|
My worry about this type
of work is that the speech (with added reverb or noise) is often played back to
the listener through a single loudspeaker, pair of loudspeakers or over
headphones. I am not convinced that by presenting the listener with such an
unrealistic sound field and a 'fake' source signal (by which I mean adding
reverb on to the speech signal after it has been recorded) that we can really be
confident in our conclusions of what they can/can't hear.
What do other people
think about this type of testing?
|