Deep coal mining is increasingly exposed to coal bursts and associated secondary hazards such as roof collapse and spalling. Coal bursts occur when highly stressed coal is subjected to dynamic disturbances, triggering a sudden release of stored strain energy; understanding the dynamic properties of coal under coupled static-dynamic loading is therefore essential. This paper investigates the dynamic behaviour of bedded coal from a burst-prone mine, considering bedding angles of 0°, 30°, 45°, 60° and 90°. Dynamic biaxial compression tests were conducted on cubic specimens using a triaxial Hopkinson bar system with high-speed imaging, at impact velocities of 5-20 m/s. The bursting process was found to be governed by bedding orientation. Repeated-impact tests were then performed, with ultrasonic measurement and synchrotron X-ray CT after each impact to track damage accumulation. With successive impacts, peak stress and P-wave velocity decreased, and the stress-strain response shifted from Class I to Class II behaviour. The number of impacts to failure depended strongly on bedding angle, being lowest at 30°-60°. Tensile cracking dominated at 0° and 90°, whereas mixed tensile-shear cracking developed at intermediate angles, with cracks initiating along bedding planes. Three-dimensional reconstruction showed that crack volume grew slowly at first and then rapidly near failure. To capture fracturing in real time, an acoustic emission (AE) methodology was developed for split Hopkinson pressure bar tests. Signal unreliability, saturation and impact noise from the dynamic stress waves were addressed by using lightweight, wide-band piezoelectric sensors, an attenuator, and 80 kHz high-pass filtering based on the spectral characteristics of the impact noise. The AE system was synchronised with high-speed cameras, allowing AE signals to be compared with full-field strain measurements during dynamic tensile tests. Few AE events were detected before peak stress, and high-frequency components appeared only near the peak. A sharp jump in cumulative AE energy coincided with the onset of crack growth observed in the strain fields, providing a reliable indicator of crack initiation. AE counts and duration further distinguished failure modes, with tensile failure producing lower values than mixed tensile-shear failure. These results clarify how bedding orientation controls damage accumulation and failure of coal under dynamic loading, and show that AE monitoring can identify crack initiation and failure mode in high-rate tests.