Since the early 1990s and for the 13 years that followed I was in charge of specifying and managing the design and development of voice-quality enhancement algorithms and systems in Lucent Technologies, and later in NMS Communications. When taking on the task, I was astonished by the secrecy surrounding the make-up of the minute algorithm details that separated the exceptional from the second-rate performers; the ones that elevate the performance of the voice-quality algorithms to significant heights versus their counterparts, which adhere to standard textbook and public-domain prescriptions.

And although I found out that there was no lack of technical material addressing the subject of voice quality, I learned that the many books, articles, and papers devoted to the subject focused on the math, while steering clear of the practical heuristics that are the backbone of any successful implementation. Their analysis was mostly academic – deep, technically precise, but nonetheless narrow. It addressed a single aspect of voice quality, whether it was electrical or hybrid-echo cancelation, non-linear acoustic-echo control and suppression, adaptive and automatic gain control, or noise reduction, rather than an interactive blend. It was intended for the few subject experts and algorithm designers rather than the user, the product manager, the troubleshooter, the marketing and sales personnel, and – most importantly – those responsible for making decisions affecting quality of service and investment in voice-quality products.