To apply neural sequence models such as the Transformers to music generation tasks, one has to represent a piece of music by a sequence of tokens drawn from a finite set of pre-defined vocabulary. Such a vocabulary usually involves tokens of various types. For example, to describe a musical note, one needs separate tokens to indicate the note's pitch, duration, velocity (dynamics), and placement (onset time) along the time grid. While different types of tokens may possess different properties, existing models usually treat them equally, in the same way as modeling words in natural languages. In this paper, we present a conceptually different approach that explicitly takes into account the type of the tokens, such as note types and metric types. And, we propose a new Transformer decoder architecture that uses different feed-forward heads to model tokens of different types. With an expansion-compression trick, we convert a piece of music to a sequence of compound words by grouping neighboring tokens, greatly reducing the length of the token sequences. We show that the resulting model can be viewed as a learner over dynamic directed hypergraphs. And, we employ it to learn to compose expressive Pop piano music of full-song length (involving up to 10K individual tokens per song), both conditionally and unconditionally. Our experiment shows that, compared to state-of-the-art models, the proposed model converges 5--10 times faster at training (i.e., within a day on a single GPU with 11 GB memory), and with comparable quality in the generated music.
We present a novel framework for generating pop music. Our model is a hierarchical Recurrent Neural Network, where the layers and the structure of the hierarchy encode our prior knowledge about how pop music is composed. In particular, the bottom layers generate the melody, while the higher levels produce the drums and chords. We conduct several human studies that show strong preference of our generated music over that produced by the recent method by Google. We additionally show two applications of our framework: neural dancing and karaoke, as well as neural story singing.
Carnegie Mellon University（卡内基梅隆大学）；National University of Singapore（新加坡国立大学）
Many practices have been presented in music generation recently. While stylistic music generation using deep learning techniques has became the main stream, these models still struggle to generate music with high musicality, different levels of music structure, and controllability. In addition, more application scenarios such as music therapy require imitating more specific musical styles from a few given music examples, rather than capturing the overall genre style of a large data corpus. To address requirements that challenge current deep learning methods, we propose a statistical machine le…
Peking University（北京大学）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）
Melody is one of the most important components in music. Unlike other components in music theory, such as harmony and counterpoint, computable features for melody is urgently in need. These features are highly demanded as data-driven methods dominating the fields such as musical information retrieval and automatic music composition. To boost the performance of deep-learning-related musical tasks, we propose a set of interpretable features on monophonic melody for computational purposes. These features are defined not only in mathematical form, but also with some considerations on composers 'in…
Queen Mary University of London（伦敦玛丽女王大学）；Academia Sinica（中央研究院）；Indian Institute of Technology Kharagpur（印度理工学院卡拉格普尔分校）；Dadabots；Taiwan AI Labs（台湾人工智能实验室）
Originating in the Renaissance and burgeoning in the digital era, tablatures are a commonly used music notation system which provides explicit representations of instrument fingerings rather than pitches. GuitarPro has established itself as a widely used tablature format and software enabling musicians to edit and share songs for musical practice, learning, and composition. In this work, we present DadaGP, a new symbolic music dataset comprising 26,181 song scores in the GuitarPro format covering 739 musical genres, along with an accompanying tokenized format well-suited for generative sequenc…
Structure is one of the most essential aspects of music, and music structure is commonly indicated through repetition. However, the nature of repetition and structure in music is still not well understood, especially in the context of music generation, and much remains to be explored with Music Information Retrieval (MIR) techniques. Analyses of two popular music datasets (Chinese and American) illustrate important music construction principles: (1) structure exists at multiple hierarchical levels, (2) songs use repetition and limited vocabulary so that individual songs do not follow general s…
