Bryan Pardo

Orr Paradise, Pranav Muralikrishnan, Liangyuan Chen et al.

Sperm whales communicate in short sequences of clicks known as codas. We present WhAM (Whale Acoustics Model), the first transformer-based model capable of generating synthetic sperm whale codas from any audio prompt. WhAM is built by finetuning VampNet, a masked acoustic token model pretrained on musical audio, using 10k coda recordings collected over the past two decades. Through iterative masked token prediction, WhAM generates high-fidelity synthetic codas that preserve key acoustic features of the source recordings. We evaluate WhAM's synthetic codas using Fréchet Audio Distance and through perceptual studies with expert marine biologists. On downstream classification tasks including rhythm, social unit, and vowel classification, WhAM's learned representations achieve strong performance, despite being trained for generation rather than classification. Our code is available at https://github.com/Project-CETI/wham

Max Morrison, Brian Tang, Gefei Tan et al.

Human perceptual studies are the gold standard for the evaluation of many research tasks in machine learning, linguistics, and psychology. However, these studies require significant time and cost to perform. As a result, many researchers use objective measures that can correlate poorly with human evaluation. When subjective evaluations are performed, they are often not reported with sufficient detail to ensure reproducibility. We propose Reproducible Subjective Evaluation (ReSEval), an open-source framework for quickly deploying crowdsourced subjective evaluations directly from Python. ReSEval lets researchers launch A/B, ABX, Mean Opinion Score (MOS) and MUltiple Stimuli with Hidden Reference and Anchor (MUSHRA) tests on audio, image, text, or video data from a command-line interface or using one line of Python, making it as easy to run as objective evaluation. With ReSEval, researchers can reproduce each other's subjective evaluations by sharing a configuration file and the audio, image, text, or video files.

Hugo Flores Garcia, Prem Seetharaman, Rithesh Kumar et al.

We introduce VampNet, a masked acoustic token modeling approach to music synthesis, compression, inpainting, and variation. We use a variable masking schedule during training which allows us to sample coherent music from the model by applying a variety of masking approaches (called prompts) during inference. VampNet is non-autoregressive, leveraging a bidirectional transformer architecture that attends to all tokens in a forward pass. With just 36 sampling passes, VampNet can generate coherent high-fidelity musical waveforms. We show that by prompting VampNet in various ways, we can apply it to tasks like music compression, inpainting, outpainting, continuation, and looping with variation (vamping). Appropriately prompted, VampNet is capable of maintaining style, genre, instrumentation, and other high-level aspects of the music. This flexible prompting capability makes VampNet a powerful music co-creation tool. Code and audio samples are available online.

Noah Schaffer, Boaz Cogan, Ethan Manilow et al.

Despite phenomenal progress in recent years, state-of-the-art music separation systems produce source estimates with significant perceptual shortcomings, such as adding extraneous noise or removing harmonics. We propose a post-processing model (the Make it Sound Good (MSG) post-processor) to enhance the output of music source separation systems. We apply our post-processing model to state-of-the-art waveform-based and spectrogram-based music source separators, including a separator unseen by MSG during training. Our analysis of the errors produced by source separators shows that waveform models tend to introduce more high-frequency noise, while spectrogram models tend to lose transients and high frequency content. We introduce objective measures to quantify both kinds of errors and show MSG improves the source reconstruction of both kinds of errors. Crowdsourced subjective evaluations demonstrate that human listeners prefer source estimates of bass and drums that have been post-processed by MSG.

Julia Barnett, Hugo Flores Garcia, Bryan Pardo

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 and CLAP embeddings to similarity measurement in a set of 5 million audio clips used to train VampNet, 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.

Hugo Flores Garcia, Aldo Aguilar, Ethan Manilow et al.

We present a software framework that integrates neural networks into the popular open-source audio editing software, Audacity, with a minimal amount of developer effort. In this paper, we showcase some example use cases for both end-users and neural network developers. We hope that this work fosters a new level of interactivity between deep learning practitioners and end-users.

Christodoulos Benetatos, Frank Cwitkowitz, Nathan Pruyne et al.

