POST-MASTERING
Making Your Music Ready for the Digital Marketplace
Editor’s Note: For this month’s DIY feature, we are pleased to hand over the reins to mastering engineer Mike Wells for an inside look at preparing your music for digital distribution. As you’ll read, control over the fidelity of your recordings is changing hands from you and your mastering engineer to the big digital distributors. Wells takes you through the process of getting your music sold online while hopefully maintaining the best possible audio fidelity in the process.
So you’ve made it to the finish line: You’ve recorded, mixed, mastered, and replicated your project for release into the marketplace. But have you made it to the end? A new bar has indeed been set — and that is encoding for digital distribution. With the addition of digital distribution as a standard part of the release process, new questions have emerged: Who is managing the audio quality of these digitally encoded assets? When the final product that reaches the consumer might be nothing more than a highly compressed, low bit-rate mp3, how are artists and labels either empowered to deliver their own assets or verify the audio quality of encoding done by an outside service?
The State of the Digital Distribution Marketplace
Let’s consider the most popular current form of physical music distribution — the compact disc. The music is recorded, mixed, mastered, and mass produced, resulting in this product. The product enters the distribution channel through which it then reaches retailers and, finally, the consumers.
Within this business model, there has always been a clear distinction between those who monitor the fidelity of the product (the recording, mixing and mastering engineers along with the artists and labels), and those who is monitor the distribution of the product (the distribution companies).
But with the focus shifting to digital distribution, the fidelity of recordings is up in the air all the way through the distribution process. Who’s monitoring it? What kind of product is ending up in consumers’ hands after going through the distribution channel?
With the majority of digital distributors handling the encoding process (turning physical CDs into digitally encoded computer files), the task of monitoring fidelity has been removed from the creators and placed upon the distributor. Should you be concerned?
I spoke with three major players in the indie digital distribution marketplace — the Independent Online Distribution Alliance (IODA), The Orchard, and CD Baby — about their policy on accepting songs (“assets”) that were already encoded by artists or labels. The feedback was consistent. The simple answer is no.
Why? In short, it’s an issue of standardization and quality control. The leading digital distributors support 40-plus online retailers each. Each retailer has unique requirements for what kind of digitally encoded assets they’ll sell via their web storefront (generally concerning file format specifics like .mp3 files vs. .aac files, 128kps vs. 192kps). That means that each song will need to go through over 40 file conversions. So, if you were to perform your own encodings for your release of 10 songs (let’s assume for the moment that you have the format requirements for each retailer), you would be responsible for delivering upwards of 400 files to your digital distributor. Now multiply that by any number of releases that these distributors are getting every day to convert into digital assets for distribution. According to them, supporting proper quality assurance methods on their end of this process is simply too much work.
If you feel very strongly about supplying your own encoded digital assets for distribution, it won’t hurt you to ask them if it’s possible and see if they will consider it for your project. There are cases — given strong trust of the mastering engineer or your product in general — where it’s happened, but distributors say these are extremely few and far between.
“Ok,” you say, “so if I can’t supply my own encoded assets to a digital distributor, how are they getting all these files converted and maintaining the best fidelity?”
The companies I spoke with all use the same process—transcoding. Transcoding is the process of going from one encoded format to another (for example converting an .OGG file to an .MP3 file). The current marketplace appears to favor the open-source, lossless codec “.FLAC” (Free Lossless Audio Codec) as its parent archive format, where the physical CD is encoded into a lossless format (.FLAC). From there, the files are transcoded into every format required for each retailer through an automated batch process. Various system tests are performed to ensure the resultant encoded files play correctly, and some companies offer a mini-review process to their clients to allow them the ability to assist in the system test process. All parties stated, however, that actually listening to every version of every song would be logistically impossible.
Does this give you the willies? Most people say no. There are arguments all over the internet debating the lowest common denominator of encoding where humans can “hear” the effects of lossy encoders. 192kps is a popular threshold, with some parties feeling strongly that even 128kps is acceptable to the majority of consumers. Don’t forget, it wasn’t long ago, in the early days of internet radio, when listeners were happy with the quality of broadcasts on Real Player bitrates of 96kps.
Additionally, all parties feel that as internet bandwidth increases, the encoded bitrates of distributed assets will rise and eventually be above the average and discerning listeners hearing threshold (possibly even distributing lossless formats) and become a moot point. An example of this model exists today with iTunes, where iTunes Plus delivers encoded assets in 192kps rather than the previous standard, 128kps. Perhaps we will see a phenomenon similar to the “re-mastering” craze of the 1990s, where entire back catalogs of music were re-mastered and re-sold to the public, this time with the material simply re-encoded at a higher bit rate.
Controlling the Fidelity On Your Own
Some artists may want to have a tighter grip on the fidelity of your digitally distributed assets. If you find yourself in this camp, you have a few puzzles to solve:
• Where to distribute and/or sell your encoded assets
Finding digital distributors that allow you to manage your own asset quality is easier than you think. Players such as Snocap allow you to upload your own digital assets and either create your own storefront (e.g.: Snocap) or sell songs on the online retailers’ own storefront.
If you are managing your own storefront via a service such as Snocap, you can determine what encoded formats you want to support and sell. If you go with one of the many services that allow you to upload and sell digital assets through a retailer’s site, you will need to obtain the encoder requirements for each store/service.
• How to perform and evaluate encoding results
Encoding services do exist, including some mastering studios, which offer a more boutique and personal approach to “qualifying” the encoded results. Learning how to do this yourself is also within the realm of possibilities. The tool that offers the most control over the lossy encoding process is the LAME codec. This encoder offers so much control that it can initially seem daunting to new users, especially those unfamiliar with Windows/UNIX command windows and abstract programming language like “passing parameters,” “executables,” and “arguments.” A fantastic resource for becoming familiar with the LAME encoding options and to find friendly responses to your questions is the Hydrogen Audio forum at www.hydrogenaudio.org.
