In a previous work, we showed that a kind of ω-automata known as Transition-based Generalized Testing Automata (TGTA) can outperform the Büchi automata traditionally used for explicit model checking when verifying stutter-invariant properties. In this work, we investigate the use of these generalized testing automata to improve symbolic model checking of stutter-invariant LTL properties. We propose an efficient symbolic encoding of stuttering transitions in the product between a model and a TGTA. Saturation techniques available for decision diagrams then benefit from the presence of stuttering self-loops on all states of TGTA. Experimentation of this approach confirms that it outperforms the symbolic approach based on (transition-based) Generalized Büchi Automata.