New research benchmarks digital filters for accurate phonograph playback

Phonograph records have been cut with intentionally non-uniform frequency response since electrical recording began in 1924, maximizing information density and signal-to-noise ratio. To achieve a flat playback, each disc needs an inverse filter—but until the RIAA standard in 1953, equalization curves varied by record company and era. Restoring these recordings digitally requires converting the continuous-time analog curves into discrete-time filters, a transformation that inevitably introduces deviations, especially near the Nyquist frequency. This survey, presented at the AES 157th Convention and winning Best Student Paper, systematically compares established discretization methods (e.g., bilinear transform, impulse invariance, matched z-transform) in the context of phonograph playback equalization. The authors quantify performance, computational cost, and inherent latency for each approach, offering a benchmark for developers building digital playback equalizers or similar systems that must approximate continuous-time filter responses.

The work is explicitly intended as a resource for engineers restoring vintage audio. By clearly documenting the trade-offs among accuracy, speed, and delay, it helps practitioners choose the right discretization technique for their application. The paper also highlights that while digital filters are cheaper and more versatile than analog hardware, the quality of the digital approximation depends heavily on the method used. This survey fills a gap in the literature, providing a structured comparison that can guide both academic research and practical implementation in audio restoration and playback software.