Audio Compression Shootout

MP3 is now a distribution standard. iTunes has sold over a billion compressed audio tracks in AAC format. DVDs, Blu-ray Discs and digital TV all use audio compression.

Besides these, many other compression formats are available to the video producer. Some use lossy compression like MP3 and others are “lossless.” The trick is determining which are useful or even practical. This month, we’ll compare the major players and see how they fit in the production process.

Lossy Formats

The January 2008 issue of Videomaker has an in-depth article on lossy audio compression. Here’s the thumbnail version: Lossy compression discards audio data based on a model of human hearing. Using some fancy math, the encoders can dramatically reduce the size of the file with a minimum of audible impact. What follows is a brief outline of popular lossy formats.

What we call MP3 is actually short for MPEG-1, Audio Layer 3 encoding. It is one of three audio compression methods designed for MPEG-1 video streams. Conveniently, the system also works without video and became popular for distributing and playing CD audio tracks in the late 1990s.

The quality of MP3-encoded audio has been the subject of great debate over the years. Early encoders produced poor-quality files, but modern encoders can sound quite good. The Fraunhofer encoder is the de facto standard included in mainstream audio applications like Adobe Audition and Apple’s iTunes. Many people currently regard the LAME open-source encoder as the highest-quality MP3 encoder, but the title isn’t permanent. There is also an MP3 variant called MP3Pro, but so far it hasn’t gained mainstream acceptance.

MP2 is a lossy compression method originally designed for digital broadcasting. As a sort of predecessor to MP3, it requires slightly higher bitrates for encoding, but it is quite reliable as a streaming format. In fact, much of the music you hear on analog radio is encoded with MP2. Remote production and satellite feeds often use the MP2 format. Good luck finding a portable player that reads MP2 files.

We can’t forget the MPEG-4 format and AAC, Advanced Audio Coding. AAC is the format of choice for Apple’s iTunes store and the venerable iPod. Playstation 3 and Wii game consoles, along with some digital radio services, also use AAC. On paper, AAC is dramatically better than MP3, with greater flexibility and careful attention paid to various encoding options. In the real world, consumers seem more interested in compatibility and, given similar sound quality, prefer an audio format that works on their existing devices. Apple has certainly sold a ton of AAC tracks, but they’re not widely used in other applications.

Microsoft has its own proprietary format with Windows Media Audio or WMA files. WMA sounds great and is quite flexible but, like AAC, hasn’t gained widespread acceptance in hardware or software circles.

The final lossy format is Ogg Vorbis. The name may sound funny, but many audio experts believe that Ogg files sound better than other lossy formats at similar bitrates. Ogg Vorbis is deeply rooted in the open-source software movement, and there are no licensing fees or other purchases necessary to use the compression method. While not exactly mainstream, Ogg Vorbis continues to gain ground in software acceptance, and there are a few hardware players that support the format. This is a format to watch.

Lossless Formats

If you’ve downloaded much from the internet, you’re familiar with Zip files. These files contain one or more other files, bundled together and squeezed for absolute minimum file size. The magic happens when you unzip them – all the files reassemble perfectly in the original form. That’s the way lossless audio compression works. Audio files are fairly large, with an average song taking up 30-50MB of storage. Lossless encoders strip out the redundant data, silence and anything else they can find. Then, they make a list of what they extracted and create a new, smaller version of the file. While a lossy encoder might achieve a compression ratio of 10:1, lossless encoding is typically 50-75% of the original file size. This may not seem like a serious reduction, but it’s convenient for archiving and collaborating over the internet, with a complete, perfect copy of the original material.

There are several lossless compressors on the market. The one everyone’s talking about is FLAC (Free Lossless Audio Codec). Another open-source project, FLAC is supported in several popular audio and video applications, with more on the way. FLAC can use multiple bitrates and sampling depths and will squeeze up to eight channels in a single file. WavPack uses different methods with similar results, including the option to create a “lossy” version of the file for immediate playback. Unfortunately, the production world does not widely support WavPack, but this could change. Monkey’s Audio lossless APE format isn’t as full-featured as FLAC or WavPack, but it produces similar results. Support is limited for this newcomer.

Not to be outdone, Apple and Microsoft each have lossless audio encoders tied to their proprietary platforms. Apple Lossless encoding works much the same as other lossless compression and has support primarily in Apple software. Microsoft’s WMA Mathematically Lossless format is another lossless method, with full support in Microsoft software, Zune players and some production packages.

Real-World Testing

We couldn’t find one audio or video application that supported all these audio formats, but we discovered Cockos’ Reaper would import and export most of the batch. So, we took an original audio mix and rendered it in MP3, OGG, FLAC, APE and WavPack and created an uncompressed WAV file for comparison. Using our WAV file in iTunes, we created an AAC and Apple Lossless version. In Adobe Audition, we generated WMA and Windows Lossless files. We rendered MP3, OGG, AAC and WMA files at 192kbs. Using Reaper again, we placed the WAV, MP3, OGG, FLAC, APE and WavPack files on individual tracks and used the solo function to listen to each separately. The first thing we noticed was an offset in the MP3 track; it was several milliseconds behind the other tracks. Interestingly, its total length was slightly shorter than the other files. This created a slight but noticeable timing problem when compared to the other tracks. In addition, the MP3 file was slightly duller-sounding, and the bass was a bit muddled. The OGG file fared slightly better, with excellent timing and a clearer, more open sound. We compared AAC and WMA files directly with the original. Each was on a par with the quality of the OGG version, but each exhibited its own unique compression artifacts. Comparing compressed audio codecs is a bit like deciding which fast-food chain makes the best hamburger. They all have the same standard elements along with their own twist. There is no clear winner, since everyone always has different tastes.

Since the other files claim to be lossless, there should be no audible difference between them and the original track. That was our experience, although, admittedly, this was a purely subjective comparison. However, based on our tests, it’s safe to say that the lossless formats accomplish their purpose admirably.

One of our goals was to see how an average video producer might leverage these formats in a production, so the next step was to import them into popular NLEs. Sony Vegas scored the highest, with support for all the formats except APE, WavPack and the Apple files. Adobe Premiere wouldn’t import FLAC, APE, WavPack or Apple Lossless, but it worked fine with AAC files. Apparently, Apple is the only vendor that supports its proprietary lossless format.


Conclusions

Ultimately, the question is this: why would you go to the trouble of using compressed audio formats in a video production?

One reason is file size. Compressed formats can quickly cut a huge audio file to a fraction of its original size. This makes internet collaboration possible and can save volumes of storage space. At their best, the lossless formats achieve only a 50% reduction in file size. In our tests, they all squeezed a 35MB file into roughly 24MB – not exactly a huge savings, given the size and price of storage today.

If you choose to use compressed audio, keep in mind the reduction in quality from lossy formats. There’s also the possibility that there may be no support for certain formats in the future.
In the final analysis, it’s up to you. If your software supports it, why not? You’ll save some storage space and live on the cutting edge of audio production.

Contributing Editor Hal Robertson is a digital media producer and technology consultant.

Did you find this content helpful?

Videomaker
The Videomaker Editors are dedicated to bringing you the information you need to produce and share better video.