Steve described the designs of converters used for audio in past and present equipment, and showed how the converter type and design features changed as higher sample rates and larger word sizes were demanded. Even as the performance of converters went up, they became easier to implement and had a dramatic drop in cost. The analog electronics are now becoming the limiting factor for signal to noise, rather than the converters.
illustrated the precision of modern day converters, explaining that if 20
bits were used to map the distance from his hometown of
Looking to the future, Steve explained that the use of multiple digital processes in live audio has led to requirements for shorter time delays in converters. This, along with concern about pre-ringing in the anti-alias filters has lead to changes in the design of the phase response for these filters. In addition, the bandwidth of converters now easily exceeds human hearing, allowing for gentler shaped cutoffs that ring less.
Jim Johnston gave a rapid paced presentation in the afternoon. He first covered human hearing, explaining how the functioning of the cochlea determines our ability to discriminate multiple signals over a wide dynamic range. He emphasized that the ear functions well because of many highly nonlinear processes, even though the perceived result seems linear.
He then covered the basics of quantizing and sampling audio, relating the limits of digital systems to the limits of human hearing. He showed how the limits of hearing with respect to phase shifts and masking thresholds relate to predicting the audibility of the time response of anti-aliasing filters.
The third section of Jim’s talk covered the use of noise shaping. He illustrated how very high S/N ratios and very flat response in the audio band have been made possible by using higher sample rates in the converters in combination with noise shaping and digital filters.
A luncheon was served in between the presentations.