# Clearing the haze

Prev Next

Try as I might, my explanations of differences between formats lack a certain clarity. So when someone else writes a beautifully clear explanation of the differences between DSD and PCM I am obliged to share it with you.

Why? Because, while I am certain many of you are tired of hearing me jump up and down about the sonic superiority of capturing analog with DSD in the recording process at Octave Records, understanding the why of the matter is crucial to unraveling the mystery.

"Tolerpro" (an interesting handle) posted the following note in the comments section. I just had to share.

"I think there's an important point that people are missing when comparing DXD and DSD, and that is the fundamental difference between PCM and DSD.

PCM quantizes the audio waveform into a series of instantaneous voltage readings taken at regular intervals (the sample rate). These readings are stored as integer values in words of 16, 24 or 32 bits.

DSD, on the other hand, maps the analog waveform itself onto a bitstream where the greater bit density represents a positive voltage while a lower density represents a negative voltage. In essence, DSD is not digital at all but a graphic representation of the analog waveform.

You can feed the DSD bitstream directly into an audio amplifier and get audio. Not so with PCM which comes through as meaningless hash. The PCM stream must be collected into words (16, 24, 32 bits) and converted to a voltage proportional to the value of the integer contained in the sample. Earlier DACs would pass these voltage samples through a filter to remove some of the added upper frequency harmonics and present the results to the amplifier. The result was an audio signal that was somewhat degraded from the original source.

Modern DACs use a process called a Fourier Transform which mathematically extracts the frequency information from the PCM sample stream. With this information, the DAC typically resamples at a higher sample rate reducing the artifacts and the required complexity of the output filter.

So, in essence, DSD is analog in nature, which explains why computers have so much trouble handling it. When it is pointed out the DXD and DSD have the same number of bits, while true, those bits are entirely different in nature and cannot be interchanged.

Ted Smith developed a means to take the frequency domain output from the Fourier Transform and render it directly to DSD and thus bypass the negative attributes of PCM conversion resulting in the sound quality we now enjoy from the PS Audio Directstream DAC.

Hope this helps."

Indeed it does.

Thank you.

Paul McGowan

Founder & CEO

Paul McGowan

Paul McGowan

Paul McGowan

Paul McGowan
• ### The talent myth

Paul McGowan

We are greatly saddened by the passing of Rondi D’Agostino (78) of Krell Industries. She was one of the most-loved and warm-hearted people in high-end audio. It was always a...

• ### PS Audio Celebrates 50 Years, Part Two: The Beg...

I will share some stories with you of our history over the last 50 years – half a century of making audio equipment. There are a lot of stories, some of...