I tested using a Nexus 5X with the same version of Chrome.  It works for me. I hear 3+ sec of audio.

I reproduced it on 2 phones that I have around - Motorola Moto G and Highscreen Power Rage, both Android 5 something.

If I'm reading the code right, you're creating a stereo convolver with an impulse response of length 4 sec.  At a sample rate of 44100 kHz (fairly typical for an Android phone).  That's a huge convolver, and doubly complex because there are two impulse responses.

It's possible the Moto G and Power Rage are not powerful enough to perform the necessary convolutions.  You can easily test by making the impulse response shorter.

An ancient Galaxy Nexus plays sounds but is very glitchy.  Drop outs throughout the clip, but audio is heard for more than 3 sec.

Made it shorter and the sound did not end that quick but still fairly quick. So that's a performance issue? How come it works on iPhone 4s from 5 years ago with 512MB RAM and dual-core 800MHz and not on 2GM RAM quad-core 1300MHz Android 5 device?

Did some more testings and yes, seems like performance issue. That's too bad, seems that convolution effects are not yet within the reach of mid-range handhelds, will have to cut those effects out of my app. The bug could probably be closed.

The bulk of the processing is in computing FFTs.  Chrome has NEON SIMD optimized FFT routines, but only for devices that have NEON. I don't know if your test phones have NEON.  It's very possible iPhone 4s has NEON and that the proprietary Apple FFT performs much better than the one used in Chrome on Android.

There are some further optimizations that can be done on Android that haven't been yet, but the FFT in this case is the biggest issue.

Thanks for testing.

I'll close this for now, but if you find some new information or tests, feel free to reopen this or file a new issue.

Closing (WAI).