We´ve noticed over the last few years that we´re sent an increasing number of tracks that have compression and/or limiting on the main/2 bus from home and professional mix studios. As a generalisation we ask people to try to leave compressors and limiters off the main/2 bus when they submit their audio for mastering. Where it´s unavoidable (ie where it´s used for the tonality rather than gain) we usually ask for alternative mixes with and without the compressor and make an informed judgment with the client on which to master.
Far too often though a compressor or limiter is dumped on the main bus just to get volume at mixing as a ´loud´ mix will sound better to a client. One difficulty when you come to master these tracks is that too often the audio will be clipped and may well be distorted. Once this has happened it´s very difficult to correct. This is then made worse as any additional processing that you attempt at mastering may well drive the track in to more distortion and clipping.
A couple of week ago we received an album to master. All the tracks had been over compressed or limited on the main bus at mixing and the files had been dithered to 16 bit. We asked the client to contact their mix engineer and ask for new files at 24 bit and with out any processing on the 2 bus, etc. Time went on and the client contacted us on Monday to say that the mix engineer was refusing to answer their calls and could we do anything at all with the files: they couldn´t afford to have the album remixed etc and were running out of time.
Here´s an image file of one of the original tracks - note that a lot of the audio peaks hav flat tops, i.e. they are clipped. Also, though it´s not as noticeable here the song had been limited by the mixing engineer, who then applied volume automation to the track so that the choruses and verses were at different volumes (but still clipped). This gave an initial impression of the songs having louder and quieter passages but with very noticeable pumping/breathing.
Click to view attachment
Our intern/trainee told the client that there probably was nothing that we could do as, ´once compressed it is impossible to remove´. And she´s right but there are sometimes things a mastering engineer can do even in a case like this...
Downward Compression and True Upward Expansion
Compression is generally thought of as a means to reduce the dynamic range of audio once it reaches a given point (the threshold) by a set amount (the ratio). This is standard, or downward, compression and is all about making loud parts in a track quieter. True upward expansion is a compression technique that is rarely used. However, if done with care it can be used effectively to restore some dynamic range to audio that has been compressed or limited. In downward compression ratios are always greater than 1:1 - ie they may be 2:1, 3:1, 5:1 etc. Here the ratio means that for a 2:1 ratio every increase in the input level of 2 dB will lead to an increase in the output of only 1 dB. (The output is at a lower level than the input - we´ve reduced the gain). Here´s a standard downward compression graph, Input level on the bottom-horizontal axis and output on the vertical axis.
Click to view attachment
In true upward expansion the ratio however is fractional, i.e 1:1.1, 1:1.5, 1:2, 1:3, so for a ratio of 1:1.3 it means that we increase the output signal relative to the input by 1.3 dB per 1dB at a given threshold. So what are we doing in true upward expansion? We´re increasing the loudness during the attack part of the audio. This brings out the transients and the higher frequencies and make the audio sound more dynamic and lively. Essentially we´re trying to remove, or minimise, the compression that has already been applied.
What we did
The track/s in this case were all over compressed and lacked dynamic range. They sounded dull and lifeless and so we wanted to restore the dynamic range of the material. Also the tracks sounded a little dull, partly due to the compression/limiting and partly due to the 16 bit format. So we want to add a little high frequency air. However we found that for these tracks that true upward expansion here placed too much emphasis on the fast transients at the expense of some loss of stereo width and warmth. We also wanted to gently limit the track after expansion to avoid it going in to clipping on fast high transients. Finally we wanted to do all of this and to adjust the gain stage so that we had some room to work and add any other processing that we needed. (We eventually also added EQ and another compressor before re-dithering and taking back to 16/44.1.)
To do all of this we first changed the bit depth from 16 to 24 and then upsampled from 44.1 to 96kHz. At this stage all of the processing was post fader and we also adjusted out gain stage to reduce the level of the track by 3dB coming out of the processing and in to the ADC. We added a mastering compressor to the chain and applied true upward expansion, adjusting the attack time so that we could emphasize the transients but without them poking through too much - we wanted more more dynamics and percussive attack on the snare etc but not to the point where it dominated the vocals. Added some very gentle limiting to again reduce the percussiveness a touch. At this point we had restored much of the lost dynamics but at the expense of the low end and some overall coherence. So we added a downward compressor set on a very low ratio (1.8:1) and at a low threshold to bring up the low end and help bring the track together.
Here´s a graph of downward compression followed by true upward expansion and finally some limiting (which is what we did):
Click to view attachment
We also adjusted the mid/side to bring up the audio in the side channel slightly relative to the mid to give an impression of an increased stereo width. once we were happy with this we sent the track in to an EQ before taking it back into the DAW and final bounce down. Here´s an image of the final track - what you should notice is that it has more peaks and fewer flat tops and that there is also more dynamic movement, particularly in the quiet sections.
Click to view attachment
[Don´t] try this at home - unless you can...
Did it work - yes the final tracks and album have much more dynamics, both for individual components and also between verse and chorus and, there is much less pumping than the original mixes and the client is happy. it was also a very good learning experience for our trainee: she now understands why I keep saying that everything you do is specific and that there are very few rules in mastering
.Should you try this at home? Probably not as to do this takes a bit of experience using compressors and also requires a compressor that is capable of fractional ratios. There are not many that do and the few that do are expensive.
We should start an "Advanced" section in the Wiki just for stuff like this.
I guess it's the subtitle.