Audio Recording

A combination of “golden ear”, highly accurate nearfield monitors and well-treated room acoustics are a perfect recipe for audio mixing ad mastering success. But in reality, I would estimate only 20% of home recording engineers and producers possess all these three combinations.

So what now? The resulting mix or master would not still sound OK and would not translate to other audio monitors during playback. The primary cause of the problem is the lack of any of those three vital combinations. This is where frequency analysis of the mix would be very helpful as validation tool.

Importance of doing Frequency Spectrum Analysis of your Mix

Doing this would validate whether what you actually heard from your mix is true or not. For example in using your ears; you noticed that the bass are strong. However this is only subjective description which can either be correct or false. By doing a frequency spectrum analysis of your mix and analyzing the response on the bass regions; you will have an objective data regarding how much bass content that you have in your mix. This is not only used for bass level analysis but for mid frequency to high frequency range also.

You can then use this data to compare with the frequency characteristics of commercially released tracks to check if your mix frequency characteristics are actually out of spec or not. You can however use your ears to compare the levels between a reference audio and your mix but it takes a golden ear, great monitoring environment and well-treated room acoustics to notice this difference objectively.

However, it is highly discourage to rely purely on frequency spectrum analysis to make proper mixing and mastering decisions. And you should use a combination of your ears and this tool to validate the frequency response of your mix. This should produce the best results.

How to do this

1.) First, do a render copy or mix down of your mix in stereo at 16-bits/44.1KHz WAV. Make sure there is no clipping on your mix and the maximum peak signal should not exceed -3dB to -1dB.

2.) Measure the average volume of the mix in RMS level. Most mix is still not loud since they are not yet processed by mastering limiters, etc. It is expected that the RMS level should somewhere -30dB to -20dB. Take note of these statistics. For example, below is a waveform statistics taken using Adobe Audition 1.5:

In the above screenshot, the average RMS level of the mix down is around -26.25dB (for left channel) and -26.34dB (for right channel). There are two channels since the mix down is in stereo. The average between the left and right channel is ((-26.25dB) + (-26.34))/2 = -26.295dB

3.) Then select a reference master (commercially produced and released) that you would like to compare head-to-head with your mix. Make sure it is in CD format (16-bit/44.1KHz stereo).

4.) Measure the average RMS level of the reference master. Of course this is loud since it is already mastered. It is expected that RMS level for this one would be around -17dB to -10dB.

5.) Compute the amount you need to reduce the volume of the reference tracks so that it will have the same volume as your mix. Use the formula below:

dB need to reduce (de-amplify) = Your Mix volume (average) – Reference master average RMS level

For example if the reference master level is -13dB and your mix down is -26.295dB (average between the left and right channel); then:

dB need to reduce (de-amplify) = (-26.295dB ) – (-13dB) = -13.295dB

6.) Then using an audio editor (Adobe Audition for example) reduce/de-amplify the overall reference volume by the amount determined in step 5. The reference master waveform would become smaller, see below:

If you try to play this reference audio and your mix, they should now have similar volumes.

7.) Perform a Frequency spectrum analysis on the reference audio first. There are different tools for this; the best tool is the one that can provide a complete chart. Set the spectrum analyser at 65536 for FFT size. Do a screenshot for the result.

8.) Perform frequency analysis for your mix using the same FFT settings as before. Do a screenshot of the results.

Interpretation and Recommendations

Below is the complete screenshot of the frequency analysis results (plot using Adobe Audition frequency analysis tool):

To interpret the results, start with the bass regions. It is shown that the mix down has significantly higher bass content compared with the reference master. The mix down bass levels can be found somewhere -60dB to -72dB while the reference master is lower around -72dB to -84dB. To make the bass levels comparable with the reference master, you would be reducing bass content in your mix.
The levels from 100Hz to 2000Hz are comparable except for a dip in 200Hz region. However the mix down lacks high end from 5000Hz above.
Moving forward, the mixing engineer would double check the mix settings for sound improvement considering the above inputs.

Finally this type of analysis would also be helpful to a mastering engineer that would need to re-confirm the frequency characteristics of the audio before applying EQ.