Phonetic Tremolo/Glitch Effect — User Guide

What this does

This script implements an intelligent phonetic effect processor that automatically analyzes speech sounds and applies different audio effects based on phonetic classification. Using acoustic feature extraction (pitch, intensity, formants, harmonicity), the system identifies vowels, fricatives, silence, and other sounds, then applies targeted processing: tremolo modulation to vowels, time-shift glitches to fricatives, heavy attenuation to silence, and mild amplification to other sounds.

Key Features:

• Automatic Phonetic Classification — Real-time detection of speech sound types
• Targeted Effect Application — Different effects for different sound categories
• Acoustic Feature Analysis — Uses pitch, formants, harmonicity, intensity
• Frame-based Processing — Precise temporal control over effect application
• Debug Output — Shows classification statistics and processing results

Technical Implementation: (1) Feature extraction: Creates Pitch, Intensity, Formant, and Harmonicity objects from the input sound. (2) Frame-based analysis: Processes audio in small time frames (default 10ms). (3) Classification: Uses acoustic thresholds to categorize each frame as vowel, fricative, silence, or other. (4) Targeted processing: Applies tremolo to vowels, glitch effects to fricatives, attenuation to silence, and mild boost to other sounds. (5) Debug output: Provides classification statistics for tuning and analysis.

Quick start

1. In Praat, select exactly one Sound object (preferably speech).
2. Run script… → phonetic_tremolo_glitch_effect.praat.
3. Adjust Effect Parameters:
   • tremolo_rate_hz: Speed of vowel modulation (2-15 Hz)
   • tremolo_depth: Intensity of tremolo (0.3-0.9)
   • shift_amount_seconds: Glitch time shift (0.005-0.03 s)
4. Set Feature Extraction parameters for analysis quality.
5. Adjust Classification Thresholds to tune sound detection.
6. Click OK — effect applied, result named "originalname_glitched".
7. Check Info window for classification statistics and debug information.

Phonetic Analysis Theory

Acoustic Feature Extraction

Fundamental Acoustic Properties

The system analyzes four key acoustic features:

Praat Analysis Objects

Feature extraction pipeline:

    Time: 0.0s     0.5s     1.0s     1.5s     2.0s
    Text:  "S"     "A"      "Y"      "SH"     "EE"
    
    Pitch:  ___    /¯¯¯\     /¯¯\     ___     /¯¯¯\
           0 Hz    120 Hz   130 Hz   0 Hz    125 Hz
    
    HNR:    ___    /¯¯¯\     /¯¯\     ___     /¯¯¯\
           2 dB    18 dB    16 dB    1 dB    20 dB
    
    F1:     ___    /¯¯¯\     \___/    ___     \___/
           0 Hz    600 Hz   300 Hz   0 Hz    280 Hz
    
    Class: Fric    Vowel    Vowel   Fric     Vowel
    Effect:Glitch  Tremolo  Tremolo Glitch   Tremolo
    

Frame-Based Processing

Temporal Analysis Windows

The audio is processed in small, overlapping frames:

Why Frame-Based Processing?

Complete Processing Pipeline

Sound Classification System

Classification Logic

Acoustic Thresholds Explained

Effect Parameters

Debugging and Tuning

Interpreting Classification Output

Tuning for Different Audio Material

Parameters

Effect Parameters

Feature Extraction

Classification Thresholds

Applications

Creative Vocal Processing

Use case: Adding rhythmic interest to vocal tracks

Technique: Use musical tremolo rates synchronized to tempo

Settings:

• tremolo_rate_hz = BPM/60 (one cycle per beat)
• Medium tremolo_depth (0.5-0.7) for clear modulation
• Subtle shift_amount_seconds (0.008-0.012) for gentle glitches
• Default classification thresholds for clean vocals

Result: Vocals with rhythmic pulsation on vowels and subtle texture on consonants

Experimental Sound Design

Use case: Creating glitchy, fragmented vocal textures

Technique: Use extreme settings for dramatic effects

Settings:

• Fast tremolo_rate_hz (10-15 Hz) for intense modulation
• Large shift_amount_seconds (0.02-0.03) for obvious glitches
• High tremolo_depth (0.8-0.9) for dramatic volume swings
• Adjust thresholds to target specific sound types

Result: Highly processed, glitch-art vocal effects

Speech Enhancement and Effects

Use case: Adding character to spoken word

Technique: Use subtle effects with careful threshold tuning

Settings:

• Slow tremolo_rate_hz (3-5 Hz) for gentle movement
• Low tremolo_depth (0.3-0.5) for subtle effect
• Small shift_amount_seconds (0.005-0.01) for texture
• Tune thresholds for specific speaker characteristics

Result: Enhanced speech with added character while maintaining intelligibility

Audio Restoration and Noise Control

Use case: Reducing background noise in speech recordings

Technique: Use aggressive silence detection and attenuation

Settings:

• Increase silence_intensity_threshold to detect more noise as silence
• Use default or lower tremolo settings
• Minimal glitch effects (small shift_amount_seconds)
• Adjust vowel_hnr_threshold higher to avoid detecting noise as vowels

Result: Cleaner speech with reduced background noise during pauses

Practical Workflow Examples

Advanced Techniques

Troubleshooting Common Issues

Technical Deep Dive

Acoustic Phonetics Basis

Phonetic Class Acoustic Signatures

Typical acoustic values for different sound classes:

Praat Analysis Methods

Effect Algorithm Details

Tremolo Implementation

The tremolo effect uses a sine-wave amplitude modulation:

Glitch Effect Implementation

The glitch effect uses time-domain shifting:

Performance Considerations

Computational Complexity

Memory Usage

The script creates multiple analysis objects: