An Interview with Chris Connaker of computeraudiophile.com

Chris Connaker of Computeraudiophile.com

LB: Chris, welcome to Dagogo. Please begin by telling us about your background, and what lead you to embrace computer-based audio and to start computeraudiophile.com.

Chris Connaker: I have been an audiophile for as long as I can remember. In 1982, after talking my second grade teacher Mrs. Van de Weigh into allowing students to bring our favorite music to class, I brought the vinyl version of Pink Floyd’s The Wall. The low fidelity elementary school turntable didn’t reproduce the album very well but the class enjoyed the rebellious lyrics of Another Brick in the Wall nonetheless. My computer experience began that same year with my family’s purchase of a Commodore 64.

Jumping ahead a bit, I graduated Summa Cum Laude with a Bachelor of Arts degree from St. Cloud State University. My now-extensive background in Enterprise Information Technology began immediately after college. I joined the world’s largest hair care company as a Help Desk Analyst and worked my way to Network Administrator in a few short years. During this time I designed, configured, and managed global IT networks including desktop hardware and software, network storage (NAS, SAN), switches, routers, and firewalls. Broadening this Enterprise IT experience I moved to a Fortune 300 financial company where I managed the company’s utilization and consumption of technology. This included analyzing everything from mainframe computing to virtual machines to OC-24 Internet circuits.

In early 2007 I discovered that audiophiles everywhere were struggling to find information about music servers and the corresponding technology. I thus started Computer Audiophile in November of that year to fill this large knowledge gap in the audiophile community and to demonstrate to hesitant audiophiles that music servers can equal or better the current state-of-the-art in any playback medium. Within three years Computer Audiophile had grown to over 160,000 absolute unique visitors per month, from nearly every country on Earth. Over the years I have built many music servers from low power absolutely silent Windows computers to high power Macintosh-based machines. Some of these servers have been used by the audio industry’s leading manufacturers; for example I custom built the silent music server used in the TAD/Pioneer Electronics suite at the Venetian for the 2009 Consumer Electronics Show. The sound quality from the TAD suite was mentioned by many audio magazines and was selected as one of the “Best of Show” selections in The Absolute Sound’s March 2009 issue. In addition to writing articles, reviewing audio components, speaking at seminars, and hosting events such as the inaugural Computer Audiophile Symposium, I am a computer audio consultant to manufacturers, dealers, distributors, record labels, and consumers all over the world.

I currently reside in Minneapolis, MN with my wife Katie, our cat Chloe, and our two Chihuahuas Archie and Jayda.

LB: There is a widespread belief that computer audio is the wave of the future, and will soon replace the CD. However, for many of us computer audio is rather intimidating. As with many technological fields, computer audio has its own terminology, which adds to the confusion for the beginner. Perhaps you could help even the playing field for providing definitions – as non-technical as possible – for terms commonly used in the field.

CC: I look at vinyl playback the same way as traditional audiophiles look at computer-based playback. When I see a state of the art turntable I’m immediately intimidated and think I could never get such a complicated piece of equipment to sound good. Thus, I’m in a good position to understand the frustration, hesitation, and intimidation traditional audiophiles feel when thinking about this next phase of high end playback using music servers. Here are some easy to understand explanations for audiophiles. I don’t consider them definitions because too often a definition is very general; I would rather focus on explaining the technology and how it fits into our wonderful high-end hobby.

Music Server: Music servers come in many different forms. The two most common types of music servers are (1) personal computers and (2) canned systems like Sooloos, QSonix, and Olive. I could get very technical and split hairs as to what actually constitutes a server but that would only serve to confuse many readers. Personal computers usually run Windows, Mac OS X, or the Linux operating system. What many audiophiles might not know, or might not think about, is that the canned systems are also computers running the same operating systems and using the same internal memory, processor, and hard drives. The only difference is what is presented to the user as an option or graphical user interface (GUI).

Ripping: Ripping is the process of copying music from a Compact Disc to one’s music server. This slang term has become common nomenclature in the world of computer audio. Ripping can be done numerous ways including using applications such as iTunes, J River Media Center, dBpoweramp, or Exact Audio Copy.

