10MB/minute would be for a 16bit, 44.1KHz STEREO file.
A speech recording would most likely be a MONO file, and probably recorded at a lower sampling rate, maybe as low as 11.025KHz, and at only 8bit - and still be good enough to capture voice.
At that rate, it could be 645KB per minute or about 97MB for a 150 minute file.
I can't see going much lower than that though, and if your file is only 36.5MB to start with, it sounds like there is already some compression.
Sometimes compressed files subjected to another compression algorithm can actually grow larger.