VoiceFilter separates target speaker from multi-speaker signals.
problem Speech recognition in multi-speaker environments.
method Speaker recognition network and spectrogram masking network trained together.
result Significant reduction in speech recognition WER on multi-speaker signals.
AaSP improves audio self-supervised learning by addressing aliasing issues.
problem Alias issues in audio spectrogram transformers.
method AaSP combines aliasing-aware patch representation, teacher-student masked modeling, cross-attention predictor, and contrastive regularization.
result AaSP learns more stable representations that integrate high-frequency cues.
MASnet enhances speech on mobile devices with low latency.
problem Efficiently enhancing speech on mobile devices with low latency.
method MASnet processes linear-scale spectrograms, using ratio masks to enhance noisy frames, and operates in low-latency incremental inference mode.
result MASnet achieves efficient speech enhancement with low latency, reducing FMA/s operations.
Framework improves marine mammal monitoring in noisy underwater environments.
problem Underwater bioacoustic monitoring challenges due to overlapping calls and variable noise.
method Multi-step attention-guided framework with segmentation and mid-level fusion.
result Improved signal discrimination, reduced false positives, reliable representations.
Advanced deep learning model improves speech enhancement by estimating phase accurately.
problem Difficulty in estimating the phase of clean speech in speech enhancement.
method Proposes Deep Complex U-Net, polar coordinate-wise complex-valued masking, and wSDR loss function.
result Achieves state-of-the-art performance in all metrics, outperforming previous approaches.
Automatic feature extraction using neural networks has accomplished remarkable success for images, but for sound recognition, these models are usually modified to fit the nature of the multi-dimensional temporal representation of the audio signal in spectrograms. This may not efficiently harness the time-frequency repr…
New neural network separates singing voices from music using cross entropy loss.
problem Separating singing voices from music accompaniment.
method Deep Convolutional Neural Network (CNN) trained with Ideal Binary Mask (IBM) and cross entropy loss.
result Proposed CNN outperforms existing systems in MIREX evaluations.
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 …
This paper improves multichannel speech enhancement using complex ratio masking and channel-attention.
problem Limited performance of deep learning methods in multichannel speech enhancement.
method Introduces complex ratio masking and channel-attention mechanism inside a U-Net architecture.
result Demonstrates superior performance on the CHiME-3 dataset.
MaskCycleGAN-VC improves voice conversion without parallel data.
problem Limited ability to convert mel-spectrogram data without parallel data.
method Integrates a novel auxiliary task called filling in frames (FIF) to learn time-frequency structures.
result MaskCycleGAN-VC outperforms existing methods with similar model size.
Improved music source separation using unlabeled data remixing.
problem Music source separation with deep learning.
method Introduces a simple convolutional and recurrent model and a scheme to leverage unlabeled music.
result Waveform methods can now match spectrogram methods on standard benchmarks.
We address the problem of acoustic source separation in a deep learning framework we call "deep clustering." Rather than directly estimating signals or masking functions, we train a deep network to produce spectrogram embeddings that are discriminative for partition labels given in training data. Previous deep network …
iSTFTNet speeds up mel-spectrogram vocoders without sacrificing quality.
problem Efficiently converting mel-spectrograms to speech with minimal computation.
method Replaces convolutional layers with iSTFT after frequency dimension reduction.
result Significant reduction in computational cost with comparable quality.
Adversarial attacks on spectrograms can fool audio classifiers trained on waveforms.
problem Susceptibility of audio classifiers to adversarial attacks on spectrograms.
method Applying adversarial attacks to spectrograms and reconstructing audio waveforms.
result Perturbed spectrograms can fool 2D CNNs and 1D CNNs trained on audio waveforms.
Spectrogram-Channels U-Net separates sounds by treating each channel as a source's spectrogram.
problem Sound source separation in music information retrieval.
method Adapting U-Net to treat each channel of the output as a source's spectrogram, balancing volumes between sources.
result State-of-the-art performance on singing voice and multi-instrument separation.
This paper aims to address two issues existing in the current speech enhancement methods: 1) the difficulty of phase estimations; 2) a single objective function cannot consider multiple metrics simultaneously. To solve the first problem, we propose a novel convolutional neural network (CNN) model for complex spectrogra…
Improved speech emotion recognition using pitch-synchronous single frequency filtering spectrogram.
problem Uncertainty principles in STFT spectrogram limit time and frequency resolutions.
method Modified SFF spectrogram by averaging amplitudes between GCI locations, named pitch-synchronous SFF spectrogram.
result Improved SER accuracy (63.95% to 70.4%) on IEMOCAP dataset.