New York University（纽约大学）；University of Florida（佛罗里达大学）；Carnegie Mellon University（卡内基梅隆大学）
Music arrangement generation is a subtask of automatic music generation, which involves reconstructing and re-conceptualizing a piece with new compositional techniques. Such a generation process inevitably requires reference from the original melody, chord progression, or other structural information. Despite some promising models for arrangement, they lack more refined data to achieve better evaluations and more practical results. In this paper, we propose POP909, a dataset which contains multiple versions of the piano arrangements of 909 popular songs created by professional musicians. The m…
Regional style in Chinese folk songs is a rich treasure that can be used for ethnic music creation and folk culture research. In this paper, we propose MG-VAE, a music generative model based on VAE (Variational Auto-Encoder) that is capable of capturing specific music style and generating novel tunes for Chinese folk songs (Min Ge) in a manipulatable way. Specifically, we disentangle the latent space of VAE into four parts in an adversarial training way to control the information of pitch and rhythm sequence, as well as of music style and content. In detail, two classifiers are used to separat…
University of Hawaii at Manoa（夏威夷大学马诺阿分校）；Technical University of Madrid（马德里理工大学）
In this paper we present an attentional neural network for folk song classification. We introduce the concept of musical motif embedding, and show how using melodic local context we are able to model monophonic folk song motifs using the skipgram version of the word2vec algorithm. We use the motif embeddings to represent folk songs from Germany, China, and Sweden, and classify them using an attentional neural network that is able to discern relevant motifs in a song. The results show how the network obtains state of the art accuracy in a completely unsupervised manner, and how motif embeddings…
Academia Sinica（中央研究院）；National Taiwan University（台湾大学）；KAIST（韩国科学技术院）；Georgia Institute of Technology（佐治亚理工学院）
While there are many music datasets with emotion labels in the literature, they cannot be used for research on symbolic-domain music analysis or generation, as there are usually audio files only. In this paper, we present the EMOPIA (pronounced `yee-mò-pi-uh') dataset, a shared multi-modal (audio and MIDI) database focusing on perceived emotion in pop piano music, to facilitate research on various tasks related to music emotion. The dataset contains 1,087 music clips from 387 songs and clip-level emotion labels annotated by four dedicated annotators. Since the clips are not restricted to one c…
Cover song identification represents a challenging task in the field of Music Information Retrieval (MIR) due to complex musical variations between query tracks and cover versions. Previous works typically utilize hand-crafted features and alignment algorithms for the task. More recently, further breakthroughs are achieved employing neural network approaches. In this paper, we propose a novel Convolutional Neural Network (CNN) architecture based on the characteristics of the cover song task. We first train the network through classification strategies; the network is then used to extract music…
The University of Sheffield（谢菲尔德大学）；Universität Koblenz（科布伦茨大学）
Hit song prediction, one of the emerging fields in music information retrieval (MIR), remains a considerable challenge. Being able to understand what makes a given song a hit is clearly beneficial to the whole music industry. Previous approaches to hit song prediction have focused on using audio features of a record. This study aims to improve the prediction result of the top 10 hits among Billboard Hot 100 songs using more alternative metadata, including song audio features provided by Spotify, song lyrics, and novel metadata-based features (title topic, popularity continuity and genre class)…
THE UNIVERSITY OF TOKYO（东京大学）；IRCAM（法国声学/音乐研究中心）；INNOVATION CREATIVE, INC.（创新创意公司）
Automatic melody generation for pop music has been a long-time aspiration for both AI researchers and musicians. However, learning to generate euphonious melody has turned out to be highly challenging due to a number of factors. Representation of multivariate property of notes has been one of the primary challenges. It is also difficult to remain in the permissible spectrum of musical variety, outside of which would be perceived as a plain random play without auditory pleasantness. Observing the conventional structure of pop music poses further challenges. In this paper, we propose to represen…
Bremen Spatial Cognition Center, University of Bremen（不来梅大学不来梅空间认知中心）；Faculty of Information and Communication Technology, Mahidol University（玛希隆大学信息与通信技术学院）
