Audio Basis - Educational program
DSD converter is software for conversion DSD audio files to other formats and back. DSD audio files have extensions DSF, DFF, ISO. As target or source formats PCM multibit formats (WAV, FLAC, AIFF, mp3, etc. files and CD-audio disks) can be used. Also such converter can resample DSD files. Below will described how to work the converter.
Format DSD is implementation of sigma-delta modulation. There used sample rates that multiple 44100 Hz:
Bit depth is 1 bit.
Software DSD converter can have inside:
1. PCM (multibit) modulator to delta-sigma modulation (1-bit), or
2. Sigma-delta demodulator (1-bit) to PCM (multibit), or
3. Both: modulator and demodulator.
Structure of DSD converter
DSD modulator provide decreasing number of bits. It lead to increasing noise level (error).
Most useful part of audio signal placed in range 0 ... 20 kHz. The modulator "push" energy of noise in range above 20 kHz (called as noise shaping).
Demodulator provide filtering this noise. After the filtering remains only the most useful audio signal.
Watch video about structure and how to work DSD converter
Sigma-delta modulator (PCM to DSD) is sophisticated audio processing. It is system with feedback. Hence it can lose stability. If sigma-delta modulator (DSD modulator) lose stability, need restart it. As rule it can't return to stable mode without external impact.
In general, dosing of sigma delta modulator is compromise between stability and noise level. More significant pushing of noise energy out of audible range, may lead to lose stability. DSD modulator can lose stability due overload.
For achieving better signal/noise ratio inside audible range and keeping stability, frequency band out of audible range must have reserve by width. It may be achieved via increasing sample rate.
Sample rate increasing is easier way to have wider band with lower noise.
Unfortunatelly, I don't know exactly history of creating DSD, but, looks like, sample rate D64 was designed for transmitting 20 kHz band. As rule, DSD demodulator is easier, comparing PCM one (read more). So replacing 44 kHz/16 bit to DSD64 may give some technical advantages for implementation.
For transfering of 100 kHz-band signal with keeping or decreasing noise floor, easiest way is sample rate increasing.
Read more about sigma-delta modulator >
Sigma-delta demodulator (DSD to PCM) or DSD demodulator is low frequency filter, that remove modulation noise, located above audible range.
This filter have 3 features:
Wider transient band allow decrease ringing audio, but pass more energy of noise of DSD modulator.
If expand pass band, DSD's noise energy also increased, pass band become narrower, ringing increased.
So DSD demodulator is balanced technical decision.
Read more about DSD decoder >
Before resampling file DSD converted without loses to multibit audio stream suitable for processing . After it processed and converted back to DSD via modulator.
Resampling 1-bit audio cause little losses, comparable with losses for PCM resampling. There losses defined by distortions of used implementation of processing.
However, playback apparatus may have different distortion level for different sample rates. Hence resampling may give benefits, if DSD converted to sample rate that works with minimal distortions on used DAC.
Resampling DSD audio file
for sample rate that works with minimal distortions on used DAC
DSD converter is software may contains:
Playback hardware may have different distortions for different modes (sample rate, bit depth, PCM/DSD).
Resampling DSD or conversion PCM to DSD and DSD to PCM may give benefits, if target format suitable for DAC mode, that cause minimal distortions.