Proposes DNN for enhancing sound quality scores.
problem Improving sound quality assessment scores.
method Develops a DNN optimization scheme based on black-box optimization and policy gradient method.
result Significant increase in OSQA scores without minimizing MSE.
Stabilizes training of DNN for speech enhancement using PESQ scores.
problem Stability issues in training DNNs using non-differentiable OSQA scores.
method Approximate OSQA scores with a differentiable auxiliary DNN and stabilize training with reinforcement learning techniques.
result Stable training of DNN to achieve state-of-the-art PESQ scores and better sound quality.
Deep learning improves singing processing tasks.
problem Lack of data and computing resources for singing processing.
method State-of-the-art deep learning techniques.
result Advances in accuracy and sound quality.
Neural network synthesizes percussive sounds with adjustable timbral features.
problem Control over high-level timbral characteristics of percussive sounds.
method Feedforward convolutional neural network mapping input parameters to waveform.
result Changing input parameters produces a waveform congruent with desired characteristics.
We augment the nonnegative matrix factorization method for audio source separation with cues about directionality of sound propagation. This improves separation quality greatly and removes the need for training data, with only a twofold increase in run time. This is the first method which can exploit directional inform…
Deep learning models generate music with arbitrary control strategies.
problem Lack of efficient methods for generating music with arbitrary control.
method Deep generative models learn to navigate arbitrary sound spaces.
result Deep learning enables high-quality, arbitrary sound synthesis.
SpecGrad improves neural vocoder sound quality by adapting diffusion noise to log-mel spectrogram.
problem Improving neural vocoder sound quality, especially in high-frequency bands.
method Adapting the diffusion noise distribution to the conditioning log-mel spectrogram through time-varying filtering.
result SpecGrad generates higher-fidelity speech waveform than conventional DDPM-based neural vocoders.
Optimizes audio codec selection with statistical guarantees.
problem Selecting the best audio encoding scheme for various data types.
method Supervised learning with uniform convergence theory.
result Rigorous statistical guarantees for codec selection.
Proposes COALA method for learning audio representations aligned with tags.
problem Lack of annotated data for high-performance audio representation learning.
method Aligns latent representations of audio and tags using a contrastive loss.
result Audio embedding model captures both acoustic and semantic characteristics.
To improve the efficiency of Monte Carlo estimation, practitioners are turning to biased Markov chain Monte Carlo procedures that trade off asymptotic exactness for computational speed. The reasoning is sound: a reduction in variance due to more rapid sampling can outweigh the bias introduced. However, the inexactness …
This paper proposes a method that allows non-parallel many-to-many voice conversion (VC) by using a variant of a generative adversarial network (GAN) called StarGAN. Our method, which we call StarGAN-VC, is noteworthy in that it (1) requires no parallel utterances, transcriptions, or time alignment procedures for speec…
Two neural samplers improve high-quality sample generation from un-normalized densities.
problem Generating high-quality samples from un-normalized probability densities.
method Developed two neural samplers using deep neural networks to transform a reference distribution to a target distribution. Training schemes minimize Stein discrepancy variations.
result The proposed samplers generate samples instantaneously and perform better than traditional methods.
Online audio advertising is a particular form of advertising used abundantly in online music streaming services. In these platforms, which tend to host tens of thousands of unique audio advertisements (ads), providing high quality ads ensures a better user experience and results in longer user engagement. Therefore, th…
End-to-end Sanskrit TTS developed with limited data, achieving good quality.
problem Developing natural-sounding speech for Sanskrit with scarce data.
method Fine-tuning Tacotron2 model with WaveGlow and transfer learning.
result Achieved an overall MOS of 3.38 from 37 evaluators.
This paper improves universal sound separation using sound classification.
problem Separating acoustic sources from an open domain, regardless of their class.
method Utilizing semantic embeddings from a sound classifier to condition a separation network.
result Classifier embeddings provide nearly one dB of SNR gain, and iterative models achieve significant performance.
Paper introduces ToyADMOS dataset for detecting anomalous machine sounds.
problem Lack of large-scale datasets for ADMOS anomaly detection.
method Collected anomalous sounds of miniature machines by deliberate damage.
result Released dataset includes over 180 hours of normal and 4,000 anomalous sounds.
