Changes in patterns are more easily detected listening than looking at. Especially if the amount of data is very large. So, in sum, we need to consider the data we have, how they evolve in time, how they are arranged and what are the salient parts we want to use to feed our sonification system. We have to ask ourselves “what will the sound mean?” We need to understand that data is not the message! We must metabolize the data and their behavior and find what message will be triggering sound.

And, therefore, before this we need to ask ourselves what is the purpose of the sonification? Will it be applied continuously, maybe in the background, or just after some time of collecting data, or both?

Real-Time Sonification vs “A Posteriori”

Acording to the use of the sonification system (to analyze or to monitor a certain phenomena) we distinguish two “modes”:

Real-time (to monitor) - a stream of data is sonifed instantly and a sound is produced to display the value and behavior of the data in that particular moment;

“A posteriori” (to analyze) - time-series sonification of a set of pre-recorded data is converted into an audio file that displays the values and behavior of the data over the period of time covered by the time-series. 

These two methods are not mutually exclusive and can eventually display the same sounds. The difference is that in an “a posteriori” sonification, because the sound is produced after the events that originated the data, the parameters of the final piece can be adapted, i.e. the total duration. In a real-time case, you can control the time resolution: that is the time interval at which the sound can change and is played.

Mapping Protocol

The mapping protocol is the core of the sonification system. This is where knowledge of input data must be combined with creativity. According to his/her educational needs, the creator of the sonification system makes choices based on his/her character and artistic taste in translating data sets into sound pieces. The mapping protocol is the process or algorithm or function that associates particular sounds to defined data. It is the set of rules by which output sounds correspond to input data. A simple mapping can consist for example in a direct one-to-one correspondence between each value of an input data to a parameter of an output sound, like the pitch. This component of the system is key because here is where the designer of the system selects certain features of the data to be played in a particular manner, in order to highlight them, or not.

So this mapping consists in associating certain data aspects to different auditory parameters, such as pitch, loudness, timbre, and rhythm. For example, the amplitude of a sound can be mapped to the value of a light resistor, or the frequency of a sound can be mapped to the rate of change of the sea level (tides).

Usually the tendency is to map a single feature of the data to a single parameter of output sound but we humans are generally more capable of perceiving differences in sound if such differences manifest concurrently through different properties. So it is not a bad idea to map the same variable onto different psychoacoustic properties of a sound (pitch and volume as an example of the most evident) if we want to emphasize its change and dynamics.

Our sense of hearing is able to focus on a particular sound in between many others (see the “cocktail party effect”) based on timbre. Our auditory system can process information at a far higher rate than our visual system. For example, while video typically updates at 60 frames per second (60 Hz), standard audio is sampled at 44,100 times per second (44.1 kHz). This means that even a single, brief spike in an audio signal—lasting just one sample—is instantly perceived as a distinct "click." As a result, hearing allows us to monitor multiple layers of information simultaneously, often more efficiently than through visual perception alone .

Audio Output

The output sound of the system will be the first characteristic to be perceived by a user. It is its signature, its flavor. It will interact with the user’s taste and we must be aware of that. It is the auditive wrapping to be perceived by an audience and, as studies on sound perception show, it will immediately and unconsciously provoke a good or bad sensation to the listener. We should therefore get used to producing “nice” sound outputs with the device that will be used, be it a microcontroller buzzer or a pc virtual synthesizer or a DAW (Digital Audio Workstation) connected to speakers. We should practice some music, or at least make some noise!

Considering that sound perception is time-based, sonification is by and large focused on rendering continuous data stream over time: this means that the input data of a sonification system could be also come from another domain, like the profile of a territory (geographical data) but all of them will be transferred onto a representation in time which is sound. Sound exists only in time, as variation of pressure detected by our eardrums and transformed into electrical signals in our brain, or broadly in our nervous system. Without getting into the depth of such a fascinating subject we need to clarify a couple of concepts before we move on. Even those who never played or created music know some of the characteristics of sound that we describe here.

Music and Sound: Basic Concepts

Sound is detected by our brain when a variable pressure stimulates our timpani. This is a small membrane that when moved by air pressure (or water if you find yourself under water) generates electrical stimuli that the brain processes as “sound”. If this variable pressure is oscillating regularly at a certain frequency (a certain number of times per second) we hear a tone. That is why tones (or notes) are measured in Hertz (Hz), or cycles per second.

The human hearing is able to sense tones between 20 Hz and 20000 Hz (this range is unique to each one and usually gets smaller with age). The vibrations of pressure with frequencies lower than 20 Hz or higher than 20000 Hz are inaudible. They are called infrasounds and ultrasounds respectively. We do not hear them but we still can sense them, with the touch sense in the case of infrasound and with temperature sense in the case of ultrasound.

