High-Resolution Metadata and Digital Interconnects: An Interview with Jeffrey Barish of 3beez
You may have noticed a banner ad from 3beez on this webpage. 3beez founder and president, Jeffrey Barish reached out to me some months ago via email and it seems we’re on the same page on many issues relating to high-end audio (cables, DSD, and more). So I offered to interview him regarding his present efforts in the area of “music management”. I think you’ll find his initiative worth checking out.
Jeffrey Barish has developed a “music management system” called 3beez. He has been involved with digital audio for over 40 years and has several startups under his belt, including Sonic Solutions, where he invented the NoNoise system, and EuPhonics, where he developed software for audio effects and music synthesis for the computer and semiconductor industries. His product may be interesting to readers, but in any case his comments about how interconnects can affect sound quality should be.
MW: Can you tell me what a music management system is?
JB: I coined the term “music management system” to describe our Wax Box system because that phrase accurately describes what the system does. It produces music, it manages the music by storing and cataloging it, and it is a complete system, not a hardware or software component that needs other components to provide all the functionality required to enjoy a collection of digital music.
We believe that most music lovers appreciate the simplicity and elegance of a self-contained system. There are no issues with interoperability of separate components made by different companies. We avoid the constraints of the standards required for interoperability – UPnP and DLNA. The integrated software performs all operations required, so there is no need to select, install, and learn multiple software products. A self-contained system is easy to install. Even technophobes can set up a Wax Box and be listening to music in minutes.
MW: So it sounds like you’ve developed a complete solution, what else makes the 3beez system unique?
JB: Two things. First is the custom Wax software. Wax provides a uniform user interface for performing all of the operations required in a digital music system. Wax makes it easy to gather, enter, or edit metadata – as much as you want – even while simultaneously playing music. It eliminates the need to index sound files and to organize your sound files in particular ways. It eliminates “file ping-pong” – the need to bounce sound files to a desktop system to make even trivial changes to the metadata and then bounce them back. Best of all, it provides high-resolution metadata, which is our term for the rich metadata that can be stored using Wax. Having control over both the hardware and software allows us to deliver a solution that is at once easier to use and more capable.
Custom hardware is the other distinguishing characteristic. Our latest system has a custom digital interface board to improve sound quality.
MW: What are the advantages of using a custom digital interface board? Isn’t managing music a software thing?
JB: Although our product has a DAC built in, we have always offered digital outputs for listeners who wanted to connect an external DAC for the best possible sound quality. In the past, we offered only asynchronous USB and TOSLINK outputs. I was partial to USB because that interface gives control of the clock driving the DAC chip to the external DAC. However, I found that we were not getting the sound quality I expected with many DACs. The fault obviously could not be attributed to corruption of the digital data because the effects were too subtle. The only other possibility was that electrical noise generated by the circuitry in the Wax Box was reaching the external DAC over the digital interconnect and corrupting the operation of the DAC. Since electrical noise is ubiquitous, DACs should protect themselves from it, but it seems that many do not. Accordingly, I set out to design a digital interface that prevented this noise from exiting the Wax Box.
My first step was to choose the best digital format. As I said, I had always been partial to asynchronous USB, but I had also come to realize that USB introduces serious countervailing challenges. The primary rationale for using an external DAC is to isolate the sensitive DAC chip and its associated analog circuitry from the electrically noisy environment of the computer at the core of any digital music system. However, the USB protocol itself requires a computer. Moreover, the data lines in a USB interface are bidirectional, so it is very difficult to galvanically isolate them when the interface is operating at the speed necessary to support HDA. Thus, USB compounds the noise problem: Noise arriving on the interconnect still penetrates to the internal circuitry and there is additional noise being produced by the internal computer required to implement the USB protocol. Some DAC manufacturers seem to have overcome these challenges, but given a choice I wondered whether USB really is the best one.
I ruled out I2S because so few products support it. Moreover, I2S was designed to communicate between chips, not systems. That left the family of interfaces based on AES/EBU. Optical fiber ought to be best because it provides galvanic isolation automatically. However, the implementation enshrined in the TOSLINK standard has limitations (the use of plastic fiber, principally) that may increase jitter. The electrical forms of the standard avoid the limitations, and providing galvanic isolation is straightforward because AES/EBU is unidirectional. I was always wary of the need to extract the clock from the digital stream because of the possibility that the extraction could introduce jitter, but designers have solved this problem. Thus, AES/EBU was my choice as the least bad interface available.