Deep learning methods have achieved high performance in sound recognition tasks. Deciding how to feed the training data is important for further performance improvement. We propose a novel learning method for deep sound recognition: Between-Class learning (BC learning). Our strategy is to learn a discriminative feature…
Paper proposes a method to detect unknown anomalous sounds without training data using deep learning and Neyman-Pearson lemma.
problem Unsupervised detection of unknown anomalous sounds in audio data.
method Uses an autoencoder to minimize reconstruction error of normal sounds and Neyman-Pearson lemma to maximize true positive rate under low false positive rate conditions.
result The proposed method improves performance measures of unsupervised anomaly detection in audio data under low false positive rate conditions.
Batch uniformization improves anomaly detection in sound data.
problem Anomaly scores for rare and frequent normal sounds are not uniform.
method Propose batch uniformization to minimize anomaly scores by weighting samples based on their density.
result Improves performance of unsupervised anomaly detection in sound data.
Paper presents an audiovisual model to recognize sounds from weakly labeled video data.
problem Sound recognition from weakly labeled video data.
method Audiovisual fusion model with attention mechanism.
result The model achieves a mean Average Precision (mAP) of 46.16 on AudioSet, outperforming state-of-the-art models.
Deep learning models improve sound separation across various types of sounds.
problem Developing a universal method to separate arbitrary sounds of different types.
method Created a dataset of mixtures containing arbitrary sounds, investigated mask-based separation architectures, and tested different framewise analysis-synthesis bases.
result STFT outperformed learnable bases in universal sound separation tasks.
Smart city surveillance benefits from sound event recognition.
problem Improving monitoring capabilities in smart cities.
method Exploration of several classifiers on the SESA dataset.
result SGD achieved 72.13% accuracy in sound event recognition.
A new method detects anomalous sounds using self-supervised learning.
problem Detecting anomalous sounds in noisy data.
method Flow-based self-supervised density estimation using Normalizing Flows.
result The method improves anomaly detection performance significantly.
In this study we present a kernel based convolution model to characterize neural responses to natural sounds by decoding their time-varying acoustic features. The model allows to decode natural sounds from high-dimensional neural recordings, such as magnetoencephalography (MEG), that track timing and location of human …
New dataset for industrial machine sounds to aid maintenance.
problem Lack of public datasets for industrial machine sounds.
method Recorded normal and anomalous sounds of industrial machines.
result Assists in automated facility maintenance development.
Paper proposes active learning for sound event detection with reduced annotation effort.
problem Reducing annotation effort for sound event detection.
method Change point detection for candidate selection, mismatch-first farthest-traversal for selection, training with context recordings.
result The proposed system achieves similar performance to full annotation with only 2% of data, reducing annotation effort.
This paper examines the speaker identification potential of breath sounds in continuous speech. Speech is largely produced during exhalation. In order to replenish air in the lungs, speakers must periodically inhale. When inhalation occurs in the midst of continuous speech, it is generally through the mouth. Intra-spee…
Deep learning speeds sound speed inversion in ultrasound.
problem Limited high-end ultrasound hardware for shear wave imaging.
method Fully convolutional deep neural network using simulated data.
result Inversion of longitudinal sound speed at high frame rates.
The paper presents a method for sound event localization and detection using CRNN models.
problem Sound event localization and detection in complex environments.
method Consecutive ensemble of CRNN models for estimating event onset, offset, direction of arrival, and classification.
result The proposed method outperforms other participants in the DCASE2019 task3.
Paper improves sound event detection using semi-supervised learning.
problem Weakly labeled sound event detection in polyphonic audio clips.
method Combines tri-training and adversarial learning for semi-supervised learning.
result Significant performance improvement over baseline model.
Paper discusses ASD challenge for machine condition monitoring.
problem Detecting unknown anomalous sounds without labeled data.
method Design and evaluation of a large-scale ASD dataset, novel approaches.
result Several novel approaches developed, evaluation results analyzed.
The paper explores how Lipschitz-continuity improves GAN training stability and quality.
problem Failure and instability in GAN training due to unreliable gradient from optimal discriminative function.
method Investigates the property of optimal discriminative function and proves Lipschitz-continuity is a solution.
result Lipschitz-continuity condition ensures convergence and leads to more stable and higher quality generated samples.
