Quad-to-Stereo Decoding

The Concept

Decoding an Ambisonic mix to four virtual speakers in the horizontal plane — front left, front right, rear left, rear right — and summing those to stereo is a robust alternative to full binaural convolution. The approach produces a wider, more stable image than a direct HOA-to-stereo downmix, while avoiding the phase artifacts that binaural HRTF decoding introduces in the low-mids.

HRTFs work with two main principles:

• Gains, Phase shifts and inter-aural time differences — the microscopic delays between left and right ear that create the illusion of externalization.

• Spectral shaping — a frequency-dependent coloration caused by the pinna reflecting and diffracting sound before it reaches the ear canal. For sounds arriving from behind, this produces a characteristic high-frequency rolloff and a set of notches in the 6–10 kHz range (Blauert, 1997).

The auditory system uses both cues, but the spectral shaping is particularly robust — it works even in mono and is the primary cue for elevation and front/back disambiguation.

By applying a high-frequency rolloff to the rear channels before summing to stereo, the rear path provides a spectral cue consistent with "this sound is behind me," without the phase manipulation that causes metallic artifacts in full binaural decoding.

The quad-to-stereo approach typically sounds:

• Less phasey than full binaural — no inter-aural crosstalk cancellation

• Wider than a direct stereo downmix — the rear channels push energy to the sides of the stereo field

• More robust across different headphones and even loudspeakers — the spectral cue survives playback systems that would destroy phase-based cues

The Setup in Reaper

Virtual speaker placement in IEM AllRADecoder:

• Front pair: ±30°–45° azimuth, 0° elevation

• Rear pair: ±135° azimuth, 0° elevation

Keep all speakers in the horizontal plane. Elevation in the virtual array introduces height decoding artifacts when summing to stereo.

Processing on the rear bus:

The rear bus is where the character of the spatial impression is shaped. Several types of processing are useful here, and they can be combined:

Spectral shaping (the core tool)

A high-frequency shelf or low-pass is the primary means of creating the front/rear distinction. A gentle shelf starting around 5 kHz preserves some rear-channel air; a steeper low-pass at 3–4 kHz produces a stronger sense of immersion at the cost of high-frequency diffuseness. This mimics the spectral signature of rear-hemisphere HRTFs (Blauert, 1997).

Level trim

Reducing the rear bus by 3–6 dB relative to the front keeps the center image stable and prevents the rear energy from masking the direct sources.

Decorrelation

A small amount of decorrelation between RL and RR (via a short allpass or the SPARTA Decorrelator) widens the perceived rear field and prevents the two rear channels from summing to a narrow mono image. Use sparingly — too much decorrelation sounds unnatural and creates mono compatibility issues.

Diffusion / early reflections

A short, dense reverb tail on the rear bus can increase the sense of envelopment without adding direct-source energy. Keep pre-delay at zero so the rear bus does not arrive before the front.

For spectral shaping, any single-band shelf or low-pass works well:

Front/Rear Balance

The level ratio between the front and rear buses determines the perceived depth of the mix.

• Too much rear level: the mix feels diffuse and lacks a stable center image.

• Too little rear level: the spatial width collapses and the result sounds like a conventional stereo downmix.

A starting point is to set the rear bus 3–6 dB below the front bus and adjust by ear. In a dense electroacoustic mix the rears primarily carry reverb tails, spatial diffuseness, and low-mid weight — not direct sources — so they can often sit lower than expected while still contributing a clear sense of space.

Limitations

This technique is a perceptual approximation, not a geometrically accurate spatial reproduction. Its main limitations:

• No externalization — without phase-based HRTF cues, the sound remains inside the head to some degree. The spectral rear cue creates the impression of depth and width rather than true externalization.

• Front/back collapse on loudspeakers — the spectral cue is weaker on speakers than on headphones, so the front/rear distinction may be less clear outside a headphone context.

• Fixed virtual geometry — the four speaker positions are static with no head-tracking compensation.

For an accurate, externalised binaural image, full HRTF convolution via SPARTA AmbiBIN or IEM BinauralDecoder remains the reference. The quad-to-stereo approach is most appropriate where robustness across playback systems matters more than geometric precision.

References