Study improves voice disorder detection system robust to channel effects.
problem Voice signals are sensitive to recording devices.
method Bidirectional LSTM network with domain adversarial training (DAT).
result Increased PR-AUC from 0.8448 to 0.9455 (and 0.9522 with labels).
LSTM model detects voice disorders with high accuracy.
problem Automated detection of voice disorders is challenging due to continuous audio data.
method Used Long Short Term Memory (LSTM) model for feature extraction and classification of voice disorders.
result 22% sensitivity, 97% specificity, 56% unweighted average recall.
Cheap model diagnoses vocal disorders accurately.
problem Diagnosing vocal disorders without expensive equipment.
method Used Mel-Cepstrum vectors and Support Vector Machine.
result Accurately diagnosed three vocal disorders.
This paper addresses converting speech to EGG signals without hardware, improving accuracy.
problem Estimating EGG signals from speech without hardware.
method Optimization of evidence lower bound with KL-divergence minimization.
result The method generates EGG signals that agree with gold standard and outperforms state-of-the-art.
Paper proposes singing voice conversion without parallel data.
problem Convert singing voices without parallel data.
method Phonetic posterior feature, DBLSTM, vocoder.
result Successfully converts singing voices without parallel data.
This paper converts speech to match a face image and vice versa.
problem Matching speech to a face image and vice versa.
method Proposes a model with speech converter, face encoder/decoder, and voice encoder.
result Trained model converts speech to match a face image and generates a face image that matches the voice of input speech.
Solution for voice conversion with limited data using hierarchical seq2seq and attention models.
problem Voice conversion between speakers with limited parallel audio pairs.
method Hierarchical sequence to sequence model with attention-based decoder, trained on single speaker dataset.
result Improved voice conversion quality using mel spectrograms and wavenet vocoder.
Convolutional network converts speaker voices without text.
problem Speaker conversion without text-based methods.
method Fully convolutional wav-to-wav network with ASR pre-training.
result Successfully converts TTS robot's voice to narrated audiobook voices.
Deep learning converts one singer's voice to another without supervision.
problem Unsupervised singing voice conversion.
method Deep learning network with a single CNN encoder, WaveNet decoder, and classifier.
result Natural, recognizable singing voices converted without supervision.
A fast voice conversion method using diffusion models.
problem One-shot many-to-many voice conversion.
method Diffusion probabilistic modeling with Fast Maximum Likelihood Sampling Scheme.
result Superior quality compared to state-of-the-art approaches.
Voice-controlled e-commerce app enhances user experience for visually impaired.
problem Limited user experience for visually impaired in e-commerce applications.
method Proposes a voice-controlled e-commerce application using IBM Watson speech-to-text.
result Demonstrates enhanced usability for visually impaired users.
End-to-end voice conversion without vocoder.
problem Speech conversion without vocoder.
method Transformer network for raw spectrum conversion.
result Transformer model converts real voices efficiently.
Improved voice conversion with semi-supervised learning.
problem Voice conversion with limited parallel data.
method Amortized variational inference with parallel and non-parallel utterances.
result Semi-supervised training improves voice conversion performance.
One-shot VC model converts voices without parallel data.
problem Limited VC applicability due to training data restriction.
method Disentangles speaker and content representations with instance normalization.
result Model converts voices from unseen speakers with high similarity.
AUTOVC converts voices without parallel data, achieving state-of-the-art results.
problem Non-parallel many-to-many voice conversion and zero-shot voice conversion.
method Only an autoencoder with a carefully designed bottleneck is used, training on a self-reconstruction loss.
result AUTOVC achieves state-of-the-art results in many-to-many voice conversion with non-parallel data and performs zero-shot voice conversion.
Semi-supervised singing voice separation using synthetic mixtures.
problem Singing voice separation with limited labeled data.
method Trains a single mapping function g on synthetic mixtures of singing and instrumental music.
result Performance comparable to fully supervised methods, better than semi-supervised alternatives.
