Bit Crusher (8-Bit Arcade) Time Domain

Digital degradation effects: bit-depth reduction and sample-rate decimation for retro game sounds, lofi aesthetics, and creative distortion.

Author: Shai Cohen Affiliation: Department of Music, Bar-Ilan University, Israel Version: 1.0 (2025) License: MIT License Repo: https://github.com/ShaiCohen-ops/Praat-plugin_AudioTools

Contents:

What this does Quick start Digital Audio Theory Processing Modes Built-in Presets Applications

What this does

🎮 8-Bit Arcade Sound Aesthetic

Historical context: Early video games (1970