NAS: Network Attached Storage (NAS) has been in the world of enterprise computing since the 1980s. Due to large music libraries and new high resolution audio releases that consume massive amounts of disk space, audiophiles are turning to NAS drives frequently. In the simplest terms a NAS drive is one or more hard drives that connects to a computer network via Ethernet cable (though not via common USB, FireWire, or eSATA cables), and that is accessed over this network. Once connected to a home network a NAS drive is accessible to every computer on the network and can appear like a regular drive on one’s desktop. NAS drives can hold incredible amounts of music. They usually start at two Terabytes in size and can increase to over 15 Petabytes (aPetabyte is one million Gigabytes).

Audio Card: An audio card simply provides a digital or analog interface to connect to one’s stereo. Back in the day audio cards were necessary when computers didn’t offer S/PDIF outputs or have an onboard Digital to Analog Converter (DAC). Now audio cards like the Lynx AES16 or ESi Juli@ can offer increased audio performance. These cards are physically installed inside one’s computer. If using a USB DAC (see below) there is absolutely no need for an internal audio card. The USB DAC becomes the “audio card” in that it shows up in the Windows Control Panel as a sound device and in Audio Midi on OS X as a sound device. An actual add-in sound card can be advantageous or even necessary if outputting an AES/EBU audio signal from a computer to an external DAC. Coaxial S/PDIF outputs frequently require an add-in audio card while many computers already have built-in TosLink outputs.

Metadata: In general metadata is data about data. In the context of audio, metadata is information describing an album, artist, track, etc. There are major differences in how metadata support is implemented in audio applications. The most important difference is between associated metadata and embedded metadata.

Associated Metadata: This metadata is frequently stored in a proprietary database or file used by playback applications. When looking at an album within the application, users will see all the information available such as album art, artists, track title etc. For example, when iTunes automatically finds album art it only associates this art with each track of the album. The problem with associated metadata is its lack of transportability. This metadata will only be available when using the specific application that associated the metadata with the files. If an iTunes library file is lost, or an application’s database of associated metadata is lost, or if a file with associated metadata is moved to another application, all the metadata is gone for good.

Embedded Metadata: This metadata is stored as chunks inside the audio file’s container such as AIFF or FLAC. Containers / file formats such as FLAC, AIFF, M4A (ALAC) support embedded metadata that is readable and writable by many audio playback applications. These containers/file formats have guidelines or standards for embedding metadata and they allocate space within the container for this data. Once this metadata has been embedded into a container/file like AIFF the metadata is there until removed. None of the three problems described above are an issue with embedded metadata. Loss of an iTunes library file or proprietary application database or moving a file to another application have no effect on the metadata. A file with embedded metadata in iTunes will display album art, artist, track title etc. without manually entering anything or without iTunes gathering the metadata from an Internet database.

Audio codec: An audio codec is simply an algorithm or program that COmpresses/DECompresses audio data. Codecs can be lossless or lossy. Each codec has its own method of compression and decompression. Thus the audio quality varies greatly from codec to codec. Some popular codecs are as follows:

Lossless

· Apple Lossless Audio Codec (ALAC)
· Free Lossless Audio Codec (FLAC)
· Windows Media Audio Lossless (WMA)

Lossy

· MPEG-1 or 2 Audio Layer III codec (MP3)
· Advanced Audio Coding (AAC)
· Windows Media Audio (WMA)
· Vorbis/Ogg Vorbis

WAV, FLAC, lossless, lossy, etc: First a little background information. 99% of music played on our computers is formatted as PCM (Pulse Code Modulation) audio data or sometimes as LPCM (Linear Pulse Code Modulation). LPCM is PCM with linear quantization. PCM and LPCM will be used synonymously in this description. PCM is a digital representation of an analog signal. Very few music playback applications support playback of raw PCM data. Playing raw PCM data may seem like a great idea to audiophiles but in reality it’s less than ideal. Raw PCM data contains only the digital representation of an analog signal; it contains no album art, nor metadata like artist, album or track information, nor information providing instructions to the playback application. In an effort to support file interoperability between many applications created by many companies, file Containers / wrappers were developed decades ago. A container simply describes the enclosed data to the application opening the file.

Think of a container as a CD or DVD case. The case describes what’s inside. It should contain a logo stating the enclosed material meets the Compact Disc, SACD, Dual Disc, or DVD-Audio standard. Liner notes inside the case frequently describe even more about the specific music contained in the case. The disc itself contains the actual music.