Study improves voice conversion model with Mel-spectrogram augmentation.
problem Insufficient speech pairs data for training sequence-to-sequence voice conversion models.
method Experimented with Mel-spectrogram augmentation using SpecAugment policies and proposed new augmentation policies.
result Time axis warping policies showed better performance in training the voice conversion model.
CycleGAN-VC3 improves CycleGAN-VCs for mel-spectrogram conversion.
problem Ambiguity in CycleGAN-VC/VC2 effectiveness for mel-spectrogram conversion.
method Proposes CycleGAN-VC3 with time-frequency adaptive normalization (TFAN).
result CycleGAN-VC3 outperforms or matches CycleGAN-VC2 for mel-spectrogram conversion.
Improved music source separation using spectrogram feature loss.
problem Music source separation quality improvement.
method Added a high-level feature loss term from spectrograms using a VGG net to a deep learning model.
result Improvement in separation quality of drums and vocals from songs.
Improved U-Nets with various intermediate blocks enhance singing voice separation.
problem Improving singing voice separation accuracy using U-Net architectures.
method Implemented and compared U-Nets with different intermediate spectrogram transformation blocks.
result A specific block type achieves state-of-the-art SDR by 0.9 dB.
This paper proposes a multichannel source separation technique called the multichannel variational autoencoder (MVAE) method, which uses a conditional VAE (CVAE) to model and estimate the power spectrograms of the sources in a mixture. By training the CVAE using the spectrograms of training examples with source-class l…
Novel CSK kernel improves GP model generalization for non-stationary patterns.
problem Improving generalization of Gaussian process models for non-stationary data.
method Introduced convolutional spectral kernel (CSK) derived from convolution of imaginary radial basis functions, using Fourier transform for interpretation.
result CSK improves GP model generalization on spatiotemporal datasets.
In this paper, we address the problem of reconstructing a time-domain signal (or a phase spectrogram) solely from a magnitude spectrogram. Since magnitude spectrograms do not contain phase information, we must restore or infer phase information to reconstruct a time-domain signal. One widely used approach for dealing w…
A new RBM model handles both linear and log-amplitude spectrograms.
problem Handling amplitude spectra with existing models.
method Proposed gamma-Bernoulli RBM that uses gamma distribution.
result The model can naturally handle positive numbers and log-amplitude spectrograms.
Deep model generates high-quality speech from spectrograms.
problem Speech reconstruction from spectrograms.
method Deep generative model with Gaussian and von Mises distributions for magnitude and phase, variational autoencoder framework.
result Generated speech has high perceptual quality and intelligibility.
Optimal transport as a loss for machine learning optimization problems has recently gained a lot of attention. Building upon recent advances in computational optimal transport, we develop an optimal transport non-negative matrix factorization (NMF) algorithm for supervised speech blind source separation (BSS). Optimal …
WaveGlow generates high-quality speech from spectrograms.
problem Speech synthesis quality and efficiency.
method WaveGlow combines Glow and WaveNet insights, using a single network and cost function for efficient, high-quality audio synthesis.
result WaveGlow produces audio samples at over 500 kHz, matching WaveNet quality.
Paper presents a deep learning framework for classifying respiratory anomalies and lung diseases from sound recordings.
problem Classifying respiratory anomalies and lung diseases from respiratory sound recordings.
method The framework uses front-end feature extraction to transform sound into spectrograms, and a deep learning network to classify these features.
result The proposed deep learning system outperforms current state-of-the-art methods on the ICBHI benchmark dataset.
Study compares new audio representation methods for limited data music retrieval.
problem Improving machine learning for audio data with limited training data.
method Investigated mel-spectrogram and Mel scattering representations, and augmented target loss function.
result All proposed methods outperform standard mel-spectrogram when using limited data.
Improving speech system performance in noisy environments remains a challenging task, and speech enhancement (SE) is one of the effective techniques to solve the problem. Motivated by the promising results of generative adversarial networks (GANs) in a variety of image processing tasks, we explore the potential of cond…
PMI-Masking improves MLM pretraining by masking correlated spans efficiently.
problem Uniform token masking leads to inefficient and suboptimal performance in MLMs.
method PMI-Masking uses Pointwise Mutual Information to mask n-grams with high collocation.
result PMI-Masking reaches half the training time and improves performance.