New method separates multiple voices in mixed audio.
problem Separating multiple simultaneous speakers in audio.
method Gated neural networks trained at multiple steps, selecting actual number of speakers.
result Outperforms current state of the art for more than two speakers.
A new neural network separates singing voices more effectively.
problem Separating singing voices from mixed signals with high accuracy.
method MBR-FCN that processes different frequency bands with varying resolutions and filters.
result The MBR-FCN achieves better performance with fewer parameters.
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.
Privacy-preserving method protects user speech data from cloud services.
problem Privacy compromise in cloud-based speech analysis.
method Collects and sanitizes speech data before sharing, using transformation functions and voice conversion.
result Identification of sensitive emotional state reduced by ~96%.
Wave-U-Net with MHE regularization improves singing voice separation.
problem Singing voice separation from mixed music recordings.
method Wave-U-Net architecture with MHE regularization applied to 1D filters.
result Adding MHE regularization to the loss function consistently improves singing voice separation.
Improved autoencoder for F0-consistent voice conversion.
problem Non-parallel many-to-many voice conversion with prosodic information leakage.
method Conditional autoencoder with information-constraining bottlenecks.
result Controlled F0 contour and improved speech quality.
Localization of chest pathologies in chest X-ray images is a challenging task because of their varying sizes and appearances. We propose a novel weakly supervised method to localize chest pathologies using class aware deep multiscale feature learning. Our method leverages intermediate feature maps from CNN layers at di…
Proposes a new voice conversion model that preserves pitch patterns.
problem Preserving pitch patterns while changing speaker identity.
method Variational-autoencoder-based model with an auxiliary network.
result Ensures the conversion result correctly reflects specified F0/timbre information.
This study investigates how much knowledge from natural images can be transferred to pathology images.
problem Quantifying how much knowledge from natural images can be transferred to pathology images.
method Proposes a framework to quantify knowledge gain by a particular layer, conducts empirical investigation in pathology image centered transfer learning.
result Early layers of deep models can transfer knowledge to pathology image classification tasks.
Study classifies pathology reports using TF-IDF features and machine learning.
problem Classifying pathology reports for cancer surveillance and diagnostic workflow.
method Extracted TF-IDF features from pathology reports and classified them using SVM, XGBoost, and Logistic Regression.
result XGBoost achieved 92% accuracy in classifying pathology reports.
Blow converts non-parallel raw audio voices efficiently.
problem Voice conversion with non-parallel data.
method Single-scale normalizing flow with hypernetwork conditioning.
result Blow outperforms existing flow-based architectures in voice conversion.
Improved speech recognition for voice assistants by analyzing speech data.
problem Reducing false triggers in speech-enabled assistants.
method Post-processing LVCSR hypothesis lattice with a Bidirectional Lattice Recurrent Neural Network (LatticeRNN).
result LatticeRNN significantly improves detection accuracy over traditional methods.
WeSinger improves singing voice synthesis with data augmentation and specialized modules.
problem Improving the accuracy and naturalness of synthesized singing voices.
method Developed a multi-singer Chinese neural singing voice synthesis system with deep bi-directional LSTM, Transformer, LPCNet, and data augmentation.
result WeSinger achieves state-of-the-art performance on the Opencpop corpus.
This paper characterizes VAE training pathologies and their effects on tasks.
problem Characterizing VAE training pathologies and their impact on downstream tasks.
method Concretely characterizing conditions for VAE training pathologies and their connection to specific downstream tasks.
result Connects VAE training pathologies to specific downstream tasks like learning compressed and disentangled representations, adversarial robustness, and semi-supervised learning.
Many-to-Many VTN improves voice conversion across multiple speakers.
problem Voice conversion across multiple speakers.
method Sequence-to-sequence learning framework with many-to-many VTN architecture.
result Improved sound quality and speaker similarity compared to baseline methods.
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.
