In this post I brought probable proof that Rava either did not know that the letter ב has a hard and soft pronunciation, meaning that it did not yet have the dual pronunciation of /b/ and /v/ or if he did then the difference to him was far slighter than the /b/ /v/ difference to us. Perhaps this would be akin to the difference between a properly pronounced ח and ה (/h/ and /h/) which is slighter than the difference between a כ and ה (/kh/ and /h/). Of course the difference between the two kinds of bets might have been even smaller.
However in the comments Steg and especially Kylopod argued that this needn't be the case at all.
I am not yet convinced they are right, but they argued it well enough that it deserves mentioning as a third possibility, which is this:
Even if the distinction had been fairly wide it is possible that Rava was unaware of it even though he naturally made a distinction himself. Not only that, the distinction may have been our own /b/ /v/.
Let's explain. First, we are obviouyly assuming that at that time the dagesh notation didn't yet exist, which means that there was only one way to write a ב. The question is only if there were two kinds of בs as there are now, one we write with a dot and one without, but they wrote them both as ב, yet pronouncing them two ways--perhaps /b/ and /v/.
The point is that most people are unaware of the actual sounds in their language because written language is a particularly poor way to capture the sounds of spoken language. The example Kylopod gave, which I've seen elsewhere is the /s/ sound in the word dogs. Most people are unaware until it is pointed out to them that the /s/ in dogs sounds like /z/. It is certainly different from the /s/ sound in snake.
Suppose there were no letter z at all, but the s in dogs still sounded like /z/. Most people wouldn't realize that there is a sound they make which can't be written. An example of this would be the /zh/ sound in vision. Now, it is true that by writing /zh/ you probably know what I mean, but strictly speaking that sound (written as /j/ in French, as in deja vu) has no English equivalent. As a sound it exists in English, obviously, but is written with an s (vision, lesion &c).
What am I getting at?
It's possible that at the historical stage of the game where Rava was that the ב already had a dual pronunciation which was influenced by the surrounding vowels, much as our own is. Yet Rava need not have been particularly aware of it. In fact it wasn't until linguists examined the language that it was noticed. These linguists mapped out the language and they were the ones who added nekudot, including the dagesh, to mark the discoveries they made about Hebrew. It would be like a contemporary linguist first noticing that sometimes an s is pronounced like it is in hiss, sometimes as a z as in dogs and sometimes like zh as in vision.
Before anyone protests that 1) Rava certainly was an expert in Hebrew and 2) how could he have not noticed what is so obvious to us, let me point out two things:
One could certainly imagine someone saying "Don't slur the words dog's sled together," even though we are talking about the /z/ /s/ sound and not an /s/ /s/.
What is obvious is not obvious until it is discovered. I assume most people who are aware of the triliteral root theory of Hebrew think it is obvious, but it wasn't discovered until about a thousand years ago.
All in all Rava may have cautioned against slurring the words esev besadecha together thinking that the ב sound was the same at the end of the first word and beginning of the second even though it wasn't.
On the other hand he may have been fully aware that they had two separate sounds, but lacking the visual cue from the dagesh had not been predisposed, as we are, to almost think of the ב as two separate letters.
In short, this example needn't be the historical linguistic clue that it would be so cool if it was.