home - Articles

links

Painting on the stereo field

A tutorial on stereo image, phases, mid-side.

(works in progress - needs audio files and effects screenshots)

So you've recorded all your tracks. Everything sounds nice, the groove is there. Ready for the final mix.

You have a powerful, digital audio workstation and lots of plugins. One million times more powerful than a 70's, expensive, famous recording studio. You know how to control dynamics, equalize frequencies, automate levels, etc.

But the song still sounds flat. All the instruments are sitting on a small arc in front of you.

Maybe it is time to make the stereo field your friend.

Stereophonics

Ok, we have two ears, this is why we like stereo, we know.

Like with our eyes (3D cinema?), we need two sources (at least) to fake real world in an immersive experience.

Understanding how we perceive it, can help us a lot while trying to give the picture of the real space.

Stereo is a beautiful thing and, while more sophisticated (and complicated) arrangements exist (surround), it has a great advantages: it is simple, works well in the worst conditions (my car) and is portable (headphones!).

Most of all, it is how we are used to listen to the music, and this, alone, is enough to make it sound good.

Two sources, together

Stereo is two-sources. They play together and contribute to build the "sensation" of a real soundscape.

An important point to focus on, here, is that, differently from a single speaker, the combination of two sources can create what is called a "virtual image". Under certain circumstances, our brain fuses the sources (the speakers) to something we know in the real world.

These circumstances are related to time, frequencies and levels.

Two sources perfectly the same and reaching our ears at the same time are heard as a single, focused, precise, near, centered one. Very rarely you can hear such a sound in a commercial music release.

The same two, at different levels, give the same sensation but the virtual source is placed near the louder speaker. This is basic stereo panning.

Audio file - a source moves around (panning) perfectly in phase

Time, frequency and level relationships

When the sources aren't perfectly related is where the interesting part begins.

Our brain instinctively elaborates lots of information from microscopic details in the sound.

Each sound is classified, compared to other sounds, classified again, and visualized as something we know. A virtual source, an arrangement of virtual sources, an environment, our position.

Everything is transformed to an image? Almost everything. Everything else is resolved to nausea (did you ever try to exchange the poles behind a loudspeaker?) or the speakers as real sources (my flat mix!).

Audio file - the same source out of phase

There are three kinds of relationships we deal with: time, level and frequency (pitch).

Being in time (or louder!)

Now I'll start from the simplest hypothesis: the two sources (from the two speakers) are the same but not perfectly in time.

There have been interesting studies on this situation, from which the known "Haas" or "precedence" effect has been defined.

If we hear one sound, and another copy slightly delayed, we hear just one. But, here is the interesting part, it comes from the first position.

What is "slightly"? Between some milliseconds and tenths of milliseconds.

Why is it so interesting? Because the second copy makes the sound a bit bigger. And because it can be controlled to build the so called localization cues.

If we hear an echo, we try to use it to understand the ambience.

A single echo tells something. Some echoes tell a lot. The timbre of the echoes tells almost everything.

The fast echoes in the Haas range are the so called early reflections. This is why reverb is so important, not only to give front-rear or distance but also for positioning the instruments on the stage. But you don't need full reverb to make a picture.

Consider this example: only early reflections (fast, low level, echoes) are added to the sources. The coherence between the echoes makes our brain hears the instruments in an environment.

Audio file - a band plays (dry) then the reflections come in

This is what is happening in the example:

Screenshot from a delay positioning effect

Each wall reflects the sound. The coherence between the sources and the reflections give a picture.

Being similar

But what about delay being really microscopic? Smaller than the time needed for a single harmonic to go up and down? Now it is when we start calling it "phase".

The phase between the two sources can change dramatically the sound. In time (loosing punch), and in timbre (changing the spectrum). But it changes totally the stereo image.

A "virtual source" is centered and focused when the real sources are in phase.

When the phase deviates, we start loosing focus. It is not bad, it is a tool (or, maybe, a problem if you are recording from real).

Audio file - an instrument starts in phase and goes 180° out of phase, then back

There are lots of tools for phase control.

We can use them trying to match the phase for the two sources. If we have recorded something with more than one microphone this is often the case. But we can aim to avoid the perfect phase, too.

