Audio Production

EQ

Equalization (EQ) can be used to modify the frequency spectrum of a sound. It is the bread and butter of the mixing and mastering process. EQ can be used subtly or creatively and in simple terms, it consists on augmenting or reducing the sound of certain frequencies. Many EQs allow to affect multiple bands, for example we may want to increase the volume around 200 Hz and reduce it around 1 kHz, or we may want to cut all sound after 10kHz.

Imagine each EQ band as a filter and generally will have available the parameters of frequency, gain and Q. There are many types of filters, and while internally may be built differently, the basic types are:

In general we want to balance our frequency spectrum to our liking and sometimes this involves making space for instruments. For example, if we have a guitar and a piano playing at the same time, both will have a good amount of mid frequencies, and when they are playing together they may sound muddy. By removing some highs from the guitar and some low-mids from the piano, we can make them sit more harmoniously in the mix. Something I do quite often is to fully remove (high pass) the frequencies above 200-300 Hz for many instruments to make space for the bass and the kick drums. If we have many elements in the mix, I may even low-pass the bass instruments, which will change their character, but at the same time, if we crowd a given frequency range, that sound wouldn’t be easy to distinguish.

Another use of EQ is to do sound design, shaping samples or instruments to create more appropriate sounds for a given track. In this chart you can find a reference of a few instruments and some characteristic frequency ranges. For example, we may want to increase the attack of a snare drum (by boosting around 900 Hz) or reduce the boominess of a kick drum (reducing the gain around 60 Hz). This chart is to be used only as a guideline reference to where certain characteristics are frequently found, factors such as tuning, the notes being played, the style of music we are going for, the instruments or recording conditions, etc. will change these reference points. Use your ears to determine where something is missing or where there is too much of!

Compression

Compressors can be used to reduce the dynamic range of a signal. Commonly used to control peaks but they have many applications. When a signal goes past the threshold it will be attenuated. Usually compressors will have the following parameters:

Glue compression

We use glue compression when we want to integrate different mix elements together in a bus (or master). There are different ideas on the parameters for setting up glue compression, for example, some go for short attacks and long releases, some for fast attack and long releases. It may depend on the type of effect that we want to achieve, sometimes we can get pumping effects or artifacts when attempting to do so. In any case the goal is to aim for 1-2 dB of reduction (and makeup gain to compensate for it). If you want to time the release time perfectly, we can use the following formula, which will give us the time in milliseconds:

release = 60000 / bpm * 4 - attack

Shorter attack times will affect the transients of the signal more, which may not be desired, wherease longer attack times will preserve them.

Resources

Pads

We don’t want to overcompress, but if we want to control the dynamics a little bit we can set up a compressor with a relatively long attack and release and around a 3:1 ratio. It can be useful to avoid high discrepancies in volume, specially when automating volume fades.

Drums

Parallel compression (NY compression)

The idea is to overcompress a track (low threshold, medium-fast attack and release and large ratio) and blend the overcompressed signal with the original one. If your compressor has a wet/dry mix, you can achieve this effect quite easily. Works specially well for drum loops, providing a more aggressive sound.

Mixdown tips

Pink noise mixing

Pink noise have the same amount of energy across all octaves. If we mix to a pink noise reference, it will make it so our music can sound louder, since limiters will not slam the lower frequencies first.

Setup your EQ/spectrum analyzer balanced to pink noise and set a reference volume to balance around. Pure pink noise mixing can sound a bit too bright so setting a slope of 4.5dB instead of 3dB (pink noise) can mixing to the horizontal line can help deal with this issue. On the master bus, you can use an EQ set up just before the limiter for this purpose, instead of track by track. You can also use something like Ozone to dynamically adjust this stuff for you. In Bitwig you can use a multiband fx-3 (MBFX3) to split you frequency ranges and use limiters and compressors in each band independently, this is useful for drums. With MBFX3 and a limiter in each band you can raise the limiter until it is just starting to hit the ceiling.

The reference is mostly arbitrary but give enough headroom. For example select -45dB, balance the kick to peak at that line, the snare a bit lower than that, the melody a bit above, etc. These are stylistic choices.

You can do this by ear by setting a pink noise generator or wave file and then, with all faders set to -inf, slowly raise each individual track until you can hear it above the noise and then decrease the volume until is barely audible. Some people mention that is better to do this in mono for more accurate results. You can do this channel by channel by muting the rest or just adding all elements as you go. Either way, this is not a silver bullet, just a tool that can be used to get a quick mixing balance, but ultimately you need to use your ears to set your levels.

Resources

Scopes and mastering

Before a final release, you likely want to master your music so that it sounds as best as it can on as many listening environments and distribution platforms as possible.

The first thing to consider is the frequency balance, which is technically part of the mixing process, but the two are intertwined. If you want your music to sound loud, you want a balanced frequency spectrum as previously described. Typically I would have a final EQ or multiband compressor on my mastering chain to correct for overall signal imbalances. In Bitwig/Ableton I use a spectrum analyzer for this purpose, with a 4.5dB tilt in slow and/or freeze mode. This lets me know how the spectra evolves around time and which moments emphasize certain frequencies.

