Controlled multichannel room impulse response (RIR) simulation supports the analysis and evaluation of coherence-sensitive microphone-array methods under explicitly stated modeling assumptions. Two late-field properties are relevant: the energy-decay rate (reverberation time) and the inter-channel coherence of the diffuse tail. This paper presents an efficient hybrid multichannel RIR generator that jointly controls a prescribed reverberation time, measured on the assembled early-plus-late response, and a target late-field inter-channel coherence structure consistent with an isotropic diffuse-field model. Early reflections are generated using an image-source method (ISM), while late reverberation is synthesized as a bandwise decaying-noise process energy-matched to the early component at the transition. Late-field spatial statistics are imposed bandwise by Cholesky coloring of the stochastic tail using the isotropic diffuse-field coherence matrix, and reverberation time is controlled through a self-calibration procedure that tunes a scalar tail multiplier via bracketed bisection based on the Schroeder-estimated reverberation time. Monte Carlo experiments show that the proposed generator closely realizes the prescribed targets, with lower target-consistency errors than ISM and hybrid baselines lacking joint control. The method is further assessed through comparison with measured multichannel RIRs, quantifying the discrepancy between the imposed diffuse-field statistics and those of measured rooms and assessing downstream direction-of-arrival (DOA) localization with steered-response power with phase transform (SRP-PHAT) on measured and matched synthetic responses. The resulting framework provides a computationally efficient generator of controlled multichannel RIR conditions for model-based analysis of coherence-sensitive array methods under explicit diffuse-field assumptions.
