The same speech sounds (phones) produced by different speakers can sometimes exhibit significant differences. Therefore, it is essential to use algorithms compensating these differences in ASR systems. Speaker clustering is an attractive solution to the compensation problem, as it does not require long utterances or high computational effort at the recognition stage. The report proposes a clustering method based solely on adaptation of UBM model weights. This solution has turned out to be effective even when using a very short utterance. The obtained improvement of frame recognition quality measured by means of frame error rate is over 5%. It is noteworthy that this improvement concerns all vowels, even though the clustering discussed in this report was based only on the phoneme a. This indicates a strong correlation between the articulation of different vowels, which is probably related to the size of the vocal tract.

This paper presents a novel method to overcome problems of finite set-model-based predictive torque control (MPTC) which has received a lot of attention in the last two decades. Tuning the weighting factor, evaluating a large number of switching states in the loop of the predictive control, and determining the duty cycle are three major challenges of the regular techniques. Torque and flux responses of deadbeat control have been developed to overcome these problems. In our method, firstly, the prediction stage is performed just once. Then, both the weighted cost function and its evaluation are replaced with only simple relationships. The relationships reduce torque ripple and THD of stator current compromisingly. In the next step, the length of the virtual vector is used to determine the duty cycle of the optimum voltage vector without any additional computations. The duty ratio does not focus on any relation or criteria minimizing torque or flux ripple. As a result, torque and flux ripples are reduced equally. The proposed duty cycle is calculated by using a predicted virtual voltage vector. Hence, no new computation is needed to determine the proposed duty cycle. Simulation and experimental results confirm both the steady and dynamic performance of the proposed method in all speed ranges.