Graph neural networks improve music genre classification on audio datasets.
problem Difficulty in applying deep learning on spectrograms due to lack of quality data and augmentation.
method Combination of CNN and Graph Neural Networks (GNN) with Siamese Neural Networks.
result Achieved state-of-the-art results on GTZAN and AudioSet datasets.
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.
X-DC improves speech separation by making DNNs more interpretable.
problem Black-box nature of DNNs in speech separation tasks.
method Introduces X-DC, a DNN architecture that interprets as spectrogram template fitting followed by Wiener filtering.
result X-DC achieves comparable speech separation performance to DC but with enhanced interpretability.
ST-MTM models complex time series by decomposing and masking seasonal and trend components.
problem Forecasting complex time series with intricate temporal variations.
method Seasonal-Trend Decomposition with Masking and Contrastive Learning.
result ST-MTM achieves superior forecasting performance compared to existing methods.
CLCNet improves noise reduction in hearing aids with deep learning.
problem Noise reduction in hearing aids is challenging due to real-time and frequency resolution constraints.
method Proposes CLCNet, a deep learning framework based on complex linear coding.
result CLCNet outperforms traditional methods in noisy environments.
Expands MLM by masking token positions, improving performance and convergence.
problem Improving language model performance and convergence.
method Masking token positions along with [MASK] tokens, using a fully connected classifier stage.
result Shows .3% improvement and 50% faster convergence for BERT Base with position masking.
Deep learning detects atrial fibrillation with high accuracy.
problem Detecting atrial fibrillation in ECG signals.
method Extracted deep features from spectrograms using convolutional networks.
result Convolutional network achieved 93.16% classification accuracy.
SMART training improves mask-predict translations.
problem Closing the performance gap between semi-autoregressive and autoregressive models.
method SMART training method for conditional masked language models.
result SMART-trained models produce higher-quality translations.
New method reduces diffusion model function evaluations for discrete data.
problem High computational burden in generating samples from masked diffusion models.
method Modified causal attention mask and speculative sampling mechanism for non-factorized predictions.
result Achieved ~2x reduction in required network forward passes.
Proposes a proportional masking strategy for better tabular data imputation.
problem Heterogeneity of tabular data disrupts uniform random masking in MAEs.
method Computes missingness statistics, generates proportional masks, uses MLP token mixing.
result Proportional masking preserves missingness distribution, improves imputation performance.
End-to-end neural network based approaches to audio modelling are generally outperformed by models trained on high-level data representations. In this paper we present preliminary work that shows the feasibility of training the first layers of a deep convolutional neural network (CNN) model to learn the commonly-used l…
iSTFTNet2 improves iSTFTNet's speed and lightness with 1D-2D CNN.
problem Efficiently synthesizing high-fidelity speech.
method Improved iSTFTNet using 1D-2D CNNs for temporal and spectrogram structures.
result iSTFTNet2 is faster and more lightweight with comparable speech quality.
BshapeNet improves object detection and instance segmentation using bounding shape masks.
problem Improving object detection and instance segmentation performance.
method Proposes bbox and bshape masks, and combines them with Faster R-CNN and Mask R-CNN frameworks.
result BshapeNet+ achieves highly competitive results with state of the art models, improving detection and instance segmentation.
New analysis reveals masked self-supervised learning's effectiveness in extracting data structure.
problem Analyzing masked self-supervised learning in high-dimensional data.
method Developed precise high-dimensional analysis of masked modeling objectives.
result Identified phase transitions and structured regimes for masked self-supervised learning.
ResNet with Focal Loss improves speech emotion recognition.
problem Speech emotion recognition using plain text features is insufficient.
method Residual Convolutional Neural Network (ResNet) trained with Focal Loss.
result Focal Loss enhances model's focus on hard examples.
ConvNet classifies whale vocalizations and ambient noise in acoustic recordings.
problem Automated detection and classification of marine mammal vocalizations in acoustic recordings.
method Convolutional Neural Network with a novel acoustic representation.
result Classifier accurately detects and classifies whale vocalizations and ambient noise.