Recent advances in generative artificial intelligence (AI) have created models capable of high-quality musical content generation. However, little consideration is given to how to use these models for real-time or cooperative jamming musical applications because of crucial required features: low latency, the ability to communicate planned actions, and the ability to adapt to user input in real-time. To support these needs, we introduce ReaLJam, an interface and protocol for live musical jamming sessions between a human and a Transformer-based AI agent trained with reinforcement learning. We enable real-time interactions using the concept of anticipation, where the agent continually predicts how the performance will unfold and visually conveys its plan to the user. We conduct a user study where experienced musicians jam in real-time with the agent through ReaLJam. Our results demonstrate that ReaLJam enables enjoyable and musically interesting sessions, and we uncover important takeaways for future work.

This paper presents a study of participants interacting with and using GaMaDHaNi, a novel hierarchical generative model for Hindustani vocal contours. To explore possible use cases in human-AI interaction, we conducted a user study with three participants, each engaging with the model through three predefined interaction modes. Although this study was conducted "in the wild"-- with the model unadapted for the shift from the training data to real-world interaction -- we use it as a pilot to better understand the expectations, reactions, and preferences of practicing musicians when engaging with such a model. We note their challenges as (1) the lack of restrictions in model output, and (2) the incoherence of model output. We situate these challenges in the context of Hindustani music and aim to suggest future directions for the model design to address these gaps.

To make music composition more approachable, we designed the first AI-powered Google Doodle, the Bach Doodle, where users can create their own melody and have it harmonized by a machine learning model Coconet (Huang et al., 2017) in the style of Bach. For users to input melodies, we designed a simplified sheet-music based interface. To support an interactive experience at scale, we re-implemented Coconet in TensorFlow.js (Smilkov et al., 2019) to run in the browser and reduced its runtime from 40s to 2s by adopting dilated depth-wise separable convolutions and fusing operations. We also reduced the model download size to approximately 400KB through post-training weight quantization. We calibrated a speed test based on partial model evaluation time to determine if the harmonization request should be performed locally or sent to remote TPU servers. In three days, people spent 350 years worth of time playing with the Bach Doodle, and Coconet received more than 55 million queries. Users could choose to rate their compositions and contribute them to a public dataset, which we are releasing with this paper. We hope that the community finds this dataset useful for applications ranging from ethnomusicological studies, to music education, to improving machine learning models.

Machine learning models of music typically break up the task of composition into a chronological process, composing a piece of music in a single pass from beginning to end. On the contrary, human composers write music in a nonlinear fashion, scribbling motifs here and there, often revisiting choices previously made. In order to better approximate this process, we train a convolutional neural network to complete partial musical scores, and explore the use of blocked Gibbs sampling as an analogue to rewriting. Neither the model nor the generative procedure are tied to a particular causal direction of composition. Our model is an instance of orderless NADE (Uria et al., 2014), which allows more direct ancestral sampling. However, we find that Gibbs sampling greatly improves sample quality, which we demonstrate to be due to some conditional distributions being poorly modeled. Moreover, we show that even the cheap approximate blocked Gibbs procedure from Yao et al. (2014) yields better samples than ancestral sampling, based on both log-likelihood and human evaluation.

Generating musical audio directly with neural networks is notoriously difficult because it requires coherently modeling structure at many different timescales. Fortunately, most music is also highly structured and can be represented as discrete note events played on musical instruments. Herein, we show that by using notes as an intermediate representation, we can train a suite of models capable of transcribing, composing, and synthesizing audio waveforms with coherent musical structure on timescales spanning six orders of magnitude ( 0.1 ms to 100 s), a process we call Wave2Midi2Wave. This large advance in the state of the art is enabled by our release of the new MAESTRO (MIDI and Audio Edited for Synchronous TRacks and Organization) dataset, composed of over 172 hours of virtuosic piano performances captured with fine alignment ( 3 ms) between note labels and audio waveforms. The networks and the dataset together present a promising approach toward creating new expressive and interpretable neural models of music. Since the beginning of the recent wave of deep learning research, there have been many attempts to create generative models of expressive musical audio de novo. These models would ideally generate audio that is both musically and sonically realistic to the point of being indistinguishable to a listener from music composed and performed by humans. However, modeling music has proven extremely difficult due to dependencies across the wide range of timescales that give rise to the characteristics of pitch and timbre (short-term) as well as those of rhythm (medium-term) and song structure (long-term). On the other hand, much of music has a large hierarchy of discrete structure embedded in its generative process: a composer creates songs, sections, and notes, and a performer realizes those notes with discrete events on their instrument, creating sound.

Music relies heavily on self-reference to build structure and meaning. We explore the Transformer architecture (Vaswani et al., 2017) as a generative model for music, as self-attention has shown compelling results on tasks that require long-term structure such as Wikipedia summary generation (Liu et al, 2018). However, timing information is critical for polyphonic music, and Transformer does not explicitly model absolute or relative timing in its structure. To address this challenge, Shaw et al. (2018) introduced relative position representations to self-attention to improve machine translation. However, the formulation was not scalable to longer sequences. We propose an improved formulation which reduces the memory requirements of the relative position computation from $O(l^2d)$ to $O(ld)$, making it possible to train much longer sequences and achieve faster convergence. In experiments on symbolic music we find that relative self-attention substantially improves sample quality for unconditioned generation and is able to generate sequences of lengths longer than those from the training set. When primed with an initial sequence, the model generates continuations that develop the prime coherently and exhibit long-term structure. Relative self-attention can be instrumental in capturing richer relationships within a musical piece.