ABSTRACT
Direct Stream Digital (DSD) is widely misunderstood as simply a higher-resolution version of standard digital PCM audio. This paper argues for a more precise framing: DSD is a modulated carrier format, not a sampled measured digital value format. Like vinyl, with a physical material being the carrier, or FM radio, using a high frequency analog sine wave as a carrier, DSD produces a continuous bitstream carrier encoded with an audio signal. Understanding this distinction clarifies why DSD-recorded and played music reproduces differently from PCM-based formats, and why the analog comparison is technically grounded rather than merely aspirational.
SECTION 1
The Carrier Principle
Every audio format is, at root, a medium for transporting an audio signal from one point to another. What distinguishes formats is not whether they carry a signal, but how. The critical dividing line in audio is between formats that measure a signal at intervals and formats that carry it continuously.
A carrier format holds an impression of the original waveform in the medium itself. The signal is not described in steps or stored as a series of numeric values. It is encoded as a continuous modulation of the medium. Recovery of the original signal requires only reversing that modulation, not interpolating between stored measurements.
Vinyl is the most familiar example. The cutting lathe impresses a continuous groove that follows the waveform directly. The cartridge reads that groove back out. No sampling, no reconstruction filter, no staircase. The groove is the signal.
SECTION 2
Modulation as the Unifying Principle
FM radio operates on the same carrier principle at a different physical layer. A carrier sine wave is modulated: its frequency shifts in proportion to the instantaneous audio signal level. The receiver demodulates the carrier and recovers the original waveform. The audio is never expressed as a number. It lives entirely in the change of frequency of the modulated carrier.
This is the conceptual bridge to DSD. DSD also encodes audio through modulation; producing a bitstream of 1’s and 0’s bits at a specified high frequency rate (2.822 MHz for DSD64 through 22.4MHz for DSD512). (See Figure 1). A bitstream can be stored and retrieved by a digital computer as a stream of ones and zeros. What that bitstream encodes, however, is not a series of amplitude measurements. It encodes the audio levels through the continuous variation of carrier's 1’s and 0’s bit density over time, at a multi megahertz carrier frequency.
SECTION 3
How DSD Encodes Audio
DSD uses single-bit delta-sigma modulation. The modulator’s function is to produce a one/zero (1/0) bitstream whose bitstream’s density of ones-bits to zeroes-bits is proportional to, and accurately describes the modulating audio signal level magnitude and direction. At zero audio signal level, the resulting output bitstream is a square wave shape with an equal quantity of alternating 1s and 0s bits. As the modulating audio signal level increases in the positive direction, the population (density) of 1s bits versus the 0s bits increases proportionately. Conversely for negative direction audio signals, the density of 0s bits increases proportionately over the 1s bits decreasing population.
The modulator’s DSD output is a continuous 1s/0s digital bitstream accurately describing the audio signal in a modulated digital bit density form. Simple analog signal integration is all that’s needed to reconstruct the audio signal.
What makes this encoding approach so accurate is the encoding process granularity. The bitstream clocking speed frequency at 11.2 MHz (DSD256) compared to the input audio frequency spectrum is well over 100 times higher than the highest audible frequency. Instead of chopping music into amplitude “snapshots” at a much lower sample rate like PCM does, DSD tracks the waveform as a continuous flow, converting it directly into a high-speed pattern of 1s and 0s.There are no multi-bit amplitude values—only the shape of the audio signal expressed through the 1s to 0s/0s to 1s bit density. The result is a simpler, more natural transfer of the original waveform, exceptional transparency. Like the groove of a vinyl record, the signal exists as pure modulation—preserving timing, nuance, and musical detail.
SECTION 4
Implications for Music Reproduction
The carrier framing carries a practical implication: DSD recordings made natively in DSD from the microphone forward have undergone no quantization of the audio waveform at any stage of production. (See Figure 2) The continuous signal captured in the recording session is carried through the format to the listener's DAC without ever being expressed as finite amplitude values. The output of a DSD DAC is the result of demodulating (integration) of a continuously varying bitstream, much as an FM receiver demodulates a continuously varying carrier sinewave.
This is not simply a resolution argument. Higher PCM resolutions reduce quantization artifacts significantly and produce excellent results. The DSD distinction is categorical rather than incremental: a different class of format, not merely more of the same. The comparison is less "better digital" than it is "analog principles, digital medium."
Conclusion
The analog comparison applied to DSD is not “marketing language". It is a technically accurate description of how the format works. Like vinyl and FM radio, DSD encodes audio in a continuously modulated medium, rather than as a sequence of stored values. The bits are the medium; the signal they carry is continuous. For listeners who have long valued vinyl for its carrier fidelity — the sense that the music was impressed into the medium rather than transcribed from it — DSD offers the same principle without the physical constraints of imprinting a physical groove: greater sonic detail, no surface noise, no wear, no degradation across plays. The groove has become a bitstream. The principle has not changed.