Waversa is not alone, the new Sparkler S512 DAC does something similar:
16-bit, reconsideration and new approach ? Development history of “D/A processor engine”
The reason why digital audio became popular was the CD (compact disc) that appeared in 1982. Since then, the digitalization of music media has progressed remarkably and has been further supported by the spread of computer technology and the Internet. On the other hand, analog discs such as LP records have been declining temporarily due to the wave of digitalization, but in recent years, not only audiophiles but also many people have come to review again. That’s because there is a world of elaborate and rich sound that can only be enjoyed with analog records, and we also feel that there is something missing in digital audio. So what is the missing part?
When the CD was born, of course, there was no “high-resolution” music distribution as it is today, and it was thought that 16-bit sound could surpass the quality of analog records. Surely, CDs are easier to handle, reduced noise, and the track can be selected with remote control. Once the CD was recognized as a standard format suitable for music media, its playback devices also spread dramatically on the market, and along with that, it became possible to listen to numerous types of music on the CD. It is certain that there are points beyond the vinyl disc in this way, but it cannot be denied that there are some points that are not so in the sense of hearing.
Then, it is not so easy to say that it will be better if it is digitized more finely. The aim of so-called “high-resolution” is to transmit analog signals more finely and with higher precision digitizing. It can be acceptable because it is inevitably associated with the development of digital technology. But then, no matter where we go, we won’t leave the field of digitization. Many new formats that exceed the performance of CDs are defined one after another, and the advantages of each are being promoted, but there is still no overwhelming market to replace CDs. Because it is annoying to have a variety of formats for ordinary listeners, and it’s enough because we can usually enjoy music on CDs already on the market. Also, the CD in a beautiful jacket has a presence rather than just the data on the computer, and we want to keep it like a favourite book.
Our main focus is on the lack of usefulness of CDs. There is a feeling that we also want to try high-resolution, but it is a shame that we don’t have time to change and handle it halfway. So we decided to do our best with the CD for a while. However, if we listen to the music carefully, the sound quality of the CD is still very impressive, and we respect the excellent achievement of the basic research of the predecessors. Certainly, 16-bit performance is too good for the human ear, and there is no doubt that there are more possibilities if it can be played properly.
We will approach the main topic from now on, but one thing we are interested in recently is that compared to analog records, the sound quality of a CD is that the harmonic content is not clear. In other words, the sound does not “stand” like a live musical instrument, and there is a feeling of “sleepy” that is peculiar to digital. It is not a lucid sound. Interestingly, it is not limited to CDs, and we get a similar impression when we listen to a high-resolution sound source. Perhaps it was missing or altered by digitization itself. On the other hand, the playback of analog records is more like the way a musical instrument sounds because there is a process of physically rubbing a groove.
To explain such an impression in technical terms, it means that “the higher the frequency, the more unclear the phase”. The phase represents a position on the time axis of a periodically changing wave. If the phase of the harmonic component is significantly shifted, the timbre will be different, so there will be no instrument-specific sound. If the phase is disturbed, the chords is not played in harmony, the focusing of the sound image is ambiguous, and the parts of the instrument that contain their specific overtones cannot be determined. Moreover, the loss or alteration has already occurred when the analog signal is converted into a digital signal, and even if the reproduction device restores the digital signal to analog faithfully, the phase once shifted is not restored. The sound quality is so smooth and transparent that it is aimed at faithful reproduction such as oversampling method, but it alone is boring. For example, the tension of a piano string that vibrates with a strong keystroke, or the sparkling overtone in the prompt sound, will not come out even if the digital signal is reproduced faithfully. The high range is delayed by one tempo no matter how many times we listen. But we aim to reproduce it. The vibration of a piano string actually contains over 7th harmonic components (more than 3 octave overtone), which determine the unique sound of the piano. When playing back musical tone, we feel that it is very important to be able to hear the pitch clearly.
When the phase shift between the fundamental and overtone occurs,
the waveform changes every moment and can be heard as a beat.
The foot of the lighthouse was really dark. That is probably because the original digital signal itself is already out of phase. As a result of forcibly rounding the position on the time axis at the sampling frequency of 44.1 kHz, the phase is not correctly recorded, and the phase may already be shifted from the original analog signal. They are also called “phase distortion”, which means that they are not in line. We won’t go into detail because it goes into the difficult sampling theory, but we think that you can understand the audibility from experience. In an extreme example, it looks like white noise when not receiving FM broadcasts, and the sound image is not focused anywhere. The natural space where musical tone resonate is on the other side. CDs that are not good at recording or digitizing will not produce a decent sound no matter how the original sound is played, but the key to solving it should be elsewhere. Therefore, after the digital signal is converted to an analog signal by a device called a digital to analog converter (DAC), it is necessary to correct this shifted phase in the entire audio system. There is not much problem with a modern amplifier with a simple characteristic, but with a classic amplifier with a narrow band, the phase is gradually delayed in the high frequency range, and even in the ultra high frequency range, oscillation may occur. In comparison, loudspeakers are not very accurate in phase and have a strong character. Cables that are often overlooked are also spiders. When listening in total, the phase is still out of phase. Jitter is also a temporal fluctuation of a digital signal like an arrhythmia, so it also affects the sound quality. Apparently the human sense seems to be particularly sensitive to time fluctuations.
Furthermore, it is also necessary to electrically compensate for the super-high frequency overtone component that has been lost due to the digitization of a limited band, as occurs in an actual musical instrument. Many of them will not be heard by human ears, but they feel as if the instrument is sounding there, and the tension and atmosphere of the air. It becomes even more difficult to reproduce the sound field and the sound of the hall. However, this can be improved by correcting the phase shift, and the sound reverberates beyond the presence of the loudspeakers. However, if we make a big deal like multi-channel surround, we can’t use the existing audio system as it is. Can we manage it in the range of two channel stereo? This is because it is the most desirable and practical way to make it closer to a realistic acoustic space by simply modifying a part of the existing system.
Up to this point, we haven’t gone out of speculation and contemplation, so we decided to prototype a DAC with such phase correction. Non-oversampling (NOS), which does not process digital signals, is rather convenient for experiments because the phase characteristics remain as they are. After converting to an analog signal with a standard 16-bit NOS-DAC, we first passed a filter that gradually advanced the phase in proportion to the frequency in the audible band where the human ear is sensitive. Next, in order to generate a missing harmonic component, a circuit that emulates a triode vacuum tube with a relatively high harmonic distortion was assembled with a semiconductor. As a result of repeating the audition with various parameters changed, it became a very realistic and pleasant sound with a certain correction value. It was unnatural if it was overcorrected, but it doesn’t have to be as strict as pinpointing. The older recordings is more effective, so the old equipment was probably out of phase. If the correction is appropriate, the delicate passages that have been buried will come out. It doesn’t hear like a CD, but rather close to the sound of a fine analog record. The sound quality is crisp and it is almost like a record needle physically rubbing a groove. In addition, the space expression of the performance venue becomes more natural, and we can feel the sound of being wrapped anywhere in the room despite the two channels. A person happened to hear this prototype and was wondering where it was sounding. Of course, when I answered that they were from these two speakers, he couldn’t hear that way, and looked around mysteriously that there were other speakers somewhere. We are working hard to further refine this prototype and commercialize it as a 16-bit DAC with a new approach. <10 October, 2019>
