| Recent Comments |
| Categories |
| Archives |
| Tags |
Volume level calibration for mixing: the K-N system
For recording and mixing it is important to set a consistent reference volume level to which you can always return. Below I explain how I did it for my studio.
More information about calibrating volume levels:
See also my post about (true) peaks, RMS, and Loudness.
Integrated approach to monitoring (Bob Katz) (pdf) 162.95 KB 119 downloads
An Integrated Approach to Metering, Monitoring, and Levelling Practices by Bob Katz...
Volume level calibration in brief
Equal loudness contours
In the 1930’s two acoustic researchers, named Harvey Fletcher and Wilden Munson, studied the phenomenon of human perception of frequencies vs. loudness. They developed a loudness contour, called the Fletcher-Munson Curve, which indicates the volume (dB SPL) levels across the frequency spectrum necessary for the listener to perceive a constant loudness level when presented with pure steady tones. This calculation was refined in 1956 and is now used as the ISO 226 standard for Equal Loudness Level Contours.
The K-system
The K-system is an audio level measuring technique proposed by mastering engineer Bob Katz in the paper “An integrated approach to Metering, Monitoring and Levelling” (see document above). The proposed K-System is a metering and monitoring standard that integrates the best concepts of the past with current psychoacoustic knowledge. This is also described in his book and in this book. However, the K-system is no longer present in the latest third edition of his book as he considers it obsolete. Although following the same principles, the new approach to calibration is based on Loudness and PLR (Peak to Loudness; cress factor) are better metrics for monitor calibration.
We should avoid meters which have 0 dB at the top–this discourages operators from understanding where the message really is. Instead, we move to a metering system where 0 dB is a reference loudness, which also determines the monitor gain. There are three different K-System meter scales, with 0 dB at either 20, 14, or 12 dB below full scale, for typical headroom and SNR requirements. Users can calibrate their system’s electrical levels with pink noise, without requiring an external meter. RMS also makes a reasonably-effective program meter that many users will prefer to a VU meter. Note that the integration time of a VU meter is 300 ms, which makes it too fast to act as a loudness meter.
The three K-System meter scales are named K-20, K-14, and K-12. The K-20 meter is intended for wide dynamic range material, e.g., large theatre mixes, audiophile music, classical (symphonic) music, and so on. The K-14 meter is for the vast majority of moderately-compressed high-fidelity productions intended for home listening (e.g. some home theatre, pop, folk, and rock music). And the K-12 meter is for productions to be dedicated for broadcast.
Note that full scale digital is always at the top of each K-System meter. The 83 dB SPL point slides relative to the maximum peak level. Using the term K-(N) defines simultaneously the meter’s 0 dB point and the monitoring gain.
This 83dB SPL level comes from the Fletcher-Munson Curve that shows that at 83 dB the loudness curve is reasonably flat.
Note: Bob Katz proposed to calibrate of 83 dB SPL (C-weighted) for a single/mono channel. This gives about 86 dB for a stereo channel (using uncorrelated pink noise). This is his interpretation of ‘forte’ in the mastering-room.
The use of 83 dB SPL calibration is for several reasons:
- Monitoring music at a slightly louder level encourages the engineer to produce a wider dynamic range. Engineers working at a greatly reduced SPL tend to over compress their masters. Nevertheless, it is still important to check the song at various levels.
- 83 dB is consistent with the SMPTE film standard.
- 83 dB is the most linear point on the ear’s loudness contours. It enables the flattest reproduction and translates to the widest variety of user levels and venues.
There are also some potential issues:
- Only large rooms with high-headroom monitors located far enough (e.g., 2.1 meters) from the listener can support this loud forte. Otherwise, the maximum level has to be reduced (e.g., 79 dB).
- The smaller the room volume, the greater the perceived loudness for a given SPL. This is due to psycho-acoustic reasons.
The K-System in Cubase
In Cubase 14 Pro it is not possible to select the predefined K-(N) scales for the channels in the MixConsole. However, the Control Room allows the selection of, among others, K-20 and K-14. For the MixConsole I redefined its color scheme of the +3dB scheme to reflect both the K-20 and K-14 system:
Image. The -20dbFS in the MixConsole corresponds perfectly (as expected) with 0db on the K-20 meter of the Control Room.
