I understand what you are saying and when someone is speaking quickly a lot of French is going to get lost in the shuffle. But in these exercises, you can slow it down and listen to one word at a time. So for learning purposes, you can listen for that "n" sound in "viennent" until you can hear it. In a live conversation, it may easily be missed.
It's subtle for native English speakers. We do nasalize the vowel preceding 'n' when when the 'n' precedes a 'g' ("sing", "wrong") or 'k' ("think"), but nasalization is not phonemic in English.
The two are SUPPOSED to sound different. My point was that I can't hear the lack of nasalization in the full-speed version, and of course the "s" on the end of "elles" isn't pronounced, so that's no help when it's purely an audio exercise. It's all run together so quickly. In real-life conversation, presumably there would be some context so that it would be clear which was meant, no matter if the person was speaking quickly or not. When I ran into that sentence again and played the slow version, it was more obvious. I reported it for lack of audio clarity. Meanwhile, I'll just remember that it's meant to be singular!
There's nothing they can do about the crappy audio quality on many of these exercises. They use a third party text to speech service over which they have no control, and I have read they have no plans to change it. It's a shame because some of the audio examples are of truly atrocious audio quality.
They've gotten better audio (the new language lessons like Japanese and the male voice for French use much better audio). Just sometimes it chooses to use the inferior audio quality for no reason.
Though I will continue to give them negative feedback until it's changed entirely pretty much.
But uh, in this case there totally is a difference between Viens and Viennent unless it's getting distorted. They're completely different vowel sounds.
My point was that I couldn't tell which was meant in this specific exercise, with no context, because it was spoken so quickly and the words were run together. I took a shot at it and I was wrong. I didn't try the slowed-down audio at first. When I did, it was quite obvious indeed. I have a lot of trouble understanding French radio, etc. at normal native-speaker speed. It's my weakest skill.