In the recording studio, producers of Electronic Dance Music (EDM) spend more time creating, shaping, mixing and mastering sounds, than with compositional aspects or arrangement. They tune the sound by close listening and by leveraging audio metering and audio analysis tools, until they successfully creat the desired sound aesthetics. DJs of EDM tend to play sets of songs that meet their sound ideal. We therefore suggest using audio metering and monitoring tools from the recording studio to analyze EDM, instead of relying on conventional low-level audio features. We test our novel set of featu…
Taiwan AI Labs（台湾人工智能实验室）；National Taiwan University（台湾大学）；Academia Sinica（中央研究院）
Transformers and variational autoencoders (VAE) have been extensively employed for symbolic (e.g., MIDI) domain music generation. While the former boast an impressive capability in modeling long sequences, the latter allow users to willingly exert control over different parts (e.g., bars) of the music to be generated. In this paper, we are interested in bringing the two together to construct a single model that exhibits both strengths. The task is split into two steps. First, we equip Transformer decoders with the ability to accept segment-level, time-varying conditions during sequence generat…
Graduate School of Information Science and Technology, The University of Tokyo（东京大学信息科学与技术研究生院）
Thanks to developments in machine learning techniques, it has become possible to synthesize high-quality singing voices of a single singer. An open multispeaker singing-voice corpus would further accelerate the research in singing-voice synthesis. However, conventional singing-voice corpora only consist of the singing voices of a single singer. We designed a Japanese multispeaker singing-voice corpus called "JVS-MuSiC" with the aim to analyze and synthesize a variety of voices. The corpus consists of 100 singers' recordings of the same song, Katatsumuri, which is a Japanese children's song. It…
University of Science and Technology of China（中国科学技术大学）
Most existing neural network models for music generation explore how to generate music bars, then directly splice the music bars into a song. However, these methods do not explore the relationship between the bars, and the connected song as a whole has no musical form structure and sense of musical direction. To address this issue, we propose a Multi-model Multi-task Hierarchical Conditional VAE-GAN (Variational Autoencoder-Generative adversarial networks) networks, named MIDI-Sandwich, which combines musical knowledge, such as musical form, tonic, and melodic motion. The MIDI-Sandwich has two…
National Institute of Informatics（国立信息学研究所）；SOKENDAI（综合研究大学院大学）
Generating melody from lyrics is an interesting yet challenging task in the area of artificial intelligence and music. However, the difficulty of keeping the consistency between input lyrics and generated melody limits the generation quality of previous works. In our proposal, we demonstrate our proposed interpretable lyrics-to-melody generation system which can interact with users to understand the generation process and recreate the desired songs. To improve the reliability of melody generation that matches lyrics, mutual information is exploited to strengthen the consistency between lyrics …
The ICASSP 2026 Automatic Song Aesthetics Evaluation Challenge
Shanghai Conservatory of Music, Shanghai, China（上海音乐学院，中国上海）；College of Computing and Data Science, Nanyang Technological University, Singapore（计算与数据科学学院，南洋理工大学，新加坡）；Nanjing University, Suzhou, China（南京大学，中国苏州）
Music genre classification is a widely researched topic in music information retrieval (MIR). Being able to automatically tag genres will benefit music streaming service providers such as JOOX, Apple Music, and Spotify for their content-based recommendation. However, most studies on music classification have been done on western songs which differ from Thai songs. Lukthung, a distinctive and long-established type of Thai music, is one of the most popular music genres in Thailand and has a specific group of listeners. In this paper, we develop neural networks to classify such Lukthung genre fro…
We explore the feasibility of using triplet neural networks to embed songs based on content-based music similarity. Our network is trained using triplets of songs such that two songs by the same artist are embedded closer to one another than to a third song by a different artist. We compare two models that are trained using different ways of picking this third song: at random vs. based on shared genre labels. Our experiments are conducted using songs from the Free Music Archive and use standard audio features. The initial results show that shallow Siamese networks can be used to embed music fo…