K-20 calibration and Cubase metering
Calibration should be done according to the standard originally proposed by Dolby’s Ioan Allen, with pink noise at -20dBFS. Why pink noise? See [here]. If I send in -20 dB of pink noise from the Cubase TestGenerator plugin then the Control Room meter peaks at about 0 dB (K-20 scale) with an RMS of about -9.9 corresponding to the Waves VU readout (which I have set for 20db headroom). The Waves Dourrough meter also peaks at -20 dB (red leds) and the RMS is about 10db below this peak level (yellow leds) (see Wikipedia for relation peak and RMS). Thus, all these meters are in agreement. The RMS of pink noise is about 10 dB lower then the peak level.
Note: a sinus test signal of -20dbFS shows slightly above the 0 dB on the VU (not sure why this is), -19dbFS peak on the Dourrough meter (why not -20dB?), and slightly above 0dB (0.3dB) on the K-20 meter (with RMS=-2.7). Thus the difference between peak level and RMS depends on the input material (here pink noise vs sinus wave).
As recommended, I filter the pink noise from 500Hz to 2kHz (12db/oct) with FabFilter Pro Q-4 to account for low-frequency standing waves or strong local mid/HF reflections in the room. Since this reduced the volume, I subsequently increased the level of the pink noise generator to -14.50db to get exactly 0dB peak level of the Master meter (control room; K20 scale ). Note: alternatively you may download a pink noise (-2odB RMS; 500-200Hz) audio file from Digido.com. However, note that the frequency spectrum is different from the result obtained with FabFilter. Increasing the slope in FabFilter gets a more similar output, but the digido file still contains more low and high-end.
The standard is now 83db SPL from a single speaker as measured on a full-bandwidth SPL meter with C-weighting and slow averaging from a -20dbFS RMS pink noise source..
This is too loud for a small room. My room is <60m3 and, therefore, the recommended calibration level is 76dB SPL(C) (see table in SOS article). This can be considered as a starting point.
I measured with the SLM-500 SPL meter (which should be pointed towards the speaker from the listening position). If I measure from listening position that the SPL fluctuates heavily been 64 up to 75 SPL(C). I think this is caused by reflections in my room. Therefore, I just settled for a level that I found comfortable. Measuring few cm from the speakers this gives about 80db per speaker.
Audient Nero
I use the Audient Nero monitor controller to work at calibrated volume levels during mixing. Although possible solely with the X32, I now can leave the X32 faders close to 0db (where there are more sensitive). The 12 o’clock position of the Big Knob is calibrated at a comfortable volume level (80 db SPL(C) per speaker measured at about one centimeter from the cone). This corresponds to K-20 in my DAW. In addition I have marked the K-14 position on the device.
Volume level calibration for mixing: Loudness
The K-system is no longer present in the latest third edition of the mastering book of Bob Katz as he now considers this approach obsolete. A better approach for monitor calibration is based on loudness and the PLR (Peak to Loudness; cress factor) are better metrics for monitor calibration. However, from his book it seems that calibration is based on loudness only, and that an appropriate PLR automatically follows from this. PLR can be measured with Cubase Supervision or with Dynameter, which currently is the better option.
I proceeded as follows:
- Windows output level: set to max (100%)
- X32
- Input gain -5 dB
- Send level to output bus: 0 dB
- Note: these settings are chosen such that the 3 o’clock position on my monitor controller corresponds with -23 LUFS / -14 dBFS.
- Cubase
- SoundID reference plugin: safe headroom was -6.3 dB
- Gain: 0 dB
- Faders: 0 dB
- All levels in the Cubase control room set to 0 dB
- Setup audio track with Cubase TestGenerator to produce full-band pink noise at -23 LUFS (this corresponds with a level in the TestGenerator plugin of -13 dB)
- This produces a monitor level of about 77 dB SPL(C) with the monitor controller set to position 8 (3 o’clock)
- I have taken a reference track with a level of -14 LUFS
- This produces a monitor level of about 77 dB SPL(C) with the monitor controller set to position 5.5 (3 o’clock)
- I have taken a reference track with a level of -7 LUFS
- This produces a monitor level of about 77 dB SPL(C) with the monitor controller set to position 4.5 (3 o’clock)
Thus for a range of loudness, I can now listen to a song a the same volume (SPL) by setting my monitor controller. However, as explained in the book of Bob Katz, for mastering one should work in reverse. If you want to make a loud master (say, -7 LUFS) then the monitor controller should first be set to position 4.5. Then mastering for the level (77 dB SPL) that you are used to, your master should result in about -7 LUFS with a corresponding PLR.
Last updated on January 3rd, 2025 at 07:16 pm