Our custom digital interface board – we call it the “BitScrubber board” – uses transformer coupling at both its input and output for maximum common-mode noise reduction and for galvanic isolation. The line driver chip has high bandwidth and low output impedance so that it can properly drive the interconnect even at the high frequencies required by HDA. Careful layout of the PCB maintains the balance of all differential signals to maximize the rejection of common-mode noise by the transformers. Objective measurements of the noise rejection capabilities of the board, which we published at our web site, show how effective the design is. Careful subjective tests confirm that the board improves sound quality, at least with some DACs.
MW: What do you mean by “high-resolution metadata”?
JB: Just as “high-resolution audio” provides a more complete representation of the audio, “high-resolution metadata” (HRM) provides a more complete description of the music. People who care about sound quality appreciate the fidelity provided by high-resolution audio; people who care about music appreciate the enlightenment provided by high-resolution metadata. Most music manager programs provide only a little information about a recording – typically the album title, the name of the artist, the title of the track, the genre, and a date. Music lovers like to know more. For example, a jazz lover might like to know that the 1947 Charlie Parker recording “Encores” has a young Miles Davis playing on several of the tracks. An opera lover would generally want to know the cast of the opera. All music lovers might want to be able to access liner notes, and they might be interested in an image of the back cover as well as the front cover. Not everyone cares about these things, just as not everyone cares about higher fidelity, but many do.
HRM is useful for more than just supplemental information; it also aids in distinguishing recordings. The small set of common tags may be sufficient for a small collection of pop music, but HRM is essential for large collections, especially when the collection includes classical music. In my collection, I have six versions of the show West Side Story. I distinguish them by cataloging the date of the original production along with the name of the show and other things. I have 12 versions of Mahler Symphony No. 2. Without the name of the conductor, I have no way of knowing which is which. Some collectors have multiple versions of a concerto with the same soloist, conductor, and orchestra, but recorded on different occasions. HRM makes it easy to handle these situations.
The limitations of standard metadata arose because the standards were developed at a time (1996) when disk space was precious and by people who thought only about support for pop music. The most widely used standard is ID3, which is the standard for embedding metadata as tags in MP3 files. ID3 actually defines a surprisingly large set of tags, but few music management programs handle them well. You can provide the name of the conductor with ID3 tags, but doing so doesn’t help if your player does not display the conductor tag.
Another problem with tag standards like ID3 is that they are too rigid. It doesn’t help to have a tag for conductor when you are cataloging pop recordings. Likewise, it doesn’t help to have a tag for album title when you are cataloging symphonic works. Always displaying all available tags clutters the display. With Wax, users can define the set of metadata keys best suited to each genre. That way, you get a “conductor” field only when you are in a genre for classical music and an “album title” field only when you are in a genre for pop. With flexible metadata, Wax can display exactly the information required to differentiate the recordings in each genre.
There are many more aspects to high-resolution metadata, but being able to define as much metadata as you want using exactly the metadata fields you want are the two essential characteristics.
MW: Don’t existing media servers in conjunction with Roon solve the problems that you are describing without requiring any effort on the part of the user?
JB: Roon in conjunction with a streaming service like Spotify or Tidal is a good solution for many listeners. Together, they provide a vast selection of music and a lot of information related to the music. If you are only interested in exploring new music or in having some background music, Roon’s user interface can be seductive. However, problems arise when you are looking for a particular recording or work. For example, the overview of your collection in Roon might display two versions of Pearl Jam Vitalogy. The two versions are indistinguishable in this display. You have to click on each cover in turn to get the information that one is for the standard CD release and the other is for the 24/96 HDtracks remaster. This problem gets much worse with classical music. No doubt Roon has metadata for even more than the 12 versions of Mahler Symphony No. 2 in my collection. It would be very tedious if I had to click on each cover in turn to get the name of the conductor and the orchestra.
The way that Roon handles CDs with multiple works is also problematic. Roon at least has a nicer way of displaying such CDs than iTunes does. The tracks for each work are grouped under a heading identifying the work. However, music lovers typically want to be able to survey all the versions of a particular work to select the one they want to hear, even when each version is one work on a multi-work CD. Roon makes it possible only to survey entire albums, and then only by cover art and scant metadata.