The behavior of users of music streaming services is investigated from the point of view of the temporal dimension of individual songs; specifically, the main object of the analysis is the point in time within a song at which users stop listening and start streaming another song ("skip"). The main contribution of this study is the ascertainment of a correlation between the distribution in time of skipping events and the musical structure of songs. It is also shown that such distribution is not only specific to the individual songs, but also independent of the cohort of users and, under station…
MuDiT & MuSiT: Alignment with Colloquial Expression in Description-to-Song Generation
College of Computer Science and Technology, Zhejiang University（浙江大学计算机科学与技术学院）；Caichong Artificial Intelligence Research, DuiNiuTanQin Co., Ltd.（追牛探琴有限公司彩创人工智能研究院）
University of Vienna（维也纳大学）；University of Bamberg（班贝格大学）
We look into the connection between the musical and lyrical content of metal music by combining automated extraction of high-level audio features and quantitative text analysis on a corpus of 124.288 song lyrics from this genre. Based on this text corpus, a topic model was first constructed using Latent Dirichlet Allocation (LDA). For a subsample of 503 songs, scores for predicting perceived musical hardness/heaviness and darkness/gloominess were extracted using audio feature models. By combining both audio feature and text analysis, we (1) offer a comprehensive overview of the lyrical topics …
Zhejiang University（浙江大学）；Alibaba-Zhejiang University Joint Institute of Frontier Technologies（阿里巴巴-浙江大学前沿技术联合研究院）；Jingchu University of Technology（荆楚理工学院）；Shandong University（山东大学）
Real-time music accompaniment generation has a wide range of applications in the music industry, such as music education and live performances. However, automatic real-time music accompaniment generation is still understudied and often faces a trade-off between logical latency and exposure bias. In this paper, we propose SongDriver, a real-time music accompaniment generation system without logical latency nor exposure bias. Specifically, SongDriver divides one accompaniment generation task into two phases: 1) The arrangement phase, where a Transformer model first arranges chords for input melo…
Queen Mary University of London（伦敦玛丽女王大学）；NYU Shanghai（上海纽约大学）；MBZUAI（穆罕默德·本·扎耶德人工智能大学）
Lyric interpretations can help people understand songs and their lyrics quickly, and can also make it easier to manage, retrieve and discover songs efficiently from the growing mass of music archives. In this paper we propose BART-fusion, a novel model for generating lyric interpretations from lyrics and music audio that combines a large-scale pre-trained language model with an audio encoder. We employ a cross-modal attention module to incorporate the audio representation into the lyrics representation to help the pre-trained language model understand the song from an audio perspective, while …
The ability of deep neural networks to learn complex data relations and representations is established nowadays, but it generally relies on large sets of training data. This work explores a "piece-specific" autoencoding scheme, in which a low-dimensional autoencoder is trained to learn a latent/compressed representation specific to a given song, which can then be used to infer the song structure. Such a model does not rely on supervision nor annotations, which are well-known to be tedious to collect and often ambiguous in Music Structure Analysis. We report that the proposed unsupervised auto-…
Universidade Estadual de Maringá（马林加州立大学）；National Institute of Science and Technology for Complex Systems（国家复杂系统科学技术研究所）
We report a statistical analysis over more than eight thousand songs. Specifically, we investigate the probability distribution of the normalized sound amplitudes. Our findings seems to suggest a universal form of distribution which presents a good agreement with a one-parameter stretched Gaussian. We also argue that this parameter can give information on music complexity, and consequently it goes towards classifying songs as well as music genres. Additionally, we present statistical evidences that correlation aspects of the songs are directly related with the non-Gaussian nature of their soun…
Music structure analysis (MSA) systems aim to segment a song recording into non-overlapping sections with useful labels. Previous MSA systems typically predict abstract labels in a post-processing step and require the full context of the song. By contrast, we recently proposed a supervised framework, called "Music Structural Function Analysis" (MuSFA), that models and predicts meaningful labels like 'verse' and 'chorus' directly from audio, without requiring the full context of a song. However, the performance of this system depends on the amount and quality of training data. In this paper, we…
