About two years ago, I set out to create a better-sounding MP3 file. I tried all the different encoders, bit-rates, and technical options. To my ears, there wasn't much of a difference – they all sounded bad.
As a mastering engineer, it disappointed me to hear the musicians' hard work end up like this. In the end, it didn't matter why MP3 was technically inferior; all that mattered was that it didn't sound as good as the 16-bit 44.1 kHz source, not to mention the 24-bit masters from which the CD was made. I concluded, then, that encoding a better MP3 was impossible. So now what?
MP3 was not going away. Even now it's still the most-used and player-compatible lossy compressed format for audio, and I imagine that it will be for some years to come. So, if a better MP3 through improved compression and encoding is not possible, is there something else that can be done? The answer is yes. Create a better mix.
Creating a better mix creates a better MP3. Yes, of course this is obvious – but maybe not for the reasons you might think. Just like engineers in the ’60s, ’70s, and ’80s mixed with the limitations of lacquering (the vinyl record master), I believe mixing engineers should mix with some degree of awareness of the limitations of MP3 since it's the way most people will listen to recordings. The good thing about this is that mixing with MP3 in mind will also create better mixes for CD or high-resolution production. Here is a short list of tips with explanations:
Limit your limiting. Digital peak limiting during the recording or mixing process raises your noise floor, reduces your dynamic range, and adds to your overall distortion.
Use your available bandwidth. Mix to 0db. That means your peak should be at or near 0dB. For every bit of “headroom” you leave, you lose perceived bit resolution.
Record at the highest bit rate and sample rate possible. As Rupert Neve once pointed out, harmonics of higher frequencies exist in lower frequencies. Recording at higher sample and bit rates, even if the final source is 16-bit 44.1 kHz compressed MP3 files, will still sound better.
Avoid dithering. Despite what people say about dithering, it really only sounds good when added at the end conversion from 24 to 16. Dithering adds noise and noise doesn't compress well.
Be aware of the MP3 compression process. Mid-range frequencies (vocals, guitars, snare drums) compress well. Everything else, especially broadband noise (bass, cymbals, air frequencies), are difficult to compress and can distort your MP3. Use this knowledge, especially in the mastering process, to shape some of your EQ decisions.
Monitor the side (difference) channel. Make sure it's not distorting. Distortion in the side can really mess up the compression process. To monitor the side channel, invert the phase of either the left or right channel of the master bus and then put it in mono. The resulting audio will be what you want to make sure is not distorting.
Use a mastering studio that understands the MP3 process. There are limitations of MP3 that go beyond the audio quality. If you take one song and download it from different sources (legal and non-legal), you quickly realize that there is no standard to MP3 creation. They go from bad to worse. Beyond that, the metadata (the artist information that's embedded into the file itself) is not consistent. Lacquer Channel (my mastering studio) launched enhancedMP3 in January 2009. It's a CD-ROM portion on the audio disc (much like enhanced CD) that contains the highest-quality, artist-approved 320 KB MP3 files available. We use a custom proprietary process to ensure the file compresses the cleanest way possible. Read more about it at www.enhancedmp3.com.
MP3 is here for a while. Using some recording and mixing smarts, and using a mastering studio that understands the limitations of an MP3 file, will go a long way to ensuring that the sonic intent of the music is not lost.
Noah Mintz is a Mastering Engineer at Lacquer Channel Mastering and the creator of enhancedMP3.com technology.
Gifts For
Audio Professionals at Music Books Plus