Other popular distributed codecs are the .AAC/.M4A format, with Nero offering an excellent free encoder as well. Additionally, there are a variety of user interface “shells” that can “drive” the encoders from a more familiar windows-style interface. Popular shells are the db PowerAMP Music Converter (www.dbpoweramp.com) and the EAC application (www.exactaudiocopy.de). Additionally, tutorials for each along with LAME and NERO commandline encoders all exist at the Hydrogen Audio website.
Analyzing the resulting fidelity of your encodings is a controversial and sometimes religious topic. There are all kinds of tools for objective quality analysis, but the audio community seems to be leaning more and more toward the opinion that our best tools are, simply, our ears. Due to the various implementations of psychoacoustic models for every encoder, combined with the material being encoded, it’s becoming apparent that frequency analysis alone is not the best method of measurement.
• How to Find Optimal Mastering Environments
Mastering engineer Seva once said “I can’t teach you how to hear, but you can teach yourself how to hear,” regarding learning how to hear effects of wideband compression artifacts after mastering. The same can be said for temporal masking effects as heard in lossy encodings. If you spend the time listening to the effects of each lossy encoder, you will become familiar enough with their artifacts to subsequently “tune” the encoder to get the best results, individually, for every track you encode. Like with mastering environments, higher-quality playback systems will make it easier to hear artifacts. Subsequently, it will be easier to tweak the encoder settings for any given track. So, to follow the advice often given for DIY mastering, get to know how your monitors translate to the outside world In this case, it’s the downstream player — the iPod. Learn how your encoding decisions translate onto those systems-it’s through those ear buds that people will ultimately be listening to these tracks.
Another major hurdle in accurately analyzing your encoded files is your own bias. You will most likely be going through several drafts of your songs, changing parameters and trying out the preset configurations (contrary to your independent will, presets are often the best place to start). But how will you truly know which one of your versions sounds better? The very act of double-clicking a known file can influence your decision by giving away too much about the file.
Enter ABX Testing, a kind of blind taste test mediator for you and your files. ABX testing is a method of performing listening tests where you are not aware of which file is under review. Additionally, another person can be used to “drive” the test while another listens and makes choices on quality (this latter method is often called double-blind ABX testing).
The Foobar application (www.foobar2000.org) offers the ability to perform ABX testing directly within its user interface, or you can see the PCABX site (www.pcabx.com) for info on ABX tools for all platforms. Adding ABX testing to your encoding environment can greatly help in educating yourself and understanding what your ears are telling you.
Encode Your Own Or Leave It to the Distributor?
So after all this the final questions become: Should I create my own web storefront, encode and sell my digital assets? Should I just go with a digital distributor? Both? There is no easy answer to this question. As with the music itself, different artists and labels feel more and less strongly about the effects of lossy encodings and what is in the distribution channel. As an independent, becoming educated about encoding your own material will always be in your best interests as new opportunities in this quickly-changing marketplace occur, whether you decide to take control of them or turn them over to distributors for now. Who knows — maybe you will think of the next great idea using them!
Mike Wells is an audio mastering engineer in San Francisco, California. Outside the mastering studio he enjoys motorcycles, working with synthesizers, teaching, and annoying software developers.
He can be reached mike@mikewellsmastering.com.
Glossary
-Assets
“Assets” are analogous today to what “Product” was in the ‘90s, where CDs would be “product” and their digitally encoded file counterparts are “assets”. A 128kps mp3 file of your song would be considered “1 digital asset” in the world of digital distribution.
-Bit Rate
“Bit Rate” is both a measurement and a description of the measurement. In the digital music encoding world, the current unit of measurement for bit rates is kilobytes per second, or kps. The types currently available are Variable Bit Rate (VBR), Constant Bit Rate (CBR), and Absolute Bit Rate (ABR). An example being a 192kps VBR encoded asset. Different combinations of Bit Rate type and measurement are used in different situations.
-Codec
“Codec” is short for “Encoder/Decoder.” For each song, you will need both an mp3 Encoder (to create the mp3 file) and an mp3 Decoder (to play the file). The majority of codecs support various bit rates (measurements & types) to allow a variety of different encodings of the same original file. The basic goal of a codec is to reduce the file size as much as possible with minimal loss in fidelity.
-Lossy/Lossless Encoding
“Lossy” encoding has traditionally been the most popular method of encoding and involves the use of a psycho-acoustic perceptual model to “mask” certain lesser-used frequencies within the song being encoded. Lossy encoding is best thought of as “audio compression” and “data compression” combined. “Lossless” encoding, by contrast, is the application of only data compression upon an audio file, resulting in a smaller filesize, but no degredation in fidelity.
-Artifacts
“Artifacts” are unwanted sounds in your song created by the encoding process. This can mean audible “pops” or “clicks” heard during playback of the song. With lossy encodings, you will often hear things like watery high end or zippering effects. Careful encoding can minimize artifacts heard during playback.
RESOURCES
ABX Testing:
http://www.hydrogenaudio.org/forums/index.php?showtopic=16295
http://en.wikipedia.org/wiki/Codec_listening_test
http://www.pcabx.com
Foobar:
http://www.foobar2000.org/
LAME MP3 Commandline Encoding tool:
http://lame.sourceforge.net/
Nero AAC Commandline Encoding tool:
http://www.nero.com/nerodigital/eng/Nero_Digital_Audio.html
|