Having a balanced spectrum makes it so that the second step, loudness, is more easily achievable. A compressor and/or limiter are used to raise the level of the input signal. If you have an imbalanced signal, the limiter will be hit the ceiling earlier in some frequencies, which will make it so either we have distortion or we can’t push the overall signal as loud as we would like. I may use a combination of a compressor plus a limiter in some cases to lightly reduce some peaks before going to the limiter device. The loudness wars are mostly over, but it is worth trying to make your music as loud as possible according to the distributor’s specifications. For streaming services, I will tend to master to Spotify’s specs, which are shared by other services (-14LUFS with a max true peak of -1dB). I use the freely available Youlean loudness meter or dpMeter5 to monitor these metrics across the entire song and adjust my limiter as needed to achieve these specs.

I also like to monitor the overall stereo spectrum using the free MStereoScope plugin. Sometimes it’s worth considering to widen or pan certain tracks and individual channels. I also monitor the final mix in mono to ensure is sounds OK.

It’s a good idea once we have a final version to print the different stems or parts of a song if desired (vocal vs instrumental). Often we want to revisit or remix a song and can’t load the original project due to unavailable plugins, samples or something else entirely. Having the stems available makes this much easier.

As a final note, don’t forget to set up the metadata in all the tracks, it will save a lot of time when sending them to distribution.

Transitions

Here are some ideas that can be used to create smooth transitions between sections of our songs or when we want to introduce a new instrument to the mix. Often can be combined so try a few and see what feels right!

Ear candy

To keep the listeners engaged throughout the song, we can sprinkle noises, instrument hits, effects, glitches, textures and many other details. These need to be subtle and secondary to the main arrangement, but they can really help bringing a song together. Sometimes we can repurpose existing material (such as vocals) processed for different sounds. Drum or bass fills could also be considered part of ear candy, specially when not used in transitions.

Voice

Compression

Voice love compression, but specially if we are doubling vocals, we want to compress those even more to reduce the dynamic range and make mixing easier. Double vocals in general tolerate much better overdoing it on effects, like reverb and delays.

Reverb/Delay + Compressor

Reverb and delay and vocals are really important for bringing a sense of space and otherworldliness but this can often overwhelm the sound and remove a lot of clarity from the lead vocals. A very useful trick is to put the reverb and delay as send buses and add a compressor to each of them, which will be used to sidechain the FX with the lead vocal signal. This allows the reverb/delay to be heard when the voice is not playing, but they will duck when it is giving us a much more clear signal. We can set the compressors with a fast attack, slow-medium release, 6-7:1 ratio and adjust the threshold until we achieve around 5-8dB of gain reduction. Make sure to remove makeup gain for this to work.

Other

Genres

Sound structure

In all music genres there is some amount of structure to a song. Even the lack of structure (for example an ambient track or a generative patch) is itself structured as a single part song. There are classical and common musical forms (sonata, rondo, AABB, AABA, ABA, 12-bar blues, etc.) but in popular music different sections are given other names and arranged according to musical genre. Note that even within each genre, there is a wide amount of variation, ultimately structure is a creative decision, but it is useful to know common conventions.

In addition to the following section types, it is also worth considering the overall energy of the track. Even with common progressions (verse-chorus-verse) the emphasis of energy can be set at different points (last chorus is the most energetic, similar energy all around except in some spots, etc.). Ways of changing a track’s energy is via adding/removing instruments, parts or textures, increasing melodic/harmonic rhythm, variations in drum patterns, melodic lines reaching higher pitches.

Synthwave

Characterized for a modern interpretation of 80s sounds, with heavy emphasis in simplicity and synth sounds. Some staples of the genre are:

Famous artists

Resources

Glitchhop

Resources

Elektron

Digitakt

Octatrack

Live resampling

We can use a record trig in combination with a flex track to live resample audio coming in from one of the inputs or some internal playback buffer. By default, a record trig + playback trig on the first step allow us to pitch down the incoming audio live, but pitching up or reversing is not possible this way. If we micro-time the record trig forward one or two clicks we can’t no longer pitch down, but we should be able to pitch up the incoming audio. Note that in this case we may need to retrigger the sample with an additional play trig. For example, if we pitch up an octave, in addition to the play trig on the 1st step, we should also use a play trig on the 9th. This isn’t really “live”, though, instead we are playing the last recorded loop in the buffer. In case of “live” reversing, the same process as with pitch up applies, only the “sample start” position should be adjusted. For the first trigger the start should be set at 1 and on trigger 9 it should be 64.

Using a “source 3” recording instead of AB/CD, we can pitch up a sample without micro-timing. For example, we can set a live recorded on track 5 from AB and use track 6 to record from track’s 5 via source 3. Another advantage of doing so is that audio can be routed to more than one track, for example track 7 could also be sampling from track 5 and process the effects differently.

Other

Production Resources

Mid-side processing

Headphone calibration

I mix pretty much entirely on headphones, given that on my home studio I don’t have access to good monitors or a properly conditioned room. This has some caveats, and even though I know my headphones pretty well, the mixes don’t always translate to other systems. Recently I discovered a simple method to calibrate headphones for a more truthful response, which works system wide and can be implemented in pretty much any DAW and operating system. For example, I mostly use Bitwig on Linux and works like a charm there, no need for external plugins or other software like Sonarworks!

