Residual covariance multivariate model with a non-gaussian response


I am currently running a couple of different models in brms that include some form of non-gaussian response. One of these is with a bivariate model with 1 Bernoulli and 1 gaussian response. The other is a zero-inflated poisson model. These models work great, and are running smoothly. The problem I have is that I would like to estimate the residual covariance between the Bernoulli response and the gaussian one in the bivariate model, and the zi and poisson part in the zip model.

I realise that brms is set up such that set_rescor = FALSE by default in models with non-gaussian (or student) responses. But is there any way around this ? I know that I can set this up directly in stan, but it would be great if this functionality could be available in the easy-format of brms (which is also running my models a lot quicker than other packages that allow for a residual covariance to be estimated so I don’t want to switch really!)

Thanks in advance for any help!