Audio Log I

Modern music is mixed and mastered too loudly.

The old process of making music:

  1. Instruments are recorded separately, with a mic placed a small distance away. The distance and any echo effect captured is called natural reverb. Jazz bands and some older rock bands recorded together, live. This was done on the older Black Sabbath records.
  2. Instruments are mixed on a 24 bit format, with frequencies ranging from 96kHz to as much as 192kHz. Of course the human hearing range is 20-20kHz, so a lot of what was recorded was just noise. 24 bits was like the word length, basically the more bits you had the more information you can mix in. The same way your PCs work today. CDs and mp3s are 16 bit.
  3. Since the vinyl was a physical, analog medium, an engineer would be hired to ensure that the songs were literally of the right size to fit on to the vinyl pressing. Making the songs louder would take more information on an analog medium, so they would take up more space on the vinyl. To save money, music was engineered to be soft.

The vinyl has a lot of flaws. Any scratch on its surface would come out in the music. Having a frequency range of 192kHz was pointless since most of the frequencies recorded were ambient noise. Frequencies below 40hz were hard to capture on the format (hence why most older songs barely have bass).

The CD was introduced around the late 70s. The CD is at 16 bit and 44.1kHz, covering the main hearing range from as low as 20Hz all the way up to 20kHz. For some reason, the CD remains at 44.1kHz up to today, which means a choppy resampling of music from the older frequency ranges. Only the DVD is at 48kHz, which makes it easy to just half or quarter the older frequencies (96 or 192kHz). Also, since all CD players and playback devices were engineered at 16 bit then, nobody bothered to really make 24 bit players or CDs. It wasn’t just the medium that changed- recording changed too.

The new process of making music:

  1. Music is now recorded digitally. There is no mic unless you record acoustic and vocals. The processing, however, is all digital. Logically, there is barely any natural reverb either.
  2. A lot of effects and compression now available since music can be digitally altered. Individual instrumental tracks and vocals can now be processed on their own before being mixed together with the rest of the instruments. No coincidence that musicians started auto-tuning their mistakes digitally around this period.
  3.  The lower frequency range meant that bass was more audible. The new frequency and recording allowed for a cleaner, more polished sound without random noises or ambiance.
  4. The concept of digital space is very different from analog space. While there is a limit of 700mb, loudness was no longer an issue. The time limit was the bigger issue, so as long the album was less than 45 minutes, the engineer could do a lot more than he wanted.

Compression at the mastering stage isn’t necessarily bad. It adds warmth and richness to the music if applied properly.

The irony now is that the medium that allowed for a softer, more detailed sound is being abused to make louder, harsher sounding records. When music aired on the radio, they had to have the same assumed loudness, otherwise the listener would have to constantly turn down the volume for louder records and turn up the volume for softer records. The solution was crude; simply compress all music to sound relatively the same (at least in terms of volume).

Let’s look at the spectograms of two separate songs:

01. Fear Of A Blank Planet
Fear Of A Blank Planet by Porcupine Tree

The spectogram has multiple peaks. The range between the highest peak and the lowest peak is dynamic range. While not the best indicator of how good the music will sound, I have observed that the most pleasant sounding records, even in the metal genre, often have very dynamic masters. On top of that, the orange space is referred to as headspace. The more headspace a record has, the softer it will sound, giving you room to add in EQ choices that will not make it sound unbearable. Goes without saying that dynamic masters work well with complex sound system setups. For reference, this track was ripped from the 5.1 surround DVD-Audio and downmixed to stereo with the Channel Mixer plugin on foobar. The dynamic range for this track is DR12.

Here is the spectogram for Lamb Of God’s Desolation:

02. Desolation
Desolation by Lamb Of God

It’s truly a desolate sight. There is no headroom, no peaks and no variation in the loudness, meaning everything is squashed to your ears. This isn’t to say that there is a lot of detail- a lot of the mix is noise frequencies at the midrange and a bloated mid-bass. The dynamic range for this track is DR5, which is not even that squashed by today’s standards.

Of course I am using two different songs from two very different genres to exaggerate a point, but go on to the dynamic range database and look up music you like. Older Metallica records went up to DR12. Slayer’s Reign In Blood, which I would consider the gold standard for thrash metal production, is DR12 on CD. The older Cannibal Corpse albums clock in at DR10-DR12 and still sound heavy and visceral.

A mistake I used to make was thinking that vinyls had better sounding music because they registered a higher dynamic range. Not true. Since all vinyls contain noise, not just from the vinyl but from the needle playing it, the noise will be mistaken by the dynamic range meter as a super small peak. As such, the program will detect a high dynamic range, even if the same master for the CD was used for the vinyl.

The only way to know if the vinyl had an uncompressed master is to actually ask the vinyl pressing company itself. The guys at Metal-Fi have a forum thread on dedicated vinyl masters for metal records.

I am now embarking on ‘restoring’ badly mastered tracks, having seen some amazingly repaired songs floating about on the Russian internet. It’s been difficult, since wave peaks that are lost in the master are honestly lost forever, but audio technology has improved vastly today. I use iZotope RX 5, with the Neutron plugin. Primarily a mix and master software, it features a dynamic equalizer that has given some astonishing results, with the software using complex logarithms to ‘guess’ the lost frequencies. The results still aren’t perfect, but I have already achieved a lot.

Here’s the spectogram of a remastered Desolation:

02. Desolation declipped
The results of my messing about with the equalizer.

Reducing gain and increasing headspace is easy, its guessing the right frequency to increase or decrease that’s been proving to be difficult. From the picture, you can already see that the new peaks are way too high, but I have achieved the jagged peaks that are prominent on more dynamic records. I now know that most metal records have too much mid-bass (the irony that reducing bass increases the audibility of the bassline!), too little treble and an extremely pointless noise at around 800hz.

This is the first of my “audio-logs”. I don’t intend to be an actual producer, but I hope I can find a way to make my music sound like it did when it was recorded.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s