DSD and PCM encoding comparison
As can be seen from the introduction above, DSD and PCM are two completely different thinking modes, and the two specifications also have their own advantages and disadvantages. The author will very briefly introduce the advantages and disadvantages of DSD encoding compared to PCM encoding. .
DSD coding advantages
Theoretically extremely high sampling bandwidth
The standard DSD music uses a sampling frequency of 2.8224MHz. In theory, it can record sounds up to 1411.2kHz (half the sampling frequency). Compared with the 44.1kHz sampling frequency used by traditional CDs, this allows the DSD format to have a very wide sampling bandwidth. Even when compared with the high-standard 24-bit / 192kHz PCM recordings commonly used in recording studios (sampling bandwidth up to 96kHz), DSD has more than 14 times the sampling bandwidth.
2. Excellent dynamic range <0 ~ 20kHz>
The DSD signal is processed by a high-frequency noise shaping (noiseshaping) circuit. After raising the high-frequency noise in the DSD to a frequency beyond the listening range of the human ear, the DSD has a theoretical data of up to 150db in the frequency range of 0 ~ 20kHz. Compared with the dynamic range of 96db / 144db with 16-bit / 24-bit PCM encoding, DSD encoding theoretically has a better dynamic range <0 ~ 20kHz>.
Comparing DSD and PCM encoding, the red box represents the recordable range of 24 / 96kHz PCM encoding, orange is 24 / 192kHz, and green is 24 / 384kHz recordable range. The 24-bit PCM signal noise floor is outside the recording range. Theoretically, it is less than -144db; the blue line is a schematic diagram of the noise floor of the 2.8224MHz DSD in different frequency ranges. It can be found that the noise floor of the DSD code rises sharply after about 30kHz, and reaches a loudness of about -6db when it approaches the limit frequency. <Image source: Playback Designs>
3. Smaller music capacity
DSD encoding also has certain advantages in storage capacity. With the improvement of sound quality, the file size of PCM music can be said to grow exponentially. Due to the advantages of DSD in encoding, under the standard 2.8224MHz sampling frequency, the DSD file capacity is only slightly more than 24 / 96kHz PCM music. This makes DSD better than the PCM format whether it is stored or downloaded over the network.
|format||File size||Estimated download time <5Mb / sec bandwidth>|
|Redbook Specification CD <16 / 44.1kHz>||32MB||1 minute|
|PCM 24 / 88.2kHz||95MB||2.6 minutes|
|PCM 24 / 96kHz||103MB||2.8 minutes|
|PCM 24 / 176.4kHz||190MB||5 minutes|
|PCM 24 / 352.8kHz <DXD>||380MB||10 minutes|
|DSD 2.8224MHz||127MB||3.4 minutes|
4. In addition to the advantages listed above, DSD music has the advantages of no sampling frequency limitation, direct support of DAC main clock and long transmission distance during playback.
DSD encoding disadvantages
The main advantages of DSD have been mentioned above, but like most specifications, DSD also has its technical disadvantages that cannot be overcome. And these practical shortcomings, on the contrary, seriously offset the advantages of DSD, and in most cases generate more problems, let me introduce the disadvantages of DSD encoding compared to PCM .
Actually narrower frequency range
The standard DSD encoding has a inherent disadvantage. Compared to the CD-specific 16bit / 44.1kHz -96db <16 * 6db> floor noise, the DSD floor noise is as high as -6db. This phenomenon can also be observed in actual measurement. Go to <See Figure X>. When the DSD code exceeds 20kHz, a large amount of high-frequency noise will be generated. If these noises are not removed, it will cause damage to the back-end audio equipment. Therefore, it must be processed by filtering technology. The current processing method is to use noise shaping circuits to move noise outside the human ear’s listening range <> 20kHz>.
However, this also means that these noises will be removed by a low-pass filter in the future. According to Sony’s specifications, the DSD signal must be processed by a low-pass filter circuit with a cutoff point of 50kHz to remove high-frequency noise. This makes the maximum frequency range that can be used during playback, only slightly higher than 30kHz, which directly affects one of the advantages of DSD: “theoretically extremely high sampling bandwidth.” In order to reduce the problem of high-frequency noise, the latest DSD files currently use double sampling frequency (DSD128). Theoretically, the noise shaping circuit can push the high-frequency noise to a higher frequency range, but it is actually not possible Solve the actual frequency range that can be used after passing the low-pass circuit.
