In this paper, we use several combinations of feature front-end modules and attention mechanisms to improve the performance of our speaker verification system. An updated version of ECAPA-TDNN is chosen as a baseline. We replace and integrate different feature front-end and attention mechanism modules to compare and find the most effective model design, and this model would be our final system. We use VoxCeleb 2 dataset as our training set, and test the performance of our models on several test sets. With our final proposed model, we improved performance by 16% over baseline on VoxSRC2022 valudation set, achieving better results for our speaker verification system.

In this paper, the Mandarin-English code-switching speech synthesis system has been proposed. To focus on learning the content information between two languages, the training dataset is multilingual artificial dataset whose speaker style is unified. Adding language embedding into the system helps it be more adaptive to multilingual dataset. Besides, text preprocessing is applied and be used in different way which depends on the languages. Word segmentation and text-to-pinyin are the text preprocessing for Mandarin, which not only improves the fluency but also reduces the learning complexity. Number normalization decides whether the arabic numerals in sentence needs to add the digits. The preprocessing for English is acronym conversion which decides the pronunciation of acronym.

In this paper, we proposed RepVGGRNN, which is a light weight sound event detection model. We use RepVGG convolution blocks in the convolution part to improve performance, and re-parameterize the RepVGG blocks after the model is trained to reduce the parameters of the convolution layers. To further improve the accuracy of the model, we incorporated both the mean teacher method and knowledge distillation to train the lightweight model. The proposed system achieves PSDS (Polyphonic sound event detection score)-scenario 1, 2 of 40.8% and 67.7% outperforms the baseline system of 34.4% and 57.2% on the DCASE 2022 Task4 validation dataset. The quantity of the parameters in the proposed system is about 49.6K, only 44.6% of the baseline system.

This paper will introduce the use of Automatic Speech Recognition (ASR) technology to process speech content with specific domain. We will use the Conformer end-to-end model as the system architecture, and use pure Chinese data for initial training. Next, use the transfer learning technology to fine-tune the system with Mandarin-English code-switching data. Finally, use the Mandarin-English code-switching data with a specific domain makes the final fine-tuning of the model so that it can achieve a certain effect on speech recognition in a specific domain. Experiments with different fine-tuning methods reduce the final error rate from 82.0% to 34.8%.

In this paper, we investigate how to use limited code-switching data to implement a code-switching speech recognition system. We utilize the Transformer end-to-end model to develop our code switching speech recognition system, which is trained with the Mandarin dataset and a small amount of Mandarin-English code switching dataset, as the baseline of this paper. Next, we compare the performance of systems after adding multi-task learning and transfer learning. Character Error Rate(CER) is adopted as the criterion for the system. Finally, we combined the three systems with the language model, respectively, our best result dropped to 23.9% compared with the baseline of 28.7%.

In this paper, we use domain generalization to improve the performance of the cross-device speaker verification system. Based on a trainable speaker verification system, we use domain generalization algorithms to fine-tune the model parameters. First, we use the VoxCeleb2 dataset to train ECAPA-TDNN as a baseline model. Then, use the CHT-TDSV dataset and the following domain generalization algorithms to fine-tune it: DANN, CDNN, Deep CORAL. Our proposed system tests 10 different scenarios in the NSYSU-TDSV dataset, including a single device and multiple devices. Finally, in the scenario of multiple devices, the best equal error rate decreased from 18.39 in the baseline to 8.84. Successfully achieved cross-device identification on the speaker verification system.

Sound event detection (SED) system outputs sound events and their time boundaries in audio signals. We proposed an RCRNN-based SED system with residual connection and convolution block attention mechanism based on the mean-teacher framework of semi-supervised learning. The neural network can be trained with an amount of weakly labeled data and unlabeled data. In addition, we consider that the speech event has more information than other sound events. Thus, we use the specific time-frequency resolution to extract the acoustic feature of the speech event. Furthermore, we apply data augmentation and post-processing to improve the performance. On the DCASE 2021 Task 4 validation set, the proposed system achieves the PSDS (Poly-phonic Sound Event Detection Score)-scenario 2 of 57.6% and event-based F1-score of 41.6%, outperforming the baseline score of 52.7% and 40.7%.

This paper describes our system implementation for subtask V-oc of SemEval-2018 Task 1: affect in tweets. We use multi-task learning method to learn shared representation, then learn the features for each task. There are five classification models in the proposed multi-task learning approach. These classification models are trained sequentially to learn different features for different classification tasks. In addition to the data released for SemEval-2018, we use datasets from previous SemEvals during system construction. Our Pearson correlation score is 0.638 on the official SemEval-2018 Task 1 test set.

In this paper, we describe our system implementation for sentiment analysis in Twitter. This system combines two models based on deep neural networks, namely a convolutional neural network (CNN) and a long short-term memory (LSTM) recurrent neural network, through interpolation. Distributed representation of words as vectors are input to the system, and the output is a sentiment class. The neural network models are trained exclusively with the data sets provided by the organizers of SemEval-2017 Task 4 Subtask A. Overall, this system has achieved 0.618 for the average recall rate, 0.587 for the average F1 score, and 0.618 for accuracy.