HARP 2.0 brings deep learning models to digital audio workstation (DAW) software through hosted, asynchronous, remote processing, allowing users to route audio from a plug-in interface through any compatible Gradio endpoint to perform arbitrary transformations. HARP renders endpoint-defined controls and processed audio in-plugin, meaning users can explore a variety of cutting-edge deep learning models without ever leaving the DAW. In the 2.0 release we introduce support for MIDI-based models and audio/MIDI labeling models, provide a streamlined pyharp Python API for model developers, and implement numerous interface and stability improvements. Through this work, we hope to bridge the gap between model developers and creatives, improving access to deep learning models by seamlessly integrating them into DAW workflows.

Qixin Deng, Bryan Pardo, Thrasyvoulos N Pappas

Understanding and modeling the relationship between language and sound is critical for applications such as music information retrieval,text-guided music generation, and audio captioning. Central to these tasks is the use of joint language-audio embedding spaces, which map textual descriptions and auditory content into a shared embedding space. While multimodal embedding models such as MS-CLAP, LAION-CLAP, and MuQ-MuLan have shown strong performance in aligning language and audio, their correspondence to human perception of timbre, a multifaceted attribute encompassing qualities such as brightness, roughness, and warmth, remains underexplored. In this paper, we evaluate the above three joint language-audio embedding models on their ability to capture perceptual dimensions of timbre. Our findings show that LAION-CLAP consistently provides the most reliable alignment with human-perceived timbre semantics across both instrumental sounds and audio effects.

Ethan Manilow, Patrick O'Reilly, Prem Seetharaman et al.

We showcase an unsupervised method that repurposes deep models trained for music generation and music tagging for audio source separation, without any retraining. An audio generation model is conditioned on an input mixture, producing a latent encoding of the audio used to generate audio. This generated audio is fed to a pretrained music tagger that creates source labels. The cross-entropy loss between the tag distribution for the generated audio and a predefined distribution for an isolated source is used to guide gradient ascent in the (unchanging) latent space of the generative model. This system does not update the weights of the generative model or the tagger, and only relies on moving through the generative model's latent space to produce separated sources. We use OpenAI's Jukebox as the pretrained generative model, and we couple it with four kinds of pretrained music taggers (two architectures and two tagging datasets). Experimental results on two source separation datasets, show this approach can produce separation estimates for a wider variety of sources than any tested supervised or unsupervised system. This work points to the vast and heretofore untapped potential of large pretrained music models for audio-to-audio tasks like source separation.

Max Morrison, Zeyu Jin, Nicholas J. Bryan et al.

Modifying the pitch and timing of an audio signal are fundamental audio editing operations with applications in speech manipulation, audio-visual synchronization, and singing voice editing and synthesis. Thus far, methods for pitch-shifting and time-stretching that use digital signal processing (DSP) have been favored over deep learning approaches due to their speed and relatively higher quality. However, even existing DSP-based methods for pitch-shifting and time-stretching induce artifacts that degrade audio quality. In this paper, we propose Controllable LPCNet (CLPCNet), an improved LPCNet vocoder capable of pitch-shifting and time-stretching of speech. For objective evaluation, we show that CLPCNet performs pitch-shifting of speech on unseen datasets with high accuracy relative to prior neural methods. For subjective evaluation, we demonstrate that the quality and naturalness of pitch-shifting and time-stretching with CLPCNet on unseen datasets meets or exceeds competitive neural- or DSP-based approaches.

Hugo Flores Garcia, Aldo Aguilar, Ethan Manilow et al.

Deep learning work on musical instrument recognition has generally focused on instrument classes for which we have abundant data. In this work, we exploit hierarchical relationships between instruments in a few-shot learning setup to enable classification of a wider set of musical instruments, given a few examples at inference. We apply a hierarchical loss function to the training of prototypical networks, combined with a method to aggregate prototypes hierarchically, mirroring the structure of a predefined musical instrument hierarchy. These extensions require no changes to the network architecture and new levels can be easily added or removed. Compared to a non-hierarchical few-shot baseline, our method leads to a significant increase in classification accuracy and significant decrease mistake severity on instrument classes unseen in training.

Max Morrison, Lucas Rencker, Zeyu Jin et al.

Text-based speech editors expedite the process of editing speech recordings by permitting editing via intuitive cut, copy, and paste operations on a speech transcript. A major drawback of current systems, however, is that edited recordings often sound unnatural because of prosody mismatches around edited regions. In our work, we propose a new context-aware method for more natural sounding text-based editing of speech. To do so, we 1) use a series of neural networks to generate salient prosody features that are dependent on the prosody of speech surrounding the edit and amenable to fine-grained user control 2) use the generated features to control a standard pitch-shift and time-stretch method and 3) apply a denoising neural network to remove artifacts induced by the signal manipulation to yield a high-fidelity result. We evaluate our approach using a subjective listening test, provide a detailed comparative analysis, and conclude several interesting insights.