2. The file has undergone multiple format conversions
Because DSD encoding is a unit-format recording format, DSD audio can hardly be edited with current audio equipment. Therefore, in the post-production of music, DSD audio must still be converted to PCM audio for editing. The current DSD music editing mode is to first convert DSD to Digital eXtreme Definition (DXD) format. DXD is essentially 352.8kHz / 24bit PCM encoded audio. After the post-production and music editing are completed, it is converted to DSD signal. Therefore, DSD music files inherit the shortcomings of DSD and PCM at the same time, and in the process of multiple conversions, another layer of high-frequency noise will be introduced. This makes DSD’s original intention to remove the PCM coding defects completely absent Now in DSD music.
3. Relatively high resolution PCM encoding has no significant advantage
The original comparison targets of DSD encoding were mostly CD-specific PCM audio. However, at the time when high-resolution PCM audio is popular, the actual sampling range and noise control capability of high-resolution PCM audio are completely better than the standard DSD format (see Figure X>. In order to solve this problem, the DSD128 audio with double sampling just mentioned was born. Although the problem of excessive DSD noise can be reduced slightly, the inherent defects of the unit element encoding still exist, and the size of the DSD128 music file with double sampling. Compared to high-resolution PCM audio, it also has no advantage in storage capacity.
4. Hardware support is not extensive
DSD processing For a music production company that has been accustomed to PCM music for a long time, it must invest additional costs to perform post-production, so it seriously affects the promotion of the DSD format. In addition, most DACs currently on the market that support DSD playback use the method of re-encoding DSD audio back to PCM signals for playback. Only a very small part of DACs support native playback of DSD music. In addition, DSD transmission is not included in most transmission specifications, such as USB Audio 2.0 and Mac platform feature specifications, audio transmission is only a standard PCM format.
Fortunately, the DSD-over-PCM <DoP> transmission specification has recently been promoted to solve this problem. This transmission specification uses 24bit / 176.4kHz PCM transmission to package DSD signals, and uses the first 8 bits as DSD marks. And because the amount of DSD audio data sampled at 2.8224MHz is equivalent to the amount of 16bit / 176.4kHz PCM audio data, the next 16 bits can be used to package DSD audio, allowing capable hardware manufacturers to develop DSD solutions on their own, and also let IOS systems that do not support DSD transmission have the opportunity to play DSD music. Of course, this transmission specification also has its disadvantages. That is because the data is packaged by PCM signals. If the hardware misjudges the DSD as a PCM signal, a noise of about -34db will be generated at 88kHz and played. The related operation and setting process of DSD is quite complicated, so DSD music specifications are still relatively unpopular.
Schematic diagram of packaging standard DSD audio in 16 / 176.4kHz PCM format. The first 8 bits are identification tags, and the last 16 bits are DSD encoded signals.
I believe that after reading the above introduction article, readers have a certain understanding of the DSD encoding format. Some people may ask: Since DSD has so many inconveniences and shortcomings, why has a group of manufacturers recently promoted the DSD music format?
I must say here that no format is without its disadvantages, depending on the actual use and market response. However, regarding the reason why the DSD specification will be promoted recently, the author speculates that it has been more than ten years since the appearance of the DSD specification. Most technical patents have a validity period of about ten years, so basically the DSD format has now become an open system. specification. Then come the main company that promotes DSD is Playback Design, where Andreas Koch, who developed the DSD specification, originally hopes to promote DSD music.
Instead, Daniel Weiss, the owner of Weiss, a Swiss digital music company that converts DSD and PCM standard specifications in the manufacturing industry, published a short DSD white paper two years ago. It is clear that DSD cannot be post-processed. It must be processed first. Convert to PCM signal to perform necessary post-production work. Therefore, in the recording stage, Weiss recommends that it is more practical to save the sound signal in PCM format first.
By the way, look back at the Sony that originally formulated the DSD specification. Currently, with the Japan Audio Association, a new “Hi-Res” certification is developed. After carefully reading its internal specifications, you can find that although DSD is also regarded as a high-resolution sound source, In the definition of high-resolution audio source, there is no special attention to the DSD format. Sony’s attitude towards this audio format that has lost its patent dominance can be deduced.
But in general, it is not really important whether DSD and PCM are good or bad. They are just a format for storing music. What is important is the music contained in it. Because the moving of music is not determined by the format of the music unit, as long as there are enough people to continue using it, the meaning of it is retained. As for who will become the hegemon of future formats in DSD and PCM? Let the music production company worry.