One important and not so intuitive concept is: phase is related to frequency. A sound contains (generally) several frequencies. If we delay a sound, this some amount is big for "small" waveforms (high frequencies) while it is small for "big" frequencies.

This is why we cannot recover from bad phase relationships simply delaying a source!

Consider this sample.

The signal is first delayed, then restored and, finally, phase shifted:

Audio file - a stereo (mono) guitar changes delay then phase. It's different!

How big is an instrument?

Reading articles about Jacko recordings, I noticed that many mono instruments (all?) were recorded in stereo. Why? Because even with intrinsically mono instruments, "stereo" is the size.

But stereo is differences between the two sources (speakers).

So, how to give an instrument its size?

A mono source is like a point in the stereo image. A stereo source, with exactly the same signal for the two sources is mono again.

The size is in the difference.

There are several ways to make a mono source become stereo (often called "pseudo-stereo"). Some sound good, others sound bad.

This is a typical approach: double and delay the same mono track. It sounds phasey:

Audio file - two copies of a guitar track, delayed

Here we are unrelating the two sources by delay. This changes the timbre and our ear recognizes the so called comb-filtering.

Another approach is to create several, statistically unrelated, copies of the signal, each one panned differently. Statistical considerations limit the number of useful copies and so the effectiveness.

Another approach is spreading the frequencies in the stereo panorama. Under certain circumstances this sounds great.

This is a stereo effect without phasey side-effects:

Audio file - widened guitar, mono to stereo

Now that the instrument is stereo, it takes a slice of the stereo arc and it can be placed in its own angle.

A good stereo picture involves relating these "arcs" for instruments (and surrounding environments), because, as we hear every day, in mixing everything is relative.

"Wow! I'm going to make everything so huge in my mix!". No, you can't.

Again, as we hear every day, in mixing, big is only where small is. And stereo is only where mono is. Who likes big mono?

Compare the difference here:

Audio file - two wide instruments, guitar and drums

Audio file - wide drums, mono guitar

All this can be thought of as matter of frequency bands, not only the whole signal.

So, we can try to get wide high frequencies while leaving mono bass frequencies:

Audio file - guitar, mono bass frequencies, stereo high frequencies

Screenshot of an effect modifying the relative phase in a frequency band

A very diffused but not so well known technique is mid-side encoding/decoding. Nothing more than adding (mid) and subtracting (side) the two channels (left and right) and thinking in terms of these new signals.

This simple table (values are generic "units", with relative meaning only, sign is signal polarity) should help to build an useful reference in our minds:

Left	Right	Mid	Side	Image
+0.5	+0.5	+1	0	centered, mono
+0.2	+0.8	+1	-0.6	panned right, mono
+1	0	+1	+1	hard panned left, mono
+0.2	-0.8	-0.6	+1	panned right, stereo
+0.5	-0.5	0	+1	isn't mono compatible!

Row n.3 shows that the name "mid" for that component isn't so easy to understand. There is mid even when the signal is hard panned to a side! Mid-side is only matter of phase relationships, not of levels!

It is often said that stereo widening is obtained by amplifying the side components and/or attenuating the mid component. But remember: there isn't side component in a mono signal. A mono source must be transformed to stereo (pseudo-stereo) before mid-side enhancement is available.

Screenshot of a spectrogram showing a stereo signal

Which side is the instrument? How far is it placed?

Each instrument has its own size on the stage (stereo), and we can place it in the environment (cues) and add the third dimension, depth (reflections).

Depth is often associated to reverb. Reverb is just a more complex result of reflections. While early reflections (or longer, discrete, delays!) let us understand "the environment", reverb adds even more colour to the picture.

Stereo image of the reverb (or delay) greatly contributes to depth, and, most af all, again, contrast between reverbs from two instruments. The main tools are: true stereo reverb (two channels for each audio channel), stereo reverb (one channel for each stereo channel), stereo reverb with input summed to mono, mono reverb. The panning of reverb, relative to the virtual source, is very effective, too.

Audio file - drums, bass, guitar. Several solutions for the reverb on the guitar. How the depth changes

Where am I?

There may be more than placing the instruments on the stage. Some "special effects" are needed, sometimes.

The position of the listener maybe another variable. Just think of the classic effect "I'm out of the disco. Now I enter".

Here is the effect, using an equalizer and mono reverb:

Audio file - Are you on the list? Listening from outside