Roon does provide a lot of information about a recording under “credits”. However, if I am looking for the recording of Aida with Leontyne Price, I have to click once to get from the overview display to the album display and then a second time to bring up the album credits. If I guessed wrong, I have to click twice more to get back to the overview display and then start over with the next guess.
At this time, it is not possible to add metadata. Even when Roon adds this functionality, users have to remember that any changes or additions go away when they stop paying $119 per year to maintain their license.
Roon is the latest in a succession of products that promise utopian ease in cataloging a digital music collection. Some music lovers will be satisfied, but people who really care about their music collection are always going to want more control over their metadata.
MW: Any final thoughts on the state of high-resolution music and how your system enhances the user experience?
JB: It’s a great time to be an audiophile because technological advances are improving every aspect of the listening experience. It took a while to realize the potential of digital technology to provide great sound quality. As you demonstrate through your business, it is now possible through meticulous engineering and the choice of the right technology to create recordings that sound great. However, for serious listeners, listening to music involves more than just the auditory experience.
It has also taken a while to get to the point where digital technology is finally able to do a good job delivering quality metadata. Serious listeners want to be able to find their music quickly and to know things about what they are listening to: names of all the artists, background information about the composer, the venue – maybe even the name of the recording engineer. High-resolution metadata makes the listening experience complete.
9 thoughts on “High-Resolution Metadata and Digital Interconnects: An Interview with Jeffrey Barish of 3beez”
Loved their argument, hate the price. They could put their algorithms to a raspberry board, package it with universal wifi ( especially one that networks cellulary ) and call it a day with a nice enclosure. Leave the hardware to competitive OEMs. All that for a fraction of the cost and quite likely comparable profits.
Perhaps you or someone else can explain why plastic fiber optic increases jitter. That’s a new on me. Thanks.
One should keep in mind that roon is evolving. And these guys are learning fast.
In my opinion two metadata solutions to a problem that does not exist. How about instead of two clicks, modify title to read Sgt. Pepper – orig mono, or Sgt. Pepper – stereo.
Got any objective data on why USB is a problem? I am surprised you gave them coverage here. Sounds like more audiophile pablum.
Sigh….Another closed, proprietary system. When/if 3Beez goes belly up, you better hope that your HRM files are portable. No thanks. I’ll stick to Freeware solutions that allow me to organize my digital files in structured folders on NAS drives which are accessible by Foobar and EAC. As far as metadata is concerned, I have the original CD for all but a couple of files which were offered for free from HDtracks. When I want metadata, I pullout the original CD. I have avoided formats (DSD, DVD-A, etc) which resist my efforts to rip my own personal backup copies. I will not be purchasing any download files or proprietary systems as long as the physical medium is still available. Given the life expectancy of vinyl and CDs, that could be a very long time. “Oh but”, you say, “These are not HiRez formats.” Pfffffttt! Vinyl has stirred my emotions for 40 years and CDs have been doing the same for 14 years. It’s about music, not formats or systems.
This is an intetesting topic indeed, and speaks directly to those of us who require better metadata to navigate our collection, and who are stuck looking at a display with a wall full of album covers. I have a large collection and face similar problems to those discribed: several recordings/versions of the same work, poor or useless metadata and no freedom to define my own logic criterias of organization.
I use Audirvana, which is a decent player from a playback and sound quality perspective, but clumsy and precarious at best, when it comes to organizing music, finding it, let alone displaying valuable or additional artwork or liner notes information during playback. My most immediate orientation vector is artwork, which is far from practical when you have a large collection.
Although the effort of 3beez is a valid one, the arguments in the comments section, regarding the price tag and proprietary nature of the component and metadata solutions provided, in my opinion outweigh the benefit provided by 3beez. The possible obsolescence or future lack of updates of the platform are also valid points.