Andreas Bugler, Bryan Pardo, Prem Seetharaman

Supervised deep learning methods for performing audio source separation can be very effective in domains where there is a large amount of training data. While some music domains have enough data suitable for training a separation system, such as rock and pop genres, many musical domains do not, such as classical music, choral music, and non-Western music traditions. It is well known that transferring learning from related domains can result in a performance boost for deep learning systems, but it is not always clear how best to do pretraining. In this work we investigate the effectiveness of data augmentation during pretraining, the impact on performance as a result of pretraining and downstream datasets having similar content domains, and also explore how much of a model must be retrained on the final target task, once pretrained.

Ethan Manilow, Bryan Pardo

In this paper, we introduce a simple method that can separate arbitrary musical instruments from an audio mixture. Given an unaligned MIDI transcription for a target instrument from an input mixture, we synthesize new mixtures from the midi transcription that sound similar to the mixture to be separated. This lets us create a labeled training set to train a network on the specific bespoke task. When this model applied to the original mixture, we demonstrate that this method can: 1) successfully separate out the desired instrument with access to only unaligned MIDI, 2) separate arbitrary instruments, and 3) get results in a fraction of the time of existing methods. We encourage readers to listen to the demos posted here: https://git.io/JUu5q.

Prem Seetharaman, Gordon Wichern, Bryan Pardo et al.

Clipping the gradient is a known approach to improving gradient descent, but requires hand selection of a clipping threshold hyperparameter. We present AutoClip, a simple method for automatically and adaptively choosing a gradient clipping threshold, based on the history of gradient norms observed during training. Experimental results show that applying AutoClip results in improved generalization performance for audio source separation networks. Observation of the training dynamics of a separation network trained with and without AutoClip show that AutoClip guides optimization into smoother parts of the loss landscape. AutoClip is very simple to implement and can be integrated readily into a variety of applications across multiple domains.

Alisa Liu, Alexander Fang, Gaëtan Hadjeres et al.

Deep learning has rapidly become the state-of-the-art approach for music generation. However, training a deep model typically requires a large training set, which is often not available for specific musical styles. In this paper, we present augmentative generation (Aug-Gen), a method of dataset augmentation for any music generation system trained on a resource-constrained domain. The key intuition of this method is that the training data for a generative system can be augmented by examples the system produces during the course of training, provided these examples are of sufficiently high quality and variety. We apply Aug-Gen to Transformer-based chorale generation in the style of J.S. Bach, and show that this allows for longer training and results in better generative output.

Alexander Fang, Alisa Liu, Prem Seetharaman et al.

Deep generative systems that learn probabilistic models from a corpus of existing music do not explicitly encode knowledge of a musical style, compared to traditional rule-based systems. Thus, it can be difficult to determine whether deep models generate stylistically correct output without expert evaluation, but this is expensive and time-consuming. Therefore, there is a need for automatic, interpretable, and musically-motivated evaluation measures of generated music. In this paper, we introduce a grading function that evaluates four-part chorales in the style of J.S. Bach along important musical features. We use the grading function to evaluate the output of a Transformer model, and show that the function is both interpretable and outperforms human experts at discriminating Bach chorales from model-generated ones.

Max Morrison, Bryan Pardo

Early detection and repair of failing components in automobiles reduces the risk of vehicle failure in life-threatening situations. Many automobile components in need of repair produce characteristic sounds. For example, loose drive belts emit a high-pitched squeaking sound, and bad starter motors have a characteristic whirring or clicking noise. Often drivers can tell that the sound of their car is not normal, but may not be able to identify the cause. To mitigate this knowledge gap, we have developed OtoMechanic, a web application to detect and diagnose vehicle component issues from their corresponding sounds. It compares a user's recording of a problematic sound to a database of annotated sounds caused by failing automobile components. OtoMechanic returns the most similar sounds, and provides weblinks for more information on the diagnosis associated with each sound, along with an estimate of the similarity of each retrieved sound. In user studies, we find that OtoMechanic significantly increases diagnostic accuracy relative to a baseline accuracy of consumer performance.

