Digitization of high quality audio usually requires precision successive approximation A/D converter circuitry. Dynamic ranges of 80-90 dB demand long digital word lengths, typically 14 to 16 bits. Utilizing predictive coding techniques and current integrated circuit technology, a scheme similar to some used in speech applications, is presented which considerably reduces output word length while maintaining the performance required in high quality audio. The scheme utilizes samples analog predictive filters as an integral part of the standard successive approximation process. The flexibility provided by this configuration allows accurate control of frequency response, full power bandwidth and quantization noise.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.