d-prime ~ version + (1|participant)

fit = glm(
  data = my_data,
  formula = response ~ reality*condition,
  family = binomial)

response ~reality * version + (1|participant) + (1|melody)