Proposes deep learning method for GCI detection from pathological speech.
problem Detecting glottal closure instants (GCI) in pathological acoustic speech.
method Convolutional neural network with fused deep acoustic speech and linear prediction residual features.
result Significantly better than state-of-the-art methods in GCI detection.
Semi-supervised learning classifies cardiac pathology using motion features from cine MRI.
problem Classifying cardiac pathology based on motion features from cine MRI.
method Semi-supervised learning of apparent flow to generate motion features from non-segmented images.
result The model achieves 95% classification accuracy on ACDC test set.
Thanks to the growing availability of spoofing databases and rapid advances in using them, systems for detecting voice spoofing attacks are becoming more and more capable, and error rates close to zero are being reached for the ASVspoof2015 database. However, speech synthesis and voice conversion paradigms that are not…
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.
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.
Dr.VOT measures both positive and negative VOTs accurately in natural speech.
problem Accurate measurement of VOT in natural speech.
method Deep-learning model based on RNNs for structured prediction.
result Dr.VOT improves over state-of-the-art performance on VOT estimation.
Novel approach characterizes deep neural networks at initialization.
problem Characterizing the behavior of deep neural networks at initialization.
method A novel approach considering the evolution of statistical moments of signal and noise.
result Established that skip-connections in residual networks lead to well-behaved moments and no pathology.
Personal VAD detects target speaker voice activity efficiently.
problem Efficiently detect target speaker voice activity for reduced computational cost and battery usage.
method Trains a neural network conditioned on speaker embedding or verification score, outputs probabilities for three speech classes.
result Trained model with 130K parameters outperforms combined standard VAD and speaker recognition networks.
Batch normalization in the last layer reduces sharpness in wide neural networks.
problem Pathological sharpness in wide neural networks.
method Quantifying the geometry of the parameter space using Fisher information matrix and analyzing deep neural networks with random initialization.
result Batch normalization in the last layer significantly decreases pathological sharpness under specific conditions.
The paper identifies conditions for trend reversal in classification tasks.
problem Trend reversal in classification scores and dataset values.
method Algebraic conditions and numerical results for ridge regression.
result Existence of pathological regularization regimes for certain dataset conditions.
Paper introduces a new model for polyphonic music composition.
problem Creating music with multiple interwoven voices.
method Developed a coupled recurrent model using probabilistic factorization and neural network ideas.
result Trained models for single-voice and multi-voice composition on a large dataset.
A-StarGAN improves nonparallel voice conversion speed and realism.
problem Nonparallel voice conversion without parallel data.
method Augmented classifier StarGAN (A-StarGAN) for nonparallel voice conversion.
result A-StarGAN generates realistic-sounding speech quickly and efficiently.
Transformer models show distinct spectral fingerprints under voice changes.
problem Detecting architectural biases in transformer models.
method Spectral analysis of attention-induced token graphs.
result Clear architectural signatures in model fingerprints correlate with language-specific behavior.
SAPSAM trains CNNs on lung CTs with binary labels, improving CPA detection and localization.
problem Chronic Pulmonary Aspergillosis (CPA) detection and localization on CT scans using binary labels.
method Binary labels, average intensity projections, 2D RGB-like images, hierarchical CNN architectures.
result High classification accuracy, precise localization, predictive power of 2-year survival.
Study compares neural and statistical models for Parkinson's disease progression from voice data.
problem Difficult statistical analysis of longitudinal voice biomarkers due to subject correlation, small cohorts, and varied disease trajectories.
method Evaluated Neural Mixed Effects (NME), Generalized Neural Network Mixed Models (GNMMs), and semi-parametric Generalized Additive Mixed Models (GAMMs).
result GAMMs achieve stronger predictive performance and retain interpretable smooth effects and subject-level structure.
KL-regularized RL from expert demos can lead to slow, unstable learning.
problem Pathological training dynamics in KL-regularized RL from expert demonstrations.
method Empirical analysis and non-parametric behavioral reference policies.
result KL-regularized RL can be significantly improved by using non-parametric behavioral policies.