In language learning there's something called the "Input hypothesis" which states that we learn languages primarily by reading and listening (consuming input), rather than production of language.

The reasoning is that when you produce output you are by definition producing something at your own level, so little improvement occurs. When you consume input (and transcription is a great way to do it), you are consuming input of a native speaker - or in the case of Jazz, a master player - which is at a much higher quality level than what you could produce yourself. So input drives the learning process.

The exception is for motor e.g. pronunciation or playing technique, which do benefit from practice.