Container / wrapper = CD or DVD case
Metadata = Liner notes
Music data = Music on disc

Over the years many audio specific containers have been developed. These containers are frequently, and correctly, referred to as file formats such as AIFF, WAV, FLAC, and MP3. Each container or file format typically holds one of three types of audio data. The three types are uncompressed, lossless compressed, and lossy compressed.

Uncompressed: Uncompressed PCM audio is stored as a one for one copy of the original. Popular containers for uncompressed PCM audio are AIFF and WAV. Most people refer to AIFF and WAV as file formats and that’s also 100% correct. (Note: These uncompressed file formats are not the same as codecs.)

AIFF is an acronym for Audio Interchange File Format. The format was developed by Apple in 1988 as an extension of the IFF format created by popular video game creator Electronic Arts. AIFF files were originally used on Macintosh computers as the equivalent of WAV files on Windows-based computers. Today AIFF files are supported by most popular music playback applications on Windows and OS X. AIFF files support embedded metadata such as album art, artist, and track title. Many popular playback applications can read and write embedded metadata in AIFF files. AIFF and AIF files are exactly the same. Using the technical description from above, the AIFF container / wrapper stores the metadata (album art, artist, track etc.) and any additional information required, and simply holds the uncompressed PCM audio data inside this container. During playback the AIFF container is opened by the playback application to access the uncompressed PCM audio data.

WAV is a short name for Waveform Audio File Format. WAV was developed by Microsoft and IBM and released in 1991. WAV files are supported on nearly every operating system and turnkey music server available. Contrary to popular belief and experience WAV files can store metadata. However, very few playback applications can read or write embedded metadata in a WAV file. For example, iTunes cannot embed album art into WAV files. Using the technical description from above, the WAV container / wrapper can store metadata (although not in most end user systems) and simply holds uncompressed LPCM audio data inside this container. During playback the WAV container is opened by the playback application to access the uncompressed LPCM audio data.

Lossless Compressed: Lossless compression involves a codec and a container/file format. The codec and container usually go hand in hand. An audio codec is simply an algorithm or program that COmpresses/DECompresses, frequently PCM, audio data.

Popular combinations are:

Codec – Apple Lossless Audio Codec (ALAC)*
Container/File Format – M4A
Codec – Free Lossless Audio Codec
Container/File Format – FLAC
Codec – Windows Media Audio Lossless*
Container/File Format – WMA
* WMA and ALAC are proprietary codecs. FLAC is open and royalty free.

The above combinations usually start with an uncompressed audio file, remove the container (AIFF, WAV, or cda directly from a CD) compress the PCM audio by removing redundant and predictable data then package it into a container such as M4A, FLAC, or WMA. During playback of a lossless file the container is opened, the PCM audio data is uncompressed and reconstructed into the identical PCM audio data that existed before it was compressed. There is much debate in the audiophile community about whether lossless compression results in deteriorated sound quality. The debate centers around the playback application and computer’s ability to decompress a lossless file on the fly during playback. Regardless of one’s position on this issue the fact remains that lossless compression as a data storage file format does not alter the original data. For example a WAV file can be compressed into FLAC and decompressed back into WAV countless times without altering the original data. The original WAV file is identical to the decompressed WAV file. Taking it one step further, it’s entirely possible to start with a WAV file, compress it into a FLAC file then decompress it into an AIFF file without altering the PCM audio data. The container will have changed but the encapsulated PCM audio (song) remains the same. Two main benefits of lossless compression are smaller file sizes and really good support for embedded metadata.

LB: An obvious advantage of computer audio over “traditional” digital playback (i.e., using a CD player, or an optical transport connected to an external DAC) is convenience: One can readily access thousands of songs via menus, rather than having to sort through large numbers of CD jewel cases which for some of us (ahem…) are scattered around one’s room rather than being neatly stacked in racks. Is this the main selling point of computer audio, or are there also sonic benefits?

CC: As an audiophile I view the convenience as a side benefit. I hesitated entering into high end computer playback until I was satisfied I could equal or better the performance of traditional disc spinners. As a music aficionado I view the convenience a great selling point.

