Stereo Delay Splitter — User Guide

Independent channel delay processing: applies different divisor-based delay patterns to left and right channels separately, creating stereo width and spatial separation through differential temporal processing.

Author: Shai Cohen Affiliation: Department of Music, Bar-Ilan University, Israel

What this does

This script processes stereo sound by splitting channels, applying different delay arrays to each (like delay_array.praat but per-channel), then recombining. Left channel gets divisors 1-2, right channel gets divisors 3-4. Different delay patterns create stereo width and spatial interest. Result: Enhanced stereo image with independent channel transformations.

Stereo Input: [LEFT] [RIGHT] ↓ ↓ Split: L R ↓ ↓ Process: d1,d2 d3,d4 ↓ ↓ Recombine: [LEFT] [RIGHT] └───┬───┘ Stereo Output

Quick start

Select one stereo Sound object in Praat (must be 2-channel).
Run script… → Load Stereo Delay Splitter script.
Choose Preset: Default/Narrow/Wide/Alt or Custom.
If Custom: Set divisors 1-4 and channel ranges.
Click OK — processing instant, auto-plays stereo result.

Quick tip: Start with Default preset (L: 2,4 | R: 8,10) for balanced stereo width. Wide Stereo for maximum separation, Narrow for subtle effect. Requires stereo input — mono files won't work.

Presets

Preset	Left (d1,d2)	Right (d3,d4)	Character
Default	2, 4	8, 10	Balanced stereo width
Narrow Stereo	3, 5	6, 8	Subtle separation, similar channels
Wide Stereo	2, 6	12, 18	Maximum separation, extreme width
Alt Divisors	2, 3	9, 15	Asymmetric, experimental spacing

🎧 Default (L:2,4 | R:8,10)

Left: Longer delays (samples/2, samples/4) = slower echoes

Right: Shorter delays (samples/8, samples/10) = faster echoes

Effect: Left spacious, right tighter = balanced stereo image

Best for: General use, enhancing stereo width without extremes

How It Works

Independent Channel Processing

Process flow:

Split stereo: Extract left (channel 1) and right (channel 2)
Process left: Apply divisors 1-2 (iterations left_start to left_end)
Process right: Apply divisors 3-4 (iterations right_start to right_end)
Recombine: Merge processed channels back to stereo
Normalize: Scale peak to 0.99

Delay Formula (Per Channel)

Same as delay_array.praat, applied independently:

For each iteration k, calculate: b = samples / divisor[k]
Apply: output = self[position+b] - self[position]
Emphasizes transients, creates rhythmic patterns

Stereo width mechanism: Different delays per channel = different temporal patterns = perceived spatial separation. Larger divisor difference = wider stereo image.

Parameters

Parameter	Type	Default	Description
Preset	choice	Default	Preset stereo configuration
Left Channel Divisors
divisor_1	positive	2	First left channel divisor
divisor_2	positive	4	Second left channel divisor
Right Channel Divisors
divisor_3	positive	8	First right channel divisor
divisor_4	positive	10	Second right channel divisor
Iteration Ranges
left_start / left_end	natural	1 / 2	Which divisors for left (1-2 = d1,d2)
right_start / right_end	natural	3 / 4	Which divisors for right (3-4 = d3,d4)
Output Options
scale_peak	positive	0.99	Peak normalization level
play_after_processing	boolean	yes	Auto-play result

Creating Stereo Width

Divisor strategies:

Narrow stereo (subtle separation):

Similar divisors: L: 3,5 | R: 6,8
Small difference = subtle width
Cohesive stereo image

Balanced stereo (moderate width):

Moderate difference: L: 2,4 | R: 8,10 (default)
Clear separation without extremes
Most versatile

Wide stereo (maximum separation):

Large difference: L: 2,6 | R: 12,18
Dramatic spatial effect
Can sound disjointed if too extreme

Rule of thumb: Greater divisor difference between channels = wider stereo image. Example: L:2 vs R:18 (difference=16) = very wide. L:3 vs R:6 (difference=3) = narrow.

Applications

Stereo enhancement:

Add width to narrow recordings
Create spatial interest in mono-like material
Differentiate left/right channels creatively

Sound design:

Asymmetric stereo effects
Spatial movement through differential delays
Experimental stereo processing

Mixing:

Widen specific stereo elements
Create stereo versions of similar takes
Add depth to flat stereo recordings

Best with: Stereo recordings with some similarity between channels (not completely independent content). Percussion, ambiences, synth pads work well. Voice may sound unnatural with extreme widening.

Output

New Sound object: "soundObj_stereo"

Channels: 2 (stereo)
Left: Processed with divisors 1-2
Right: Processed with divisors 3-4
Normalized: Scaled to 0.99 peak
Original: Preserved (script works on copy)

Processing: Instant execution. Temporary left/right objects created and removed automatically. Only final stereo result remains.

Limitations

Stereo input required: Will not work with mono files (needs 2 channels)
Fixed iteration count: Always 2 iterations per channel (hardcoded)
No mid-side processing: Processes L/R directly, not mid-side components
Divisor-based delays only: Cannot specify delays in milliseconds directly
Same formula both channels: Only divisors differ, not processing type

Related script: For mono processing with more iteration control, use delay_array.praat