Anaglyph: Binaural Spatialization
In Binaural audio and recording , you heard how recordings can preserve the cues that help us locate sounds over headphones. Anaglyph lets you add those cues to a mono recording and choose where the sound appears to be.
Anaglyph positions a single mono source around your head. It uses filters based on measurements of a head and ears to make the sound appear to come from a chosen direction, including above, below, or behind you. How convincingly you hear that position depends partly on how well the filters match your hearing.
You will use Anaglyph for the individual sources in Project 2 . Keep the project’s stereo microphone recording and Zoom H3-VR binaural recording on their own tracks. Those recordings already contain spatial information; downmixing them through Anaglyph would turn each into a single source.
Start by installing the plugin and positioning one sound. Then automate a move. The control reference below explains the settings you can explore afterward.
Listen on headphones. Binaural processing sends different cues to the left and right ears. Over speakers, each ear hears both channels, so the intended spatial effect may not hold.
Installing
Download the release for your platform from the
Anaglyph site
and unzip it. On Windows, copy
Anaglyph.vst3 into C:\Program Files\Common Files\VST3\. On macOS, copy
Anaglyph.vst3 to ~/Library/Audio/Plug-Ins/VST3/ or Anaglyph.component to
~/Library/Audio/Plug-Ins/Components/. Rescan under Options, Preferences,
Plug-ins, VST.
The plugin ships filters computed at 44.1 kHz and 48 kHz. Run your project at one of those rates or download the extra sample-rate package, or you will get file-loading errors.
Setting it up in REAPER
- Put a mono recording on its own track. Use a separate track and Anaglyph instance for each source you want to position independently.
- Set the track pan to center and leave it there. Anaglyph does the positioning. Use its controls to move the sound.
- Add Anaglyph to the track’s FX chain. Leave the track at two channels so the plugin can output separate left- and right-ear signals.
- Set INPUT CHANNEL MAPPING to Stereo Downmix so a stereo file is summed into one source. This instance of Anaglyph positions that source at a single point.
- Put on headphones and drag the ball in the display before you automate anything.
In Top view, the display on the left shows the scene from above, with your head at the center. Top and Rear switch the viewpoint and ZOOM changes the scale of the view.
First listening exercise
Loop a short mono recording, such as a voice or a sound effect. Keep WET MIX at 100% so you hear the processed signal.
- In Top view, drag the source to one side of the head, then the other. Watch AZIMUTH, the angle around your head, change as you move it. Use the display to check which side corresponds to positive and negative values.
- Move the source in front of you and then behind you. Does it sound as though it stays behind you, or does it seem to move back to the front?
- Change DISTANCE, measured in meters, while keeping the direction fixed. Listen for changes in level and in the sense of how close the sound is.
- Change ELEVATION, the angle above or below ear level. Return it to 0° before trying another movement.
If front and back are hard to distinguish, try several options under CURRENTLY ACTIVE HRIR. Each set contains a different listener’s measured filters. Keep the one where sounds behind you stay behind you. The personalization section below explains these filters in more detail.
Automating a move
Project 2 asks for at least two automation moves. Open the track’s automation envelope window and show the Anaglyph parameter you want to automate. An envelope changes a parameter over time: draw points on it to set where the sound should be at different moments.
Fly-by. Automate AZIMUTH from one side to the other and DISTANCE to dip to the shortest distance you want at the moment of closest approach. The change in distance helps the sound seem to approach and pass you. Leave the near-field enhancements active for this exercise; their controls are explained below.
Circle. Sweep AZIMUTH through a full turn at constant DISTANCE. Listen for whether the source stays behind you as it passes through the rear half of the circle. If it seems to jump to the front, try another HRIR set or enable MICRO OSCILLATIONS, which adds a small movement to help stabilize the perceived position.
Automate ELEVATION on its own before you combine it with anything else. Elevation can be difficult to judge and varies between listeners.
Using Anaglyph in your mix
The DISTANCE control changes level and near-field filtering. To place a source in a particular room or environment, you may also need reverb that matches the space. Put that reverb after Anaglyph, or on a separate bus, so you are not spatializing a reverb tail to a single point.
For Project 2, use Anaglyph to position individual sources alongside your stereo and binaural recordings. Check the combined mix over headphones and listen for whether the sources seem to belong in the same space.
Control reference
You do not need to adjust every control to position a sound. Start with AZIMUTH, ELEVATION, and DISTANCE, then use the sections below to address what you hear. The INFO button in the lower right corner of the plugin opens its built-in control descriptions.
Source position
| Control | What it does |
|---|---|
| ELEVATION (deg) | Height of the source. 0° is level with your ears; positive values are above and negative values are below. |
| AZIMUTH (deg) | Angle around you. 0° is directly in front. Move the source in Top view to check the angle for each side. |
| DISTANCE (m) | How far away the source is, in meters. |
| MIN / MAX ATTEN. DIST. | The near and far bounds of the distance attenuation model. Inside the minimum and beyond the maximum, gain stops changing. |
Automate AZIMUTH, ELEVATION, and DISTANCE to move the source. MIN / MAX ATTEN. DIST. sets the range over which distance changes affect its level.
