The general idea is that echoic memory has evolved to last longer than iconic memory due to availability of source. For example, let's say you glance at an object. Unless that object moves very quickly, you would have if you needed to the chance to look at it again. You might choose not to but your requirement to store that image is less because it is likely to be available if you require to refresh. Sound isn't likely to work like that. If you hear a dog bark, then you don't know if it will bark again. You cannot control the ability to refresh that information so therefore your brain has been wired to retain the sound for longer - just in case. In terms of sensory memory, 3-4 seconds is a very long time.
Due to the nature of the memory test here, we don't currently measure your echoic memory performance. Indeed, even if we could it would not measure echoic memory as such as the sounds would not stay in your sensory memory for long enough. Any test of sound or speech is likely to actually test other forms of memory rather than the actual sound itself.