This is not your sport. Your sport is real conversation. This is your accessory work — and a hypothesis about what it builds.
First, the honest label: dictation may not produce any results unless accompanied by conversational lectures. However — it empowers you to participate in conversational lectures, because at least half the battle is just listening.
Now the hypothesis: understanding is merely compression of listening. The brain only "understands" to reduce its own cognitive burden when listening to strings of syllables — to get what your lecturer, and every other speaker of the language, is saying. And it is only prompted to understand when faced with an enormous volume of information to compress: chains and chains of syllables. No volume, no compression. This is why the vocabulary you can produce on a quiz still turns to noise in the air.
Once the brain accepts that you will be doing this dictation, it starts to "understand" just to make its own life easier — to take shortcuts.
The shortcuts have an order. First it notices the accords between adjectives and nouns — gender, number, in Romanian sometimes also the case, genitive or dative. Agreeing endings become predictable, and predictable is cheap. Then it goes beyond the accords: word boundaries where there used to be blur, whole constructions arriving pre-parsed. You don't decide any of this. You supply the volume; the brain, lazy like every efficient system, does the compressing. (Native speech is a pro serve; this is why standing in the box works.)
Which is why this is a program, not an app. Strength doesn't go from beginner to intermediate in one workout, and no diet that works is over in a week — no diet nor any fitness regime is independent of scheduling. Six months at least. Honestly: two years. Do your sport — take the lectures, refuse the slowed-down version. This is the accessory work that gets you on the platform.