Logs : https://pantheon.corp.google.com/storage/browser/chromiumos-test-logs/bugfiles/cr/807266

The little echo while starting the call had been observed in few other chromebooks as well.

The issue gets very Interesting. Tested in M66 Canary Build for Samus, the little echo that shows up in the beginning of the call is gone. but there is a continuous - very little echo all the time (noticeable if you listen carefully in the end of every pronunciation).  Got the Logs and Tracing Info, Added in a new folder. 

Re #6: When I look at the recording it seems like AEC3 was not on.

Its being uploaded there, u can find it in the name "Samus_AEC3_On" and indeed there is no echo observed in samus (but the other side had the little echo - Mac Book in the beginning and after a while its settled greatly ). 

Regarding #10:
Looking at Samus_AEC3_On, it basically has no signal coming from the other end, and no echo issues. Did you perceive any issues in this call from this end?

Regarding #6:
The mentioning of the tiny echo all the time. Is that captured in the recording where AEC3 was not on?

[Triage] To Per.

Yes the Tiny echo (all the time) perceived when AEC was not ON. with AEC (enabled in the chrome:flags) the echo is Gone. 

Re 13: Just double checking that by AEC you mean AEC3, right? I.e., that the tiny echo is gone when AEC3 is turned on.

Yes Per. Ravi meant AEC3 ON.

Little echo is observed to the other end user Macbook pro (64.0.3282.119). Sometimes even to the chromebook user. This is quite common in M65. This is not observed in M64.

Most of the time initial few seconds of the call have this echo issue. 

Thanks!

So from #15 I gather that:
-Low-level echo on Macbook pro is present when AEC3 is OFF in M64.
-Low-level echo on Chromebook is not present when AEC3 is OFF in M64.
-Low-level echo on Chromebook is common when AEC3 is OFF in M64.
-Low-level echo does not occur on neither Macbook pro or Chromebook when AEC3 in ON in M64.
-Low-level echo does not occur on neither Macbook pro or Chromebook when AEC3 in ON in M65.

Is that correct?

-Low-level echo on Macbook pro is present when AEC3 is OFF in M64 - Yes.
-Low-level echo on Chromebook is not present when AEC3 is OFF in M64 - No, Echo is present when AEC3 is OFF.
-Low-level echo on Chromebook is common when AEC3 is OFF in M64 - Yes, easy to reproducible while start talking together. 
-Low-level echo does not occur on neither Macbook pro or Chromebook when AEC3 in ON in M64 - sometimes happens when AEC3 On too, but not always reproducible. But when start talking together (double talk) for a while, echo shows up irrespective of AEC3 On or OFF.
-Low-level echo does not occur on neither Macbook pro or Chromebook when AEC3 in ON in M65 - No, in M65 issue still present. (in M66 Canary build, Its not present when AEC3 ON).

Added another folder Called Pit with AEC3.

Clear echo heard in Mac Book Pro in the evaluation room (where Pit is leaking Echo). 
Pit is with AEC3 On. With 3.5mm audio jack. 

https://pantheon.corp.google.com/storage/browser/chromiumos-test-logs/bugfiles/cr/807266/Pit%20-%20with%20AEC3

Ping!

Ping!

I've re-evaluated the recordings on the AEC3 and AEC2 ToT and my view is that
-AEC2 leaks echoes in some of the recordings. I don't think that is a regression in the AEC2 behavior though.
-AEC3 does not leak any echoes.
-AEC3 has some few segments of lower transparency compared to AEC2, which is caused by the fact that the echo path is unknown and AEC3, in contrast to AEC2, handles that by not leaking echoes until the echo path is estimated.

I'm not really sure of whether there is anything left to on this issue. The echo leakage I see is now handled by AEC3. 

I'm closing the issue as fixed now as the AEC3 seem to solve any issues present in AEC2 for these cases, but please feel free to reopen it if you think it isn't yet properly fixed.