Self-supervised learning (SSL) leverages large datasets of unlabeled speech to reach impressive performance with reduced amounts of annotated data.
They are usually based on a message-passing mechanism and have gained increasing popularity for their intuitive formulation, which is closely linked to the Weisfeiler-Lehman (WL) test for graph isomorphism to which they have been proven equivalent in terms of expressive power.
Despite the recent advancements in speech recognition, there are still difficulties in accurately transcribing conversational and emotional speech in noisy and reverberant acoustic environments.
Self-supervised learning (SSL) has recently allowed leveraging large datasets of unlabeled speech signals to reach impressive performance on speech tasks using only small amounts of annotated data.
Our principal innovation in this work is to use Simulated annealing in EArly Layers (SEAL) of the network in place of re-initialization of later layers.
In this paper, we introduce a new approach, called Posthoc Interpretation via Quantization (PIQ), for interpreting decisions made by trained classifiers.
Self-supervised learning (SSL) has allowed substantial progress in Automatic Speech Recognition (ASR) performance in low-resource settings.
End-to-end speech synthesis models directly convert the input characters into an audio representation (e. g., spectrograms).
In this paper, we work on a sound recognition system that continually incorporates new sound classes.
In particular, we extend our previous findings on the SepFormer by providing results on more challenging noisy and noisy-reverberant datasets, such as LibriMix, WHAM!, and WHAMR!.
Ranked #1 on Speech Enhancement on WHAM!
Although deep learning (DL) has achieved notable progress in speech enhancement (SE), further research is still required for a DL-based SE system to adapt effectively and efficiently to particular speakers.
First, we release the REAL-M dataset, a crowd-sourced corpus of real-life mixtures.
Most of the deep learning-based speech enhancement models are learned in a supervised manner, which implies that pairs of noisy and clean speech are required during training.
This result is consistent with recent neuroscience studies on emotion recognition, which found an association between these band suppressions and the behavioral deficits observed in individuals with ASD.
4 code implementations • 8 Jun 2021 • Mirco Ravanelli, Titouan Parcollet, Peter Plantinga, Aku Rouhe, Samuele Cornell, Loren Lugosch, Cem Subakan, Nauman Dawalatabad, Abdelwahab Heba, Jianyuan Zhong, Ju-chieh Chou, Sung-Lin Yeh, Szu-Wei Fu, Chien-Feng Liao, Elena Rastorgueva, François Grondin, William Aris, Hwidong Na, Yan Gao, Renato de Mori, Yoshua Bengio
SpeechBrain is an open-source and all-in-one speech toolkit.
The discrepancy between the cost function used for training a speech enhancement model and human auditory perception usually makes the quality of enhanced speech unsatisfactory.
Ranked #11 on Speech Enhancement on VoiceBank + DEMAND
This paper introduces Timers and Such, a new open source dataset of spoken English commands for common voice control use cases involving numbers.
Ranked #4 on Spoken Language Understanding on Timers and Such (using extra training data)
Learning robust speaker embeddings is a crucial step in speaker diarization.
In this work we explore a way in which the Transformer architecture is deficient: it represents each position with a large monolithic hidden representation and a single set of parameters which are applied over the entire hidden representation.
Transformers are emerging as a natural alternative to standard RNNs, replacing recurrent computations with a multi-head attention mechanism.
Ranked #5 on Speech Separation on WSJ0-2mix
In this paper, we propose to capture these inter- and intra- structural dependencies with quaternion neural networks, which can jointly process multiple signals as whole quaternion entities.
We then propose a revised encoder that better learns short- and long-term speech dynamics with an efficient combination of recurrent and convolutional networks.
End-to-end models are an attractive new approach to spoken language understanding (SLU) in which the meaning of an utterance is inferred directly from the raw audio without employing the standard pipeline composed of a separately trained speech recognizer and natural language understanding module.
Ranked #7 on Spoken Language Understanding on Snips-SmartLights
Using the Wall Street Journal Dataset, we compare our phase-aware loss to several others that operate both in the time and frequency domains and demonstrate the effectiveness of our proposed signal extraction method and proposed loss.
Building on recent advances, we propose a new deep complex-valued method for signal retrieval and extraction in the frequency domain.
Whereas conventional spoken language understanding (SLU) systems map speech to text, and then text to intent, end-to-end SLU systems map speech directly to intent through a single trainable model.
Ranked #14 on Spoken Language Understanding on Fluent Speech Commands (using extra training data)
Learning good representations without supervision is still an open issue in machine learning, and is particularly challenging for speech signals, which are often characterized by long sequences with a complex hierarchical structure.
Ranked #2 on Distant Speech Recognition on DIRHA English WSJ
Deep neural networks can learn complex and abstract representations, that are progressively obtained by combining simpler ones.
Mutual Information (MI) or similar measures of statistical dependence are promising tools for learning these representations in an unsupervised way.
Deep learning is currently playing a crucial role toward higher levels of artificial intelligence.
Ranked #3 on Distant Speech Recognition on DIRHA English WSJ
Neural network architectures are at the core of powerful automatic speech recognition systems (ASR).
Experiments, that are conducted on several datasets and tasks, show that PyTorch-Kaldi can effectively be used to develop modern state-of-the-art speech recognizers.
Ranked #1 on Distant Speech Recognition on DIRHA English WSJ
Rather than employing standard hand-crafted features, the latter CNNs learn low-level speech representations from waveforms, potentially allowing the network to better capture important narrow-band speaker characteristics such as pitch and formants.
Recurrent neural networks (RNNs) are powerful architectures to model sequential data, due to their capability to learn short and long-term dependencies between the basic elements of a sequence.
Distant speech recognition is being revolutionized by deep learning, that has contributed to significantly outperform previous HMM-GMM systems.
A field that has directly benefited from the recent advances in deep learning is Automatic Speech Recognition (ASR).
Ranked #6 on Speech Recognition on TIMIT
Deep learning is an emerging technology that is considered one of the most promising directions for reaching higher levels of artificial intelligence.
The availability of realistic simulated corpora is of key importance for the future progress of distant speech recognition technology.
Audio and Speech Processing Sound
Despite the significant progress made in the last years, state-of-the-art speech recognition technologies provide a satisfactory performance only in the close-talking condition.
This paper introduces the contents and the possible usage of the DIRHA-ENGLISH multi-microphone corpus, recently realized under the EC DIRHA project.
First, we suggest to remove the reset gate in the GRU design, resulting in a more efficient single-gate architecture.
Improving distant speech recognition is a crucial step towards flexible human-machine interfaces.
Despite the remarkable progress recently made in distant speech recognition, state-of-the-art technology still suffers from a lack of robustness, especially when adverse acoustic conditions characterized by non-stationary noises and reverberation are met.
This paper describes a multi-microphone multi-language acoustic corpus being developed under the EC project Distant-speech Interaction for Robust Home Applications (DIRHA).