Quantum states of optical pulses can be controlled with great accuracy, and in solid state precise control over quantum states of spins has been achieved. Conversion of such optical quantum states into spin quantum states, and vice versa, is highly relevant for quantum information science, but appears to be more challenging. This proposal aims at a study of such quantum-state conversion, and at developing it into a fast and robust technique with high fidelity. The project builds on the established idea that spins in semiconductors provide a promising system for such research, but pioneers two innovations. A key ingredient is to use ensembles of electron spins, instead of the more widely studied case of an individual spin in a quantum dot. Using ensembles gives access to strong interaction between spins and highly-directional optical fields in a robust manner, without a need for high-finesse cavities. This is vital for the second innovation, which is to use projective measurement with quantum optical techniques as a robust tool for preparing very pure correlations between quantum states of spins and optical pulses. The correlations originate from spin-flip Raman transitions at the single photon level, and this approach also gives access to studying entanglement between spins that are separated by a large distance.During this project, the same quantum optical techniques will be used for time-resolved probing of the loss of quantum coherence, and initialization experiments aimed at preparing the solid-state environment in a state that yields longer spin coherence times. The experiments use donor-bound electron spins in GaAs where spin decoherence is mainly due to interaction with fluctuating nuclear spins in the host lattice. The project includes a new approach for pumping these fluctuations away. Work with donor-bound electrons in ZnSe, where the lattice has nuclear spin zero, explores how the single nuclear spin of the donor can act as a long-lived quantum memory.