A new method classifies heart sounds using i-vectors and machine learning.
problem Heart sound classification for disease diagnostics.
method Extract i-vectors from MFCC features, apply PCA and VAE for dimensionality reduction, then use GMMs and SVM for classification.
result The method improves heart sound classification by 16% on the Physionet dataset.
Heart diseases constitute a global health burden, and the problem is exacerbated by the error-prone nature of listening to and interpreting heart sounds. This motivates the development of automated classification to screen for abnormal heart sounds. Existing machine learning-based systems achieve accurate classificatio…
Study presents a dataset and methods to handle noisy labels in sound event classification.
problem Label noise in sound event classification datasets.
method Developed a dataset with noisy labels and evaluated CNN baseline systems.
result Training with large amounts of noisy data can outperform training with carefully-labeled data.
A multi-head attention network improves ASC by recognizing overlapping sound patterns.
problem Challenging ASC due to overlapping sound patterns and complex event mixtures.
method Proposes a multi-head attention network to model complex temporal input structures.
result Achieved competitive performance on DCASE 2018 Task 5 dataset.
Automatically detecting sound units of humpback whales in complex time-varying background noises is a current challenge for scientists. In this paper, we explore the applicability of Convolution Neural Network (CNN) method for this task. In the evaluation stage, we present 6 bi-class classification experimentations of …
Paper detects adversarial attacks in sound classification models.
problem Adversarial attacks threaten data-driven models, especially in sound classification.
method Detects adversarial subspaces in unitary vector domain using chordal distance and generalized Schur decomposition.
result Regularized logistic regression detector outperforms other approaches on benchmark datasets.
Reduces sound event localization error by 2.6x with hybrid parametric-deep learning.
problem Sound event localization and detection accuracy.
method Hybrid approach combining parametric spatial audio analysis and deep learning.
result Reduction of localization error by 2.6x compared to baseline.
Generative replay extends sound classification models to new classes without old data.
problem Incrementally refining a sound classifier with new data causes previously learned tasks to degrade.
method Developed a generative replay procedure to generate training data in place of older datasets.
result Generative replay with 4% of old data performs as well as keeping 20% of old data.
InverSynth automatically tunes synthesizer parameters from audio input.
problem Manual tuning of synthesizer parameters is time-consuming and requires expertise.
method Strided convolutional neural networks for inferring synthesizer parameters.
result InverSynth outperforms baselines in synthesizer parameter tuning.
New task AQA tackles acoustic reasoning from sound scenes.
problem Promote research in acoustic reasoning.
method Generate acoustic scenes from elementary sounds and formulate questions.
result Preliminary results with models FiLM and MAC show promise.
New saliency evaluations focus on completeness and soundness, improving explanations.
problem Current saliency evaluations focus on completeness but ignore soundness.
method Introduces new intrinsic evaluation metrics based on completeness and soundness.
result Simple saliency method matches or outperforms prior methods in new evaluations.
Efficient neural network for audio source separation.
problem End-to-end general purpose audio source separation.
method SuDoRMRF structure with simple one-dimensional convolutions for feature aggregation.
result SuDoRMRF achieves high quality audio source separation with minimal computational resources.
Neural network based architectures used for sound recognition are usually adapted from other application domains, which may not harness sound related properties. The ConditionaL Neural Network (CLNN) is designed to consider the relational properties across frames in a temporal signal, and its extension the Masked Condi…
The ConditionaL Neural Network (CLNN) exploits the nature of the temporal sequencing of the sound signal represented in a spectrogram, and its variant the Masked ConditionaL Neural Network (MCLNN) induces the network to learn in frequency bands by embedding a filterbank-like sparseness over the network's links using a …
Even in the absence of any explicit semantic annotation, vast collections of audio recordings provide valuable information for learning the categorical structure of sounds. We consider several class-agnostic semantic constraints that apply to unlabeled nonspeech audio: (i) noise and translations in time do not change t…
TimbreTron transfers musical timbre using CQT and WaveNet.
problem Transfer musical timbre while preserving pitch, rhythm, and loudness.
method Apply image domain style transfer to CQT representation, then generate high-quality waveform with WaveNet.
result TimbreTron recognizably transfers timbre while preserving musical content.