Because they don't differentiate between P and B, this causes them to fluidly float between either when expressing a B or P without even knowing they are doing it since the mouth is in the same position. The only difference between a B and a P is aspiration, which they don't notice because their language doesn't distinguish the aspirated versus unaspirated versions of this sound.
Japanese, Koreans, and Chinese do the same thing with R and L; because those Asian languages do not differentiate between R and L. Because R in those languages is made with the tongue very close to where it will be when saying L, they will freely use either without differentiating them. If you want to hear a good example of that in Japanese, listen to some songs by Ikimono Gakari. The lead singer often uses both R and L interchangeably in her singing. Here are some examples:
Listen to how she says sakura in this video at 30 seconds for example. It sounds a lot like sakula. Also how she says kara at 59 seconds. It sounds like kala. But a few seconds later, she says futari and haru with R. If you are speaking Japanese, you can freely substitute L for R and it will sound fine but perhaps a bit feminine but really it will sound fine. If you want to sound like a macho tough guy, roll your R's really hard. Actually, Most Americans who can't say the correct Japanese R should probably just say L instead as their Japanese will sound much better to native Japanese speakers. https://www.youtube.com/watch?v=U8_53jIMDV0
On the CD version of her singing Kira Kira Train, her use of L when you usually hear R is very pronounced. You can kind of hear it in her live stage singing of the song but it isn't as obvious as it is in the albums probably because in the studio there isn't background noise and she is singing very cleanly and purely and not trying to compete with other noise. That is a great album so if you like their music, I recommend buying any of their songs anyway if you like Japanese.
In sum, I am trying to say that when a language does not differentiate between two sounds that are articulated with only slight differences, the native speakers will usually float between both/all sounds. They themselves will often not be able to even hear the difference between them unless they study another language that does differentiate them.