The application of deep learning to the detection of gravitational wave (GW) signals has attracted much attention in recent years, and nearly all of the existing deep learning-based GW detection algorithms are based on the convolutional neural network (CNN), which can be used to identify the GW merger moment and produce responding triggers. However, current CNN-based models are not best optimized tools in modeling the causal logic of sequential data in the time dimension to incorporate the information of the inspiral, merger and ringdown phases of the GW waveform. To address this issue we propose a deep learning model designed based on the bidirectional gated recurrent unit (BiGRU), a variant of the recurrent neural network (RNN) that is sensitive to the sequential logic of GW data. We train the neural network and implement the GW detection task by using the data obtained during the first, second and third observation runs (O1/O2/O3) of the LIGO-VIGRO scientific collaboration. We find that our BiGRU model achieves a false alarm probability (FAP) of $4.88\times {10}^{-4}$ (corresponding false alarm rate is about 1 per 18.2 hours) and maintains a detection ratio of about 89.6%. We also construct a model ensemble containing two independently trained BiGRU models and find it does not trigger any false alarm events on our testing dataset. Meanwhile, our BiGRU model not only recovers all the reported GW events of O1-O2 runs, including the unique (as of today) confident (with observed electromagnetic counterparts) GW event originating from a binary neutron star (BNS) merger (GW170817), but also achieves a detection ratio of about 80% of the O3 run. By calculating and comparing the receiver operator characteristic (ROC) curves we demonstrate the superiority of the BiGRU model over traditional CNN-based models, and show that the BiGRU model, being optimized to deal with GW waveform that sequentially contains the inspiral, merger and ringdown phases, is a potentially powerful tool to study the GW data in the future.

It has been demonstrated that interactive speech and noise modeling outperforms traditional speech modeling-only methods for speech enhancement (SE). With a dual-branch topology that simultaneously predicts target speech and noise signals and employs bidirectional information communication between the two branches, the quality of the enhanced speech is significantly improved. However, the dual-branch topology greatly increases the model complexity and deployment cost, thus limiting its practicality. In this paper, we propose UniInterNet, a unidirectional information interaction-based dual-branch network to achieve noise modeling-assisted SE without any increase in complexity. Specifically, the noise branch still receives information from the speech branch to achieve more accurate noise modeling. Subsequently, the noise modeling results are utilized to assist the learning of the speech branch during backpropagation, while the speech branch no longer receives the auxiliary information from the noise branch, so only the speech branch is required during model deployment. Experimental results demonstrate that under the causal inference condition, the performance of UniInterNet only marginally decreases compared to the corresponding bidirectional information interaction scheme, while the model inference complexity is reduced by about 75%. With comparable overall performance, UniInterNet also outperforms previous interactive speech and noise modeling-based benchmarks in terms of causal inference and model complexity. Furthermore, UniInterNet surpasses other existing competitive methods.

For time-frequency (TF) domain speech enhancement (SE) methods, the overlap-and-add operation in the inverse TF transformation inevitably leads to an algorithmic delay equal to the window size. However, typical causal SE systems fail to utilize the future speech information within this inherent delay, thereby limiting SE performance. In this paper, we propose an overlapped-frame information fusion scheme. At each frame index, we construct several pseudo overlapped-frames, fuse them with the original speech frame, and then send the fused results to the SE model. Additionally, we introduce a causal time-frequency-channel attention (TFCA) block to boost the representation capability of the neural network. This block parallelly processes the intermediate feature maps through self-attention-based operations in the time, frequency, and channel dimensions. Experiments demonstrate the superiority of these improvements, and the proposed SE system outperforms the current advanced methods.