 EMG-based motion estimation is important for applications such as myoelectric control, where the simultaneous estimation of kinematic information, namely joint angle and velocity, is challenging and critical. To address this challenge, we proposed a novel method which combines two streams of convolutional neural networks, CNN, to learn informative features from raw electromyography, EMG, data and accurately estimate the generated motion under three different conditions, isotonic, isokinetic, quasi-dynamic and fully dynamic. Experimental results showed that our method outperformed other state-of-the-art approaches in terms of accuracy and robustness. In particular, our method achieved better performance than existing methods when estimating joint angles or velocities under isokinetic conditions. Furthermore, our method was able to achieve comparable performance with existing methods when estimating both joint angles and velocities under dynamic conditions. We are article.tv, links in the description below.