LB: Whereas some computer audiophiles build their own machines, less savvy individuals have the option of buying pre-assembled machines whose sole function is music playback. One obvious difference between the two approaches is that with the pre-assembled music server, one is locked into using the internal DAC. Are there other difference s as well?

CC: I touched on this a little bit previously. The only real differences are the options presented to the user and the user interface. Canned music servers like Olive and Sooloos both offer digital output to an external DAC. Canned systems tend to limit the number of options visible to end users and present everything in a very aesthetically pleasing manner. These limited options and beautiful interfaces are usually much easier to use for non-technical audiophiles. However, someone with the know-how can set up a personal computer-based system that is very similar to most of the canned music servers. The price of canned music servers isn’t necessarily more or less than a personal computer used for audio playback. I’ve built custom music servers costing $7,500 or more whereas a Sooloos MC200 is close to half that much money.

LB: For those readers who would like to build their own computer-based music system, let’s run through the various steps. First of course is the computer itself. Will any computer do, or are there specific requirements? And at the risk of bringing to the pages of dagogo the never-ending battle between Macs and PCs, what are the plusses and minuses of each for computer audio?

CC: Computers used for audio playback are on the same continuum as all other high end components. It’s possible to purchase a pair of Audioengine A2 loudspeakers for $199 and it’s also possible to listen to the same music through a pair of $40,000 TAD Compact Reference CR1 loudspeakers. All computers are capable of outputting the exact same bits (ones and zeros). However, not all computers will be acceptable to audiophiles seeking the highest quality playback. For example, a legendary audio engineer recently told me about his experience measuring the impact a computer graphics card can have on digital audio output. This engineer measured drastic increases in jitter on the digital output of an internal audio card when a high powered graphics card was placed in close proximity to the audio card inside the computer’s case. Another example of differences between computers is frequently seen even with external USB DACs. I have an older MacBook Pro (Pre unibody) that will not function correctly as a music server if the USB DAC is connected to the rear left USB port. When connected to this port my audio system suffers annoying dropouts. The reason this happens is related to how this MacBook Pro was designed. The rear left USB port shares the same bus (think of a bus as an internal pathway) as the keyboard. Mac keyboard illuminate the letters from underneath the keyboard. After twenty seconds or so of non-use the illuminated keys go dim. When the keys are touched again they illuminate and cause audio dropouts if a USB DAC is connected to the rear left USB port. An external hard drive or printer connected to this port would function totally fine due to the way these devices communicate. Files copied to an external hard drive are sent in packets that can be resent if not received by the external drive. This happens invisibly to the end user. Not all computers are created equal when it comes to high performance audio playback.

The whole Mac v. PC thing is always blown way out of proportion. I always tell people to use whatever operating system or server they are most familiar and/or comfortable with. It’s possible to obtain equally impressive sound quality and user experience using Mac OS X or Microsoft Windows. Heck, even a Linux based system can be configured to work as easy as a toaster.

LB: Are there any special requirements for the operating system?

CC: Technically no. It’s possible to get sound out of almost any operating system. I prefer Windows 7 over Windows XP because v7 has more audio output capabilities such as Windows Audio Session Application Programming Interface, commonly known as WASAPI. This is a big advantage because output modes such as WASAPI and WASAPI Event Style enable users to send the audio output stream directly from a playback application like J River Media Center to a connected audio device such as a USB DAC or internal audio card. WASAPI Event Style lets the audio subsystem pull data instead of pushing data to the system, thereby allowing lower latency buffer sizes; it also removes an unreliable Microsoft layer. In layman’s terms this is about a much more direct data path between software and hardware. Many users are unknowingly sending their audio output through a maze of software layers. It’s well documented that these layers use sample rate conversion and adjust output volume levels, resulting in data coming out of the computer that differs from that which was ripped from the CD. Thus, it’s no longer bit transparent. Audiophiles, myself included, spend a sizable amount of money improving their audio hardware yet many don’t realize there are free software improvements to be made.

I also prefer the newest versions of Mac OS X 10.6.x. Starting around version 10.6.4 OS X started correctly supporting Class 2 USB audio, which enables USB DACs to handle up through 24 bit / 192 kHz music without the need for additional software installation. The DACs are plug n’ play.

LB: Let’s turn now to ripping. I know there are many options here: uncompressed vs. lossless, WAV vs, FLAC, etc. Please share your thoughts on this process.

