The Drummers Clock is a water clock and probably one of the first drum machines and musical robot ancestor. It features five mechanical drummers: two cymbal players, two drummers with a drum slung over their shoulders, and a drummer sitting in front of two kettle drums. Despite significant advances in robotics and AI- Artificial Intelligence, musical robots fall short compared to human musicians; their music lacks subtleties and is “mechanical.” The simplicity of the drum machine, in contrast to a robot violinist, helps to focus on the real issue. This post moves between explanations of the drummers’ clock and thoughts on the difficulties in creating a “musical” musical robot.
How does it work?
This is a simple version of the Castle clock with fewer mechanisms to display the time, and those that remained are simpler. The large components: The water tank, float, and flow controller are identical to the Castle clock and the “Time cart” is very similar, a little like a cheap version of a mobile phone. Al-Jazari does not explain them again but refers the reader to the first chapter (the Castle clock). I also turn directly to musicians. Al-Jazari writes:
“When an hour has passed the musicians (نوبة – nūbah, a musical genre found in the North African, it has its origins in Arabo-Andalusian music.) perform with a clamorous sound which is heard from afar.”
The technical explanation, as always, will be colored in blue, so anyone who is not interested in tipping buckets or scoops wheels can skip those bits. The diagram below is the original drawing of al-Jazari with my captions.
The water flows on the scoop wheel once an hour. This is a large clock, and every time about eight liters of water flows. It is turning the scoop wheel on its axle so that the pegs move the rod which is connected by a copper chain to the slave-girls’ hands. The pegs are an early version of a camshaft and convert circular motion to linear movement. The number of pegs and the intervals between them create different patterns of drumming. The copper strip goes through the hollow wooden body of the musician, and when pulled it goes up and later falls to hit the drum. The pegs are organized in a way that is characteristic of the work of al-Jazari, two adjacent pegs, and a third peg apart. The result is two relatively fast beats, and a third after a pause. There are also two trumpet players, but they are only “decor accessories.” The sound of trumpets is produced separately by the water pouring into an air vessel and compressing the air out through the pipe with a whistle. Al-Jazari used this in many devices, including the Perpetual Flute.
Robotics and the student’s disappointment
Robotics is an enjoyable and sometimes exciting way to teach and learn science and technology. This is true for excellent students and students with difficulties in mathematics and science. Most of the students gladly take upon themselves robotics problems, research a topic, and build a robot using original thinking and their ideas. I taught robotics in different settings: in elementary school Gavrieli, in middle school Branco Weiss and at the Davidson Institute of science education. I found it a creative learning experience in all the years I taught. Beyond the programming tools, mechanical engineering, electronics, and sensors, it teaches children to confront and overcome obstacles, builds confidence and self-esteem, and gives inspiration to science and technology. As a part of the introduction, I would present a wide range of robots, including a robot that plays the violin:
In almost every class I taught, students (happy and enthusiastic) were complaining that the robot’s performance is “mechanical” or “robotic” as a weakness. Violin has a wealth of nuances in how the violinist produces the violin sound (Timbre). This results from many choices, such as which string to use, the pressure of the bow, the point of contact, the bowing speed, to use the whole bow or only partially. All these choices reflect the musical understanding of the violinist and will echo emotionally with the listener. The Musical robot in the movie is programmed so that it “knows” to play the notes, but it has no musical understanding at all. The concept of an artistic interpretation is foreign to him. The drum machine is much simpler in comparison to the violin and will facilitate the discussion.
Musical robots and music
The drummer’s choice includes “merely” the question of the drums selection and the beating template. In terms of the drum machine of al-Jazari, this is the arrangement of the pegs for each drum and possibly changing the length to affect the volume. Al-Jazari made these decisions during “programming” or the design phase, but we can easily think of a modern drumming robot with all the parameters free to change in real-time. This will allow changing hand techniques and evolving drum beat patterns but will not progress us even one step toward musical interpretation.
This is a nontrivial challenge for robot builders. Robots in science fiction literature and toward the end of the 20th century are machines that can replicate human action, especially when it is repetitive. When Karel Čapek coined the word “robot” in the play R.U.R (Rossum’s Universal Robots), the idea was to replace humans when the work was tedious, difficult, or even dangerous. Prominent examples are the welding robots in the automotive industry or the Police Bomb Disposal Robot. In recent years, there has been a shift in direction, and a lot of research has been done on Artificial Intelligence (AI). There are exceptional results in various fields, including robotics stock traders, diagnostic medical robots, and precise surgery robots. What was considered thirty years ago insurmountable, like software playing chess or go (a Japanese board game) is a reality. Chess software like Komodo can beat any human grandmaster. The contribution of AI to music (AIM- Artificial Intelligence Music) is more modest and limited, at this time, to conferences and academia, and no robot can be matched to a human musician. People are not jamming to concerts to listen to musical robots. AIM is a broad field and includes many topics, some of which are relatively simple to understand, Like:
- Methods to produce music using musical robots
- Storage of digital music
Some are more complex, but still approachable:
- Symbolic representations of music – how to represent music beyond the note, including “human touch” and interaction between musicians.
- Human-computer interaction-music – how can a computer respond to music, including jazz improvisation.
Some are borderline science fiction:
- Cognition Computational music – the idea is to teach the computer what is needed for music playing or composition. Moreover, to treat this as a process and do the same learning that does a composer/performer.
As someone who likes robots and automatons, maybe like al-Jazari at his time, I’m surprised by the pleasure I get from reading about the difficulties of the AIM community and the human abilities which are so hard to imitate. Despite what I wrote, I would like you to see the film below: I don’t know how it was done and what part is human, and what part is AIM, but it is certainly fun to watch and listen to!