Music professionals [music information retrieval, MIR], develop machine learning algorithms to identify patterns and dependencies in compositions. The best practitioners in this area today use streaming platforms to classify music – for example, to recognize plagiarism. However, MIR is a fairly new direction, and it is just beginning to consolidate itself as a full-fledged scientific discipline.
For those looking to get a closer look at the topic, we have prepared a short selection of books recommended by industry experts and members of the thematic IT communities. The list includes the work of the author of the language for music programming and a collection of notes from an engineer at Google.
Accompanying Website: Fundamentals of Music Processing
The book written by Maynard Müller – one of the leading collaborators of the international acoustic laboratory AudioLabs – a scientific complex in Germany, where research of multimedia technologies has been carried out for thirty years.
The author examines the algorithms and methodologies of MIR on the example of specific tasks related to the search for musical information. For example, on the network you can find demo chapterdedicated to Fourier analysis, with the help of which it is determined which particular frequency components are present in a particular musical note and how to resynthesize the sound with their help. At the end of each thematic block, Müller provides links to scientific literature for additional reading.
All work is structured according to the “modular principle”, where each chapter is a separate material. So, the reader can independently choose from what moment to start the study and delve into a specific topic. However, it should be noted that the book is suitable only for those who are already a little familiar with the MIR sphere and have a certain mathematical background. By words readers, the author did not provide a transcript of most of the designations and terms, which is why it can be difficult for beginners to navigate in the material.
Musimathics: The Mathematical Foundations of Music
The author of this book was musician and programmer Gareth Loy, who during his career worked at Apple and the Center for Computer Research in Music and Acoustics at Stanford University – CCRMA, where in 1977 developed Samson Box synthesizer capable of processing acoustic signals on the fly. At that time he supported work with 256 oscillators, 128 filters and was considered one of the first specialized devices of this class. In addition, Gareth has released a framework for music programming. Musimat, the principle of operation of which was described in “Musimathics: The Mathematical Foundations of Music“.
In addition to the topic of musical programming, the publication also touches upon the basic theory of MIR. The author explains in simple language how mathematics and composition of music are so similar, touches upon the issues of composition and psychophysics of sound. In second volume of his book Gareth breaks down the basics of digital audio and sampling.
Despite the fact that the first edition was published back in 2011, the material remains relevant to this day. For those unfamiliar with signal processing and retrieval of musical information, this book is an excellent entry point. Professionals working in these areas can use it as a desktop reference.
Notes on Music Information Retrieval
It – electronic collection of notes engineer Steve Toy from Google, which he did during development seminar on the topic of music information retrieval for CCRMA at Stanford University. The entire texture is divided into eleven chapters devoted to topics such as signal analysis, feature extraction, working with rhythm and tempo, musical timing, etc.
The material is filled with a large number of video and audio examples. Also on the site you can find pieces of Python code that illustrate the solution to a particular MIR problem. For example, the author discusses the issue in some detail. factorization of a non-negative matrix (this type of decomposition is used in recommender systems) and k-means method for clustering (needed in machine learning). At first glance, it may seem that only a professional can understand Steve Toy’s notes, but those who are just beginning to understand the field of searching for musical information may well work with the material.
More book selections in our Hi-Fi World:
Our selections of audio gadgets on Habré:
What affordable turntables look like – five “turntables”
Five expert reviews of audio technology – from gags to floor-standing speakers
What’s in the “Hi-Fi World” – speakers, “turntables” and lossless tracks in the video