CC: A while back I wrote an extensive article called the Computer Audiophile CD Ripping Strategy and Methodology. I would love to provide the executive summary of Cliff’s Notes, but people should really read the whole article to understand why I make the recommendations that I do.

http://www.computeraudiophile.com/content/Computer-Audiophile-CD-Ripping-Strategy-and-Methodology

LB: Perhaps less interesting but certainly quite important is data storage. What do you recommend in terms of capacity and redundancy (i.e., back-up)? And what are your thoughts regarding solid state storage vs. hard drives?

CC: Data storage is critically important to computer audiophiles. It’s impossible to recommend any specific capacity as all users have different storage needs. I have about 4,000 albums ripped to my five Terabyte NAS drive. The NAS is configured as a RAID5 disk array, which allows one drive to completely fail without causing any data loss. I usually recommend RAID5 as it’s a great compromise between cost, capacity, speed, and simplicity. RAID is a form of redundant storage but should never be confused with backup. All users must have a backup copy of their music on another drive(s), preferably off-site. The chances are pretty high that one day a catastrophic disk failure will occur or a user will accidentally delete his music. RAID will not help in this type of situation. A completely separate backup copy of one’s music files can save the day, or many months of re-ripping CDs.

LB: An area that seem to generate considerable discussion – and controversy – is the software used for playback (and for ripping). I often read about programs such as Foobar, Media Monkey, J Rivers, iTunes, Pure Music, and Amarra, amongst others. How do these programs differ from one another? In particular, are the differences solely in terms of the user interface, or are there (as many claim) sonic differences as well?

CC: If you ask five engineers a technical question you’ll likely receive ten answers. It’s the same with audiophiles and playback software. One great thing about software is the ability to try it in one’s home for free.

LB: Which brings us to the DAC, and how it is connected to the computer. Prior to the advent of computer audio, all but a few DACs connected to the optical transport via S/PDIF (or less commonly, AES/EBU). If I understand correctly, such connections require the computer to be fitted with an appropriate sound card. So the first question is, what does one look for in a soundcard?

CC: First and foremost a sound card must feature the correct digital interface for the DAC. A sound card with only coaxial S/PDIF output will be of no use for a DAC with only an AES/EBU input. Second, the sound card must be supported by the operating system used on the music server. Some cards work with Mac OS X only. Third, I recommend cards that support good audio output modes such as ASIO, WASAPI, or Kernel Streaming. Other items to consider that can increase audio quality include the ability to receive a word clock signal from a DAC with word clock output or the ability to run in Dual Wire AES mode for DACs from manufacturers like dCS.

LB: The latest buzz in computer audio is USB DACs – in particular, asynchronous USB DACs – the primary advantage of which is said to be reduced jitter. With the understanding that a thorough description of USB and jitter is well beyond the scope of this interview, please tell us a bit about this technology.

CC: In general USB DACs are either asynchronous or adaptive. Neither technology is always better than the other. It all comes down to implementation by a skilled engineer. Asynchronous USB transfer mode should never be confused with Asynchronous Sample Rate Conversion (ASRC); these are two very different concepts. The main selling points of asynchronous USB transfer mode are clocking superiority and data flow control. An asynchronous USB DAC completely controls the clock rate (44.1, 96, 192 kHz etc.) using fixed crystal oscillators, and completely controls the flow of data (music) between the DAC and the computer sending the data. These two capabilities, when implemented well, can make an asynchronous USB DAC very tough to beat if all else is equal. On the other hand adaptive USB doesn’t control the master clock or the data flow from the computer. Adaptive USB DACs must adapt to the incoming clock signal and update every 1ms. Since the USB bus operates at 12MHz and an audio signal such as 44.1 kHz requires 11.2896MHz, an adaptive USB DAC must use a frequency synthesiser to create the correct master audio clock signal. Adaptive USB architectures can require jitter reduction techniques that also reduce sonic quality. That said, in the hands of a good engineer an adaptive USB DAC can sound wonderful. Readers will do themselves a disservice to select a DAC based solely on USB transfer mode.

LB: Do all computers have USB connectors? Are there special types for audio?