The main characteristics of sound are:

Volume or Intensity or loudness: The power of a soundwave (louder more power, softer less power).

Frequency or pitch: The number of times the sound pressure moves back and forth the timpani in our ears. According to music theory some of these frequencies are called notes in the context of tuning systems.

Timbre: Is the spectral characteristic of sound, its sound quality, its fingerprint, a sense of the “color” of the sound. This is what allows us to distinguish between a trumpet or a guitar when they are playing the same note with the same volume. It also allows us to distinguish between our human voices.

There are several other characteristics that define sound but these are the main ones we can use in this context. Other characteristics that could be easily employed in a classroom for sonification are:

Duration: How long each sound lasts.

Rhythm: How frequently the sounds repeat and in what pattern. For example, a metronome is a device that produces short, evenly spaced sounds at a set number of beats per minute (BPM). Other devices use this characteristic as a sonification output (Geiger counter - Wikipedia) and parking devices for cars.

3D Positioning: The position of a sound source in space - for example if the sound comes from the left or right speaker in a stereo system. Far more complex but basically the same concept, are the surround systems 5.1 or 7.1 up to ambisonic systems where the position of the sound source can be even more detailed by using multiple channels (Ambisonic reproduction systems - Wikipedia)

Context is important

When designing the output sounds we need to consider what will be the audience of the designed system. In which settings they will listen to its sounds. It is impossible to be sure about it and to know the taste of our target listeners but it is convenient to think about it.

What is the profile of the listener? Are they young students? What type of sound would they be interested in hearing? but also in what kind of sound-producing interaction could they be engaged in, according to their skills and potentials? Do they perceive changes in less evident sound features, (i.e. timbre)?

The sounds we produce must be considered in the context where they will be played. They should be able to capture the listener's attention and emerge from the background noise, and if possible, not be perceived as noise or annoying. For example, mapping all the values of a single variable to all the values of frequency in a certain range may sound unpleasant compared to mapping it onto a familiar music scale, like the chromatic scale in the western world. Or manipulating the speed of a regular beat instead of playing random time durations can be more effective. It depends on the listener's attitude and taste, of course. Additionally, it is important to consider the sound designers’ own taste. It is convenient to consider who will be the listener, but, on the other side, it is not mandatory to produce mainstream sounds in order to please the supposed “common taste”.

Apart from taste and esthetic considerations, we need to consider factual conditions: in the case of background continuous sound as a product of a sonification to monitor some data stream we should therefore take into account the potential listener fatigue to that type of sound. We can consider the difference of using familiar sounds (for example even recorded samples of voices and sentences of the target listeners) compared to new and special digitally synthesized sounds. Designers of sonification systems should at least be be aware of the variety of different impacts that their sounds can have upon the listener (synthesized sounds could surprise!).

We need to assemble a diverse toolkit of musical techniques and resources. Sonification designers should ensure their palette of sounds is as rich and varied as the data they aim to represent.

Quality of Output

Sonification should not only be comprehensible but also engaging, ideally offering information as effectively as, or even more clearly than visual graph. The quality of the sonification is equally important. This includes both the technical excellence of the audio and its "musical narrative" - how well it describes the evolution of the data while remaining aesthetically pleasing. While "pleasantness" is subjective, an appealing sonification helps maintain the listener’s attention and ensures the data is effectively communicated, as discussed in the Context is Important section.

Sonifications can use either physical (natural) or digital sounds, depending on resources and approach. Physical sounds come from acoustic sources like the human body, percussion, or traditional instruments, performed through notation, gestures, or improvisation. Digital sounds, however, are generated or processed using computers, digital audio workstations (DAWs), or electronic devices. While technical details like compression, sample rate, or bit depth influence digital audio quality, the key point is the impact of the playback system: a high-quality sound system (e.g., computer speakers) will deliver a richer experience than a simple buzzer.

Musical Quality

The designer should consider what type of narrative he/she is inducing in the listener. That means for example using low and scary sounds to represent parameters of global warming (A Song of Our Warming Planet or The sound of climate change from the Amazon to the Arctic).  As we want to stimulate the user to pay attention to our system output it can also be useful to have a survey about what type of music the listener appreciates. A generally and initially acceptable musical sound, with the least possible chance of being rejected by the majority of recipients, would be the one that would obey the fundamental principles of symmetry and proportion, as these have shaped our common perception of “music" in today's world. 

However, Soundscapes project encourages every approach on sonification if it satisfies the creator's inspiration or cultural demands as well as the aesthetical or informative needs of the audience, or the target group it is addressed to.  

In the following pages, practical ways to implement the above approaches with or without handling data sets coming either from measurements or from sensors, are entitled as: Unplugged activities, Real-time sonification and a posteriori sonification.

References