Search for: All records

Musicians and nonmusicians alike use rhythmic sound gestures, such as tapping and beatboxing, to express drum patterns. While these gestures effectively communicate musical ideas, realizing these ideas as fully-produced drum recordings can be time-consuming, potentially disrupting many creative workflows. To bridge this gap, we present TRIA (The Rhythm In Anything), a masked transformer model for mapping rhythmic sound gestures to high-fidelity drum recordings. Given an audio prompt of the desired rhythmic pattern and a second prompt to represent drum kit timbre, TRIA produces audio of a drum kit playing the desired rhythm (with appropriate elaborations) in the desired timbre. Subjective and objective evaluations show that a TRIA model trained on less than 10 hours of publicly-available drum data can generate high-quality, faithful realizations of sound gestures across a wide range of timbres in a zero-shot manner. 

Every artist has a creative process that draws inspiration from previous artists and their works. Today, “inspiration” has been automated by generative music models. The black box nature of these models obscures the identity of the works that influence their creative output. As a result, users may inadvertently appropriate, misuse, or copy existing artists’ works. We establish a replicable methodology to systematically identify similar pieces of music audio in a manner that is useful for understanding training data attribution. A key aspect of our approach is to harness an effective music audio similarity measure. We compare the effect of applying CLMR [50] and CLAP [55] embeddings to similarity measurement in a set of 5 million audio clips used to train VampNet [24], a recent open source generative music model. We validate this approach with a human listening study. We also explore the effect that modifications of an audio example (e.g., pitch shifting, time stretching, background noise) have on similarity measurements. This work is foundational to incorporating automated influence attribution into generative modeling, which promises to let model creators and users move from ignorant appropriation to informed creation. Audio samples that accompany this paper are available at https://tinyurl.com/exploring-musical-roots. 

Hydrophobic interactions drive the binding of nonpolar ligands to the oily pockets of proteins and supramolecular species in aqueous solution. As such, the wetting of host pockets is expected to play a critical role in determining the thermodynamics of guest binding. Here we use molecular simulations to examine the impact of pressure on the wetting and dewetting of the nonpolar pockets of a series of deep-cavity cavitands in water. The portals to the cavitand pockets are functionalized with both nonpolar (methyl) and polar (hydroxyl) groups oriented pointing either upward or inward toward the pocket. We find wetting of the pocket is favored by the hydroxyl groups and dewetting is favored by the methyl groups. The distribution of waters in the pocket is found to exhibit a two-state-like equilibrium between wet and dry states with a free energy barrier between the two states. Moreover, we demonstrate that the pocket hydration of the cavitands can be collapsed onto a unified adsorption isotherm by assuming the effective pressures within each cavitand pocket differ by a shift pressure that depends on the chemical identity and number of functional groups placed about the portal. These observations support the development of a twostate capillary evaporation model that accurately describes the equilibrium between states and naturally gives rise to the effective shift pressures observed from simulation. This work demonstrates that the hydration of host pockets can be tuned following simple design rules that in turn are expected to impact the thermodynamics of guest complexation. 

The field of polymer membrane design is primarily based on empirical observation, which limits discovery of new materials optimized for separating a given gas pair. Instead of relying on exhaustive experimental investigations, we trained a machine learning (ML) algorithm, using a topological, path-based hash of the polymer repeating unit. We used a limited set of experimental gas permeability data for six different gases in ~700 polymeric constructs that have been measured to date to predict the gas-separation behavior of over 11,000 homopolymers not previously tested for these properties. To test the algorithm’s accuracy, we synthesized two of the most promising polymer membranes predicted by this approach and found that they exceeded the upper bound for CO 2 /CH 4 separation performance. This ML technique, which is trained using a relatively small body of experimental data (and no simulation data), evidently represents an innovative means of exploring the vast phase space available for polymer membrane design. 

There are many open questions regarding the hydration of solvent-exposed, non-polar tracts and pockets in proteins. Although water is predicted to de-wet purely repulsive surfaces and evacuate crevices, the extent of de-wetting is unclear when ubiquitous van der Waals interactions are in play. The structural simplicity of synthetic supramolecular hosts imbues them with considerable potential to address this issue. To this end, here we detail a combination of densimetry and molecular dynamics simulations of three cavitands, coupled with calorimetric studies of their complexes with short-chain carboxylates. Our results reveal the range of wettability possible within the ostensibly identical cavitand pockets — which differ only in the presence/position of the methyl groups that encircle the portal to their non-polar pockets. The results demonstrate the ability of macrocycles to template water cavitation within their binding sites and show how the orientation of methyl groups can trigger the drying of non-polar pockets in liquid water, suggesting new avenues to control guest complexation.