CC: Nearly all computers produced in the last ten years have at least one USB port. Universal Serial Bus is exactly that, universal. There are no special USB specs for audio. There are a few different sizes and shapes of USB connectors but these all have designations such as A, B, Mini, and Micro. All USB DACs use the standard A connector to connect to a computer while using a standard B or mini connector on the DAC itself. Most USB DACs support either USB version 1.1 or both versions 1.1 and 2.0. USB 3.0 was released in late 2008 but has yet to gain critical mass consumers. In terms of USB DACs there is no known advantage to using USB version 3.0.

LB: Some DACS have both S/PDIF and USB inputs. Is it necessarily the case that the USB will sound better or – as with most things in audio – is implementation a critical factor?

CC: It’s all about implementation. For example a USB input that’s not galvanically isolated from a noisy computer can sound far worse than an optical S/PDIF input on the same DAC due to optical’s inherent electrical isolation.

LB: For those who have “legacy” DACs that lack USB inputs, there are adaptor devices such as those form M2Tech and Audiophilleo that allow one to use the computer’s USB output . How do this approach compare to using a soundcard and S/PDIF input on the DAC?

CC: The only difference that’s 100% true with all these adaptors is they are external to the computer. Other than that the sound quality and functionality are all over the board.

LB: Still another means of connecting a computer to a DAC, and one that does not require the DAC to have a USB input, is via ethernet, the most popular example being the Squeezebox products. Are the computer requirements as stringent as for computers connected directly? And how do the two compare sonically?

CC: I’ve used Squeezebox products and Ethernet based components from Linn quite a bit over the years. It’s been my experience that the computer requirements are nothing like the requirements for using a device directly connected to a DAC via USB. Computers sending audio over Ethernet simply need to send the data in a timely manner. That said, the receiving devices (Squeezebox, Linn, etc.) operate much more closely to an actual computer than more traditional DACs by buffering and reassembling TCP/IP packets sent over the network. The sound quality of Ethernet audio devices is all over the board just like every other audio technology ever invented. I don’t consider any wireless solutions available at the time of this writing to be true audiophile solutions.

LB: I’d like to briefly return to a point discussed earlier. Computer audio brings great convenience to music playback, though the system setup is considerably more complex than hooking up a CD player (or even an optical drive and DAC). But for those of us into high-end audio, the bottom line is sonic quality. Based on your considerable experience with computer-based systems, is it reasonable to expect results comparable to those with optical drives?

CC: It’s not only reasonable, it’s expected. There are many accepted engineering reasons why computer based audio is capable of better sound quality than a spinning disc player. For one, a disc player usually has one chance to read the variable speed optical disc and engage its error correction and interpolation. A computer does all of this in the CD ripping stage and reads the disc as many times as required to get the data off a CD. Readers have likely noticed that some CDs take much longer to rip than others. This is because many reads and possible error correction algorithms have been use to capture the correct data. This just isn’t possible with optical CD players / transports.

LB: Last but not least, where do you see computer audio heading in the next few years, and how do feel it will impact high-end audio?

CC: Physical discs are history. Analog tape and vinyl will soon be the only physical media available. Computer-based audio is currently supplementing many high end systems and will completely replace all digital front ends in the not-too-distant future. High resolution downloads are becoming more available every year. The Rolling Stones are about as big as a band gets and The Stones’ music is available for download via HDtracks. Direct Stream Digital (DSD), not to be confused with SACD, is also available via download from Blue Coast Records and hopefully additional labels soon. There are more than a few DSD-capable DACs coming out as well. Convenience and sound quality are not mutually exclusive anymore. Computer-based high end audio simply makes sense.

Also, people should not get caught up in the one technology over all others game. By this I mean there will not be a single type of music server that will outlast all others or a single type of DAC, be it USB or Ethernet, that will end up “winning” as Charlie Sheen would say. It’s impossible to pigeon-hole every music aficionado or audiophile into one camp such as Sooloos, Mac OS X, USB, Ethernet, iPhone remote, tactile remote, etc. Everything is temporary if given enough time.

LB: On behalf of the dagogo readers, I would like to thank you for taking the time to so patiently share your knowledge and experience with us. I hope you will come back and visit us again.

CC: Thank you very much for the opportunity, it’s been my pleasure. This is a wonderful hobby. I relish every opportunity to help improve people’s listening experiences.