Sound localization cues

Sound localization combines several auditory cues.

• time difference: different arrival times at the ears
• level difference: different levels at the ears
• spectral cues: filtering by the pinnae, head, and torso

This section focuses on time and level differences because they are central to stereo reproduction.

Related: Fundamentals of Sound · Sound Localization · Room Acoustics

Stereo loudspeaker reproduction

Summing localization

Correlated signals can fuse into a phantom image. Equal signals produce the center; level or time differences shift it toward the louder or earlier loudspeaker.

Equal-power stereo panning

Mono, dual mono, and stereo

• mono: one channel carrying one signal
• dual mono: the same signal in both channels; over two loudspeakers it forms a phantom center without interchannel differences
• stereo: two channels using interchannel differences to create a spatial image
• stereo width: the horizontal extent of the image between the loudspeakers

→ Dual mono can produce a phantom center; stereo width requires interchannel differences.

Spatial audio beyond stereo

Spatial audio may be represented in several ways:

• channel-based: signals assigned to loudspeaker channels, such as 5.1 or 22.2
• object-based: sources with position metadata, rendered during playback
• scene-based: an encoded sound field, as in Ambisonics

Reproduction may use loudspeaker arrays or binaural headphone rendering. Diffusion projects sound through loudspeaker arrangements, including unconventional or site-specific setups, either fixed or performed live.

Delay principles and technologies

Perceptual result

• delay: one or more distinct repetitions
• reverb: many dense reflections perceived as an environment

Implementation

• tape: loops and playback heads; speed and spacing determine delay time
• electronic: bucket-brigade circuits or digital memory
• software: flexible timing, feedback, routing, synchronization, and modelling

Insert and send/return routing

Effects can operate in series or in parallel.

Echo

An echo is a distinct repetition heard after the original sound.

Whether it separates from the source depends on the delay time and the character of the sound.

Listening examples: the echo threshold

The same voice is presented dry and with a single equal-level repetition. Only the delay time changes.

1. Dry: no delay. ▶ Play
2. 15 ms: fused; coloration rather than a separate repeat. ▶ Play
3. 55 ms: near the boundary; the repeat begins to separate. ▶ Play
4. 400 ms: distinct echo. ▶ Play

Delay parameters

• Time: interval before the repetition
• Feedback: amount returned to the delay input
• Sync: relates delay time to tempo and rhythmic divisions
• Filtering: changes the spectrum of repeated sounds
• Mix / level: balance between direct sound and delay
• Stereo placement: distributes repetitions between channels

Delay signal flow

Delay patterns and stereo placement

Single delay
One short repeat (slapback)

▶ 95 ms

Feedback delay
Several decaying repeats

▶ Low feedback

Ping-pong delay
Repeats alternate left and right

▶ 300 / 600 ms

Listening example: dub production

Dub uses sends, delay, feedback, and reverberation as compositional tools.

Compare the original vocal recording with its dub version:

▶ original · Bree Daniels, Oh Me Oh My

▶ dub version · The Upsetters, Oh Me Oh Dub

Components of reverberation

Reverberation is the persistence of sound formed by reflected energy.

• Direct sound: dry signal arriving first
• Early reflections: first reflections shaping space
• Reverb tail: dense, decaying reflections that define ambience

Artificial reverberation

Room, hall, and chamber describe spatial characters that can be produced through different approaches.

• acoustic and electromechanical: chamber, plate, and spring
• algorithmic: networks of delays and filters
• convolution: applies an impulse response
• acoustic simulation: geometric or wave-based modelling

Reverb parameters

• predelay / early reflections: onset and initial spatial cues
• decay: duration of the reverberant tail
• size / diffusion: scale and reflection density
• damping / filters: spectral change through the tail
• width: stereo spread
• mix: dry/wet balance (normally 100% wet on an effect return)

Impulse response

An impulse response describes how a linear, time-invariant system responds to a very short excitation.

• the direct sound arrives first
• early reflections follow
• the response decays over time
• the recorded waveform characterizes the measured source–system–receiver path in time

Examples: a room, a loudspeaker, a microphone, a filter, a plate or a spring

Measuring and applying an impulse response

A known test signal is used to measure the system:

1. play the test signal and record the response
2. use deconvolution to derive the impulse response
3. apply the response to dry audio through convolution

→ The impulse response acts as a time-domain fingerprint of the measured path.

Convolution

Convolution calculates the output of a linear time-invariant system from an input signal and its impulse response.

Each input sample produces a scaled and shifted copy of (the copies overlap and add).

