The traditional Delta-sigma DA chip boasts a history spanning several decades, yet certain issues remain that cannot be flawlessly resolved. These include transient distortion, phase distortion, and more. While these distortions noticeably impact our auditory perception, their influence on routine, repetitive static index tests is minimal. The concepts we frequently discuss, like the digital sense of sound and the sense of fragmentation, are interconnected with these distortions.
Decoding in NOS mode typically involves minimizing digital processing of the Delta-sigma to achieve an output that closely resembles the native signal. However, as we're aware, the introduction of frequency sampling noise significantly affects the entire DA's noise profile. Take, for instance, the small distortion tail observed in Soundaware. In normal mode, THD+N at -60dB might achieve approximately 72dB. Yet, with the utilization of NOS, this figure might only be around 40dB. This substantial gap of around 30dB not only results in a decline in the improved auditory perception associated with NOS but also renders the overall distortion and noise nearly unacceptable.
However, this approach is not without a solution. Audiophiles familiar with PC-HIFI are likely aware that when using NOS mode with HQPlayer processing, up-converting 44.1kHz to 358.2kHz results in the elimination of distortion within the audible audio range. This method stands out as the simplest approach, offering both minimal distortion and superior sound quality.
Evidently, the integration of a high-performance upscaler into a decoder is not a novel idea in the realm of HIFI. Many such decoders exist, although the cost of a feature-rich upscaler might match that of a high-end decoder. The primary factors influencing this are the performance and precision of the Sample Rate Conversion (SRC) process and the quality of the filter. The challenge lies in achieving real-time processing. You can conduct a basic experiment to validate this. By utilizing high-performance SRC software on a PC to convert 44.1kHz to 358.2kHz, the distortion in NOS decoding can also be remarkably low.
The performance of today's digital-to-analog converters (DACs) is continuously improving, with increasing strength. The original Sample Rate Conversion (SRC) solution operating within a 24-bit framework can no longer fulfill the demands, leading to the inevitable shift toward genuine 32-bit SRC. The SRC implementation utilized by DAM1 employs a 32-bit Finite Impulse Response (FIR) high-performance digital filtering process, achieving distortion levels of <-170dB. Consequently, under DAM1, the NOS decoding performance, even at 44.1kHz, can approach the product's optimal performance benchmarks. This results in a sound that is not only more coherent and natural but also surpasses the default delta-sigma method.
Presently, the most effective approach for improving NOS decoding performance and sound quality involves the integration of a high-performance SRC+ filter module prior to NOS mode decoding. This solution stands as the best available method for addressing NOS decoding indicators and enhancing audio quality.
The decoder's principle is not overly complex, but the nuances hold significance. Some individuals mention employing HQ for HQ upscaling and wonder why the disparity compared to PCM upscaling isn't conspicuous. This is primarily due to HQPlayer's operation on a PC. PCs do not execute HIFI processing; factors like delay and substantial noise impact the decoder's effectiveness during actual playback. The matter of PCM frequency upscaling might be influenced by PC enhancements.