Distance attenuation
Gain falls off as 1 / distance ^ ATTEN. EXP.
| Control | What it does |
|---|---|
| ATTEN. EXP. | The exponent in that formula. 1.0 is the physical inverse-distance law. Higher values make distance more dramatic. |
| BYPASS DIST ATTENUATION | Turns the level-versus-distance model off, leaving only the filtering cues. |
Personalization
Binaural rendering filters with a head-related impulse response, an HRIR, measured on a head. An HRIR is the time-domain representation of a head-related transfer function, or HRTF. Both describe how sound changes on its way to the ears from a particular direction.
Your ears may differ from those used for the measurements, which is one reason the same filters work better for some listeners than others.
| Control | What it does |
|---|---|
| CURRENTLY ACTIVE HRIR | The HRIR set in use. The bundled choices include sets from the LISTEN database, such as listen_irc_1008. Try several and keep the one where sounds behind you stay behind you. |
| CUSTOM ITD | Replaces the measured interaural time difference, the difference in arrival time between the ears, with one modeled from your own head size. |
| HEAD CIRCUM. | Your head circumference in millimeters, used by that model. Measure with a tape above your ears. |
| BYPASS HRIR | Turns the ear filtering off. Useful for A/B: everything collapses back toward ordinary panning. |
| IR XFADE DUR. | Crossfade time between impulse responses when the source moves, in seconds. Low values like 0.05 stay responsive; high values like 0.4 remove zipper noise on fast moves. |
| NO INTERP. | Turns off interpolation between measured positions. Leave it off unless you are testing. |
If you automate a fast fly-by and hear stepping or zipper noise, IR XFADE DUR. is the control to reach for.
Near-field enhancements
These improve the sense of distance for sources close to the head.
| Control | What it does |
|---|---|
| BYPASS PARALLAX | Parallax picks the HRIR by the direction the sound actually arrives from at each ear, rather than from the center of the head. It matters most within arm’s reach. |
| BYPASS ILD | The near-field interaural level difference model adjusts left and right levels by frequency to account for head shadowing, mostly by lowering the far ear rather than raising the near one. |
Leave both enhancements active for nearby sources, with their BYPASS controls off. Enable each bypass in turn to compare the sound with and without that enhancement.
Externalization booster (STUDIO)
Over headphones, sounds often sit inside your head. This module adds the subtle cues of hearing a monitor system in a treated studio, which pushes the image outside. It is not a reverb plugin and not a substitute for your own sends.
| Control | What it does |
|---|---|
| CURRENTLY ACTIVE STUDIO | Which studio response is used: room_01_studio_foley, room_02_studio_dry, or room_03_studio_warm. |
| GAIN STUDIO | Level of the booster, independent of the direct binaural signal. |
| H / M / L | High-, mid-, and low-frequency EQ for the booster only. These controls do not change the direct sound. |
| SPATIAL DETAIL | The spatial detail of the studio response. OMNI has no directional detail; 2D adds horizontal directions; the 3D settings include height. Higher settings use more CPU. Leave this at its default for now; we will cover the underlying ambisonic format later. |
| BYPASS STUDIO | Turns the booster off. |
| STUDIO ONLY | Mutes the direct binaural sound and leaves only the booster. For testing. |
Stabilization
| Control | What it does |
|---|---|
| MICRO OSCILLATIONS | Varies the rendered position very slightly. With a static source and no head tracking, the brain has trouble telling front from back; the small movement adds redundant cues and reduces that confusion. |
If a sound that should be behind you keeps jumping to the front even after you choose an HRIR set, try enabling MICRO OSCILLATIONS.
Output
| Control | What it does |
|---|---|
| GAIN | Overall plugin gain. |
| WET MIX | Blend of spatialized and raw signal: out = wet × spatialized + (1 − wet) × raw. Keep it at 100% for a full effect, lower it to pull a sound partly back toward the original. |
| BYPASS ALL | Passes the input through untouched. |
| CLIP | Turns red when the output has clipped. Click it to reset. |
Alt-click or double-click any control to return it to its default.
Optional: loading other HRTFs
Start with the bundled HRIR sets. For a further comparison, you can load other
measurements stored in SOFA, a file format for spatial acoustic data. Files are
available from the SOFA conventions
collection
. They have
to be preprocessed first; the
tutorials page
links
the anaglyph-hrtf-preprocess Matlab package that does it, and the processed
files go into the plugin’s resources alongside the ones that shipped with it.
Compare two HRTF sets on the same recording and at the same source position. Listen for changes in tone and in how clearly you can locate the sound.
Documentation and further examples
The website has a feature overview but no parameter manual. The reference is built into the plugin: the INFO button in the lower right corner opens a panel describing every module and every control in alphabetical order. Click it when you are unsure what a knob does.
Beyond that:
- anaglyph.dalembert.upmc.fr has the download, the About section, and release notes.
- The tutorials page covers importing your own SOFA HRTF files, the Max control messages, and the Unity package.
- The misc page lists DAW compatibility and known issues. Pro Tools is not supported on any platform; REAPER is fine.
- The gallery collects projects made with it, which is where to go for making-of material.
- The AES e-brief The Anaglyph Binaural Audio Engine by Poirier-Quinot and Katz is the academic write-up.