This article addresses the challenges of water scarcity and climate change faced by cotton cultivation in the Xinjiang region of China. In response, a precise irrigation model based on reinforcement learning and the crop model DSSAT is proposed. The experimental site chosen for this study is Changji City in northwest China’s Xinjiang province. Integrating the cotton model, CSM-CROPGRO, from the DSSAT model with reinforcement learning algorithms, a decision system was developed to provide accurate irrigation strategies that maximize cotton yield. The experimental results demonstrated that our approach significantly improved cotton yield and, compared to genetic algorithms, reduced water consumption while increasing production. This provides a better solution for developing cotton cultivation in the Xinjiang region. Additionally, we analyzed the differences in irrigation strategies among different decision scenarios, and the results showed that the reinforcement learning method achieved higher yields in water application trends during different periods. This research offers new ideas and methods for improving cotton-crop-management decisions. The study’s focus on maximizing cotton yield while reducing water usage aligns with the sustainable management of water resources and the need for agricultural adaptation to changing climate conditions. It highlights the potential of reinforcement learning methods in improving irrigation decision-making and their applicability in addressing water scarcity challenges. This research contributes to the advancement of cotton crop management and provides valuable insights for agricultural decision-makers in the Xinjiang region and beyond.

Phonetic information is one of the most essential components of a speech signal, playing an important role for many speech processing tasks. However, it is difficult to integrate phonetic information into speaker verification systems since it occurs primarily at the frame level while speaker characteristics typically reside at the segment level. In deep neural network-based speaker verification, existing methods only apply phonetic information to the frame-wise trained speaker embeddings. To improve this weakness, this paper proposes phonetic adaptation and hybrid multi-task learning and further combines these into c-vector and simplified c-vector architectures. Experiments on National Institute of Standards and Technology (NIST) speaker recognition evaluation (SRE) 2010 show that the four proposed speaker embeddings achieve better performance than the baseline. The c-vector system performs the best, providing over 30% and 15% relative improvements in equal error rate (EER) for the core-extended and 10 s–10 s conditions, respectively. On the NIST SRE 2016, 2018, and VoxCeleb datasets, the proposed c-vector approach improves the performance even when there is a language mismatch within the training sets or between the training and evaluation sets. Extensive experimental results demonstrate the effectiveness and robustness of the proposed methods.

Recently, the time-domain-based methods (i.e., the method of modeling the raw waveform directly) for audio source separation have shown tremendous potential. In this paper, we propose a model which combines the complexed spectrogram domain feature and time-domain feature by a cross-domain encoder (CDE) and adopts the hierarchic temporal convolutional network (HTCN) for multiple music sources separation. The CDE is designed to enable the network to code the interactive information of the time-domain and complexed spectrogram domain features. HTCN enables it to learn the long-time series dependence effectively. We also designed a feature calibration unit (FCU) to be applied in the HTCN and adopted the multi-stage training strategy during the training stage. The ablation study demonstrates the effectiveness of each designed component in the model. We conducted the experiments on the MUSDB18 dataset. The experimental results indicate that our proposed CDE-HTCN model outperforms the top-of-the-line methods and, compared with the state-of-the-art method, DEMUCS, achieves the improvement of the average SDR score of 0.61 dB. Significantly, the improvement of the SDR score for the $\ bass$ source has a sizable margin of 0.91 dB.

Drought stress disrupts the molecular-level water balance in plants, and severe water deficiency can be fatal for cotton plants. However, mild water deficits or short-term drought stress may enhance crop resilience, increasing yields. The present study aims to determine the optimal watering time and irrigation amount to induce drought tolerance in cotton seedlings during drought training. Specifically, the investigation focuses on identifying the ideal day for watering and the corresponding irrigation volume that effectively triggers the transition of cotton plants into a state of enhanced resistance to drought stress during the seedling stage. In this study, the CSM-CROPGRO-Cotton model was utilized, and our objectives were to (i) evaluate the predictive capability of CSM-CROPGRO-Cotton for yield estimation in field experiments in Xinjiang and (ii) simulate and assess the range of time during the seedling stage when cotton plants can withstand drought stress without reducing yields, identifying irrigation strategies that induce drought training while maintaining yield under mild water deficiency. The model was validated using yield data from field experiments conducted in 2023. The validation criteria included a normalized root mean square error $\left(nRMSE\right)>10\%$ and a coefficient of determination $\left(r^2\right)>85\%$ for yield; for the leaf area index (LAI), the criterion was $\left(r^2\right)>90\%$, with a degree of agreement of $\left(d\right)>75\%$. The results demonstrated the accuracy of the CSM-CROPGRO-Cotton model in predicting cotton yield. Based on the validated CSM-CROPGRO-Cotton model, this study employed the LINUX crop model batch-processing technique to efficiently simulate 357 different irrigation strategies by adjusting the amount of “first irrigation” and timing. The findings revealed that in the irrigation scheme for cotton during the seedling stage, when the amount of first irrigation was in the lower range of 10 mm to 15 mm, the cotton plants underwent drought training during the early growth stage, and their yields did not exhibit drastic fluctuations due to reduced amounts of first irrigation. The suitable period for first irrigation for drought training was from 25 June to 6 July, and the amount of first irrigation could save approximately 57.14% in irrigation water. This implies that subjecting cotton plants to a certain level of drought training can enhance their stress tolerance and increase yields. This finding holds great significance for cotton cultivation in drought-prone regions.