The idea is to try to match your headphone’s frequency response to the “Harman curve”, which seem to have become a standard for audio mixing on headphones. This curve tends to match the preference of a multitude of people from different demographics. Instead of a totally flat response, it tries to mimic the response of flat speakers in a treated room and gives a closer approximation to what a studio environment sound like.

Ideally you will be starting with a good pair of headphones, but you don’t need to break the bank for the most expensive pair. There is massive diminishing returns here, and some 100 buck cans are probably more than enough for most people. I personally use a set of Beyerdynamic DT 990 Pro (250 ohm) and while I like the sound and comfort, I had some issues with the build construction so I wouldn’t necessarily recommend them. Just find something that works for you within your price range and try to keep them as long as possible.

To obtain the calibration curve you can go to the AutoEQ website and type down your headphones model on the search bar. You will be greeted with a scary graph, for which we only really care about the Raw, Target and Equalized plots. On the left you can click “show advanced” and select different target profiles or make any desired adjustments, but by default the appropriate Harman curve should be selected. In the “Select equalizer app” dropdown you can select a variety of options if you want to add this curve to several equalizers (iTunes, Spotify, Rockbox, etc), but for my purposes I use the “Convolution Eq”. You should be able to see the Equalized graph very nicely matching the target curve. Click the “download” button (arrow pointing down) and wave the wav file to your desired folder.

In Bitwig (or your DAW of choice), add a convolution loader in your master and load the equalizer file as an impulse response. Make sure the mix is set to 100%, wet gain is set to 0dB and Width is set to 100%. I like to add a tool device afterwards to gain match the before/after signal. You can group these two together and save the chain as a preset. But wait, we are not done yet, you could disable this chain if you are mastering or measuring your track, but instead, I like to create an FX bus and move the chain there. Disable your master output and set the input of the calibration channel to Master (POST). Don’t forget to turn on monitoring! Lastly, set the output of the calibration track to your headphones output in your audio interface. You should now have headphone calibration directly on your DAW.

Any IR loader will work for this process, and if your DAW doesn’t have one available, there are plenty of free VSTs for this purpose.

You could potentially apply these equalization settings globally, but if you work with multiple headphones this may become a hassle. Furthermore, in that case, you can just add as many FX tracks with their respective calibration IRs and route the audio as needed.

I don’t claim this is the only or best method for headphone mixing, but seems to work for me, and hopefully it may work for you too. In any case, you should always listen to your mixes in a variety of different listening environments: different headphones and earbuds, different speaker systems, your car, mono systems, etc.

Mixing/Mastering

Loudness

Nowadays, music is most likely going to be consumed from streaming and digital platforms. These services have started to normalize the percieved loudness so that no longer music that sounds louder appears as sounding “better”. This means we no longer have to suffer from the Loudness War and we have some more headroom in dynamic range when mixing and mastering a song.

The loudness is typically measured in LUFS/LKFS (Loudness, K-weighted, relative to full scale). Different streaming services will reduce the volume of the music until is in their selected ranges. For example, at the time of writing, the following list show an example of the target LUFS in each platform:

To measure LUFS in a Digital Audio Workstation software (DAW), there are fantastic free plugins available, such as Youlean Loudness Meter.

Some people recommend mixing with a target of -23 LUFS with 6dB of headroom and then mastering for the target platform. Of course, we can always mix a bit louder (-12 to -10 LUFS depending on how much compression we are OK with) than we need to target all platforms. Beware of not overdoing it, otherwise we risk losing dynamic range.

Resources

Miscellaneous

How to create a reverse reverb effect on a hit

From Mr. Bill’s video on Ableton Reverse effects (Source)

  1. Isolate and bounce the audio track we want to apply the effect. Short hits work better.
  2. Reverse the bounced audio track.
  3. Apply a reverb and/or delay to the rev. bounced track.
  4. Freeze, flatten and consolidate the audio track with the effect.
  5. Reverse the track back.
  6. Put it back in time with the original hit.

Note that if the track was not reversed prior to consolidation, we obtain a fade-in effect instead when reversing the processed track.

How to set the gain of your microphone

Try to set the gain so that the average audio peaks at around -12dB to have plenty of headroom to play with.

Noise gate

On OBS, set the close threshold and open threshold 5 apart (Open should be smaller), e.g. ct 55, ot 50.

Compression

The effect of compression can still be heard even at lower volumes. The attack of a compressor can help us control the transients. Longer attacks left the transient through, and the faster the attack will shorten the transient. Long attacks and high compression can make for punchier attacks and faster attacks can smooth out harshness. Playing with the release can be used to play with the swing/groove. For example blending overcompressed drums with a slow release will highlight the higher frequencies. Playing around with the release can help shape high-hats. Additionally, the release can be used to create texture. Slower releases will reduce some “growl” smoothing the sound and faster releases can make the transients more obvious.

Resources

Software

Guitar amps/cabinets