Alisa Liu, Prem Seetharaman, Bryan Pardo

Audio source separation is the process of separating a mixture (e.g. a pop band recording) into isolated sounds from individual sources (e.g. just the lead vocals). Deep learning models are the state-of-the-art in source separation, given that the mixture to be separated is similar to the mixtures the deep model was trained on. This requires the end user to know enough about each model's training to select the correct model for a given audio mixture. In this work, we automate selection of the appropriate model for an audio mixture. We present a confidence measure that does not require ground truth to estimate separation quality, given a deep model and audio mixture. We use this confidence measure to automatically select the model output with the best predicted separation quality. We compare our confidence-based ensemble approach to using individual models with no selection, to an oracle that always selects the best model and to a random model selector. Results show our confidence-based ensemble significantly outperforms the random ensemble over general mixtures and approaches oracle performance for music mixtures.

Prem Seetharaman, Gordon Wichern, Jonathan Le Roux et al.

Separating an audio scene such as a cocktail party into constituent, meaningful components is a core task in computer audition. Deep networks are the state-of-the-art approach. They are trained on synthetic mixtures of audio made from isolated sound source recordings so that ground truth for the separation is known. However, the vast majority of available audio is not isolated. The brain uses primitive cues that are independent of the characteristics of any particular sound source to perform an initial segmentation of the audio scene. We present a method for bootstrapping a deep model for music source separation without ground truth by using multiple primitive cues. We apply our method to train a network on a large set of unlabeled music recordings from YouTube to separate vocals from accompaniment without the need for ground truth isolated sources or artificial training mixtures.

Ethan Manilow, Prem Seetharaman, Bryan Pardo

We present a single deep learning architecture that can both separate an audio recording of a musical mixture into constituent single-instrument recordings and transcribe these instruments into a human-readable format at the same time, learning a shared musical representation for both tasks. This novel architecture, which we call Cerberus, builds on the Chimera network for source separation by adding a third "head" for transcription. By training each head with different losses, we are able to jointly learn how to separate and transcribe up to 5 instruments in our experiments with a single network. We show that the two tasks are highly complementary with one another and when learned jointly, lead to Cerberus networks that are better at both separation and transcription and generalize better to unseen mixtures.

Prem Seetharaman, Gordon Wichern, Jonathan Le Roux et al.

Separating an audio scene into isolated sources is a fundamental problem in computer audition, analogous to image segmentation in visual scene analysis. Source separation systems based on deep learning are currently the most successful approaches for solving the underdetermined separation problem, where there are more sources than channels. Traditionally, such systems are trained on sound mixtures where the ground truth decomposition is already known. Since most real-world recordings do not have such a decomposition available, this limits the range of mixtures one can train on, and the range of mixtures the learned models may successfully separate. In this work, we use a simple blind spatial source separation algorithm to generate estimated decompositions of stereo mixtures. These estimates, together with a weighting scheme in the time-frequency domain, based on confidence in the separation quality, are used to train a deep learning model that can be used for single-channel separation, where no source direction information is available. This demonstrates how a simple cue such as the direction of origin of source can be used to bootstrap a model for source separation that can be used in situations where that cue is not available.

Zafar Rafii, Antoine Liutkus, Fabian-Robert Stöter et al.

Popular music is often composed of an accompaniment and a lead component, the latter typically consisting of vocals. Filtering such mixtures to extract one or both components has many applications, such as automatic karaoke and remixing. This particular case of source separation yields very specific challenges and opportunities, including the particular complexity of musical structures, but also relevant prior knowledge coming from acoustics, musicology or sound engineering. Due to both its importance in applications and its challenging difficulty, lead and accompaniment separation has been a popular topic in signal processing for decades. In this article, we provide a comprehensive review of this research topic, organizing the different approaches according to whether they are model-based or data-centered. For model-based methods, we organize them according to whether they concentrate on the lead signal, the accompaniment, or both. For data-centered approaches, we discuss the particular difficulty of obtaining data for learning lead separation systems, and then review recent approaches, notably those based on deep learning. Finally, we discuss the delicate problem of evaluating the quality of music separation through adequate metrics and present the results of the largest evaluation, to-date, of lead and accompaniment separation systems. In conjunction with the above, a comprehensive list of references is provided, along with relevant pointers to available implementations and repositories.