On the other hand, and although 3beez recently displayed their technology with a Benchmark Media DAC2 and AHB2 Power Amps, we know that especially Benchmark DACs will see no benefit from “bitscrubbing” or external jitter reduction devices. John Siau has been clear in this respect, but I’d still appreciate him clarifying this with regard to Benchmark’s DAC2 being exhibited with the 3beez component, implicitly suggesting an improvement in sound quality provided by “bitscrubbing”. (“CLOCK GENERATION, CLOCK RECOVERY AND JITTER ATTENUATION
No discussion of audio converters would be complete without mentioning clock generation, clock recovery and jitter attenuation. An SPDIF coaxial digital input or a Toslink digital input can be buffered and sent directly to the input pins on the ES9018. This low cost solution is popular with many manufacturers, but it is not the method used by Benchmark. The Benchmark UltraLock2™ jitter attenuation system is entirely external to the ES9018. All incoming clocks are fully isolated from the low-jitter fixed-frequency crystal oscillator that drives the D/A conversion. This oscillator is located less than one inch away from the ES9018, and the clock connection is well shielded and isolated from the surrounding circuitry through the use of a 6-layer printed circuit board.
All digital inputs on the DAC2 are asynchronously upsampled to 211 kHz to remove all traces of interface jitter. Benchmark’s UltraLock2™ upsampling system provides uniform jitter-free performance with all types of digital interfaces (coaxial, XLR, Optical, and USB). Cable length, cable type, and interface format will not cause an increase in jitter at the D/A chip in the DAC2. External jitter attenuation devices will have no positive or External jitter attenuation devices will have no positive or negative effect on the clock performance of the DAC2. effect on the clock performance of the DAC2.”
The metadata solution would be interesting as a complement to a player like Audirvana, JRiver, Amarra, etc., and would be even more interesting if it was open source or free software. I’d pay for a player that incorporated metadata improvements, but would support free software allowing more flexibility and control even more so.
Audirvana is a decent and relatively affordable player for someone with a large collection who replaced turntable, CD player and physical media for a macmini or RaspberryPi, but far from ideal. Overcoming business models that impose proprietary regimes on users and music lovers, would certainly make this a better world. This goes for the whole production, distribution and reproduction chain of our beloved music. In the meantime, offering just the software as an option, and at least as open source, would certainly be interesting.
It is gratifying that you appreciate the advantages of our Wax software for cataloging and enjoying a collection of digital music. However, you are confused about the functions of the 3beez BitScrubber board. We provide a list of its functions at our web site (http://www.3beez.com/comparison/sound_quality.html#bitscrubber). “Clock generation, clock recovery, and jitter attenuation” are not among them. In a nutshell, the purpose of the board is to optimize the digital outputs to maximize the likelihood that the receiving device will accurately interpret the data stream. We state in the first paragraph of that web page (http://www.3beez.com/comparison/sound_quality.html#The problems) that there are DACs that do a good job of protecting themselves from the noise and distortion that can interfere when connecting two components. However, not all do. A good example of a DAC that needs help is the Anthem D2V processor used inaptly for listening tests in the upcoming TAS review of our product (http://www.3beez.com/comparison/gloves_off.html#Not using audiophile-grade test equipment when testing audiophile-grade products). Providing outputs with the appropriate peak-to-peak voltage, impedance, and bandwidth helps assure that DACs can do their jobs properly.
Your concern (and Butch Patchell’s) about the “possible obsolescence or future lack of updates” of the software reflects additional misconceptions. First, our “metadata solution” is neither proprietary nor secret. Moreover, we provide an export function for encoding as much of the Wax metadata as possible in standard tags (http://www.3beez.com/manual/select.html?highlight=export#export). We also make it possible to export all metadata to an Excel spreadsheet (http://www.3beez.com/manual/config.html#export). These features assure that Wax users will be able to preserve their catalog if they need to move to another platform (and there are other uses as well).
As for the possible “future lack of updates”, we are just entering our fifth year of operation. We have continued to improve the software throughout that period and the latest software runs on every version of the product from Wax Box 1 through Wax Box 4. We respond to requests from users for additional features. Of course, all of this ends if we shut down tomorrow, but the same goes for software from Audirvana, JRiver, or any other company. Well-established companies like Apple, Microsoft, and Google regularly discontinue support for software. Open-source software dies when no one is willing to volunteer support. There are never guarantees.
You addressed your comments to Mark and John Siau. I do not know why you think that they would be able to respond. If you (or anyone else) have additional questions or comments, you are welcome to send email directly to me. There is a link at the bottom of every page at our website. You will need to use it as I will not respond further in this forum to this thread.