Impulse response libraries

• OpenAIR, University of York
  research and auralization; documented real spaces, usually measured with logarithmic sine sweeps; individual datasets available

• MIT IR Survey
  acoustic research; 271 everyday spaces measured at a standardized 1.5 m source–receiver distance

• Waves IR Library
  professional production; more than 2,000 captured responses, 4.8 GB; account required

Listening examples: convolution reverb

Three measured responses, ordered by decay time. Each row presents the response, then speech and a single click processed with it.

Dry speech ▶

Using reverb in montage

• add depth and space to dry recordings
• blend layers into one coherent environment
• lower the direct-to-reverberant ratio to suggest distance
• use short reverbs for definition and long reverbs for atmosphere

Creative reverb techniques

• reverse reverb: creates a rising tail before the source event
• gated reverb: cuts off the tail abruptly for a shaped, percussive decay
• filtered reverb: changes the spectral character of the reverberation
• automation: changes send level, wet/dry balance, or decay to create transitions and shifts in perspective

Modulation controls

An LFO (low-frequency oscillator) changes a parameter cyclically.

• rate: speed of the modulation
• depth: amount of change
• shape: contour of the modulation cycle

Modulation effects may also provide:

• feedback: returns part of the processed signal to the input, strengthening resonances
• mix: balances direct and processed signals

Tremolo and vibrato

Both effects use periodic modulation but change different signal properties.

• tremolo: modulates amplitude, producing a periodic change in level
• vibrato: modulates pitch, producing periodic variation around the original pitch

Chorus, flanging, and phasing

• chorus: combines the source with one or more short, modulated delays; small timing and pitch differences create richness and stereo width
• flanging: combines the source with a very short, modulated delay; interference produces a moving comb-filter pattern
• phasing: mixes the source with modulated all-pass filtering; interference produces moving spectral notches

Feedback can strengthen the resonances in flanging and phasing.

Stereo width and mono compatibility

• level, timing, spectrum, and decorrelation can create width
• delay and modulation may create phase differences between channels
• these differences can cancel when the signal is summed to mono

→ Check spatial effects in stereo and mono.

Mid/side representation

Mid/side processing converts left and right channels into sum and difference signals. Changing the Side level adjusts stereo width.

• Mid:
• Side:

The original channels are reconstructed as:

• ▶ Mid is the channel sum and forms the mono-compatible component
• ▶ Side is the channel difference and carries left–right information
• ▶ Stereo combines Mid and Side to reconstruct the original channels

Glitch, lo-fi, and databending

• tape, vinyl, VHS, and digital compression have characteristic artifacts
• degradation can be preserved, simulated, or exaggerated
• artifacts may suggest age, memory, distance, or technological identity
• bitcrushing, sample-rate reduction, buffer repetition, freezing, and databending treat digital systems as material

Spectral and data-driven transformation

• spectral drawing and resynthesis: create or transform sound through a time-frequency image
• cross-synthesis and morphing: combine characteristics of two sound sources
• corpus-based recomposition: analyze collections and reorganize fragments by sonic similarity
• neural resynthesis and timbre transfer: regenerate input through a model trained on another body of sounds

Examples: MetaSynth, IRCAM ASAP and RAVE, Zynaptiq MORPH, FluCoMa, and SpectraLayers.

Delay time as rhythm

Use a short percussive source to make delay part of the rhythm. Explore delay times related to the project tempo and shape feedback so that the repeats form a clear pattern without clipping.

Choose one version and explain how the delay reorganizes the source.

Convolution

Transform one dry recording with two contrasting impulse responses: one room and one short non-room sound. Compare the results at similar levels and choose the response that best supports an intended spatial or timbral character.

Explain how the chosen response changed the source.

Selected references

• Blauert, Jens. Spatial Hearing: The Psychophysics of Human Sound Localization. Revised ed., MIT Press, 1997.
• Dutilleux, Pierre, M. Holters, S. Disch, and Udo Zölzer. “Modulators and Demodulators.” DAFX: Digital Audio Effects, edited by Udo Zölzer, 2nd ed., Wiley, 2011, pp. 83–99.
• Farina, Angelo. “Simultaneous Measurement of Impulse Response and Distortion with a Swept-Sine Technique.” 108th AES Convention, Audio Engineering Society, 2000.

Selected references (continued)

• Smith, Julius O. Physical Audio Signal Processing: For Virtual Musical Instruments and Audio Effects. W3K Publishing, 2010. Online edition.
• Välimäki, Vesa, et al. “Fifty Years of Artificial Reverberation.” IEEE Transactions on Audio, Speech, and Language Processing, vol. 20, no. 5, 2012, pp. 1421–48.

Related: Fundamentals of Sound · Filters · Sound Localization · Room Acoustics