Recent works in cognitive load theory (CLT) have shown that working memory resources can be depleted during a learning task (Chen, Castro-Alonso, Paas, & Sweller, 2018; Leahy, & Sweller, 2019). Sweller and colleagues found that students performed better at a working memory test after a less demanding task than when the same test was completed immediately after a very demanding and time-consuming task. These authors thus introduce a new, dynamic variable in CLT: the depletion of working memory resources. Our aim is to investigate the working memory resources depletion effect by considering the Time-Based Resources Sharing model (Barrouillet & Camos, 2015). The TBRS model allows simple and straightforward predictions about cognitive load in working memory, as well as about the solicited resource (controlled attention). According to this model, a learning task involving the sharing of controlled attention between the processing devoted to refreshing memory traces and the processing devoted to new items, should solicitate controlled attention. If this task is long and demanding, then this resource should be depleted. Our general assumption is that it is possible to measure this depletion with a working memory test, just before and just after the learning task. If performance at the second working memory test is lower than the performance at the first working memory test, it is because resources have been depleted. In previous experiments, we failed to obtain the working memory resource depletion effect with a long transcription task (Vié et al., 2021) and with a short dictation task (Brellier & Tricot, 2021). In the present experiment we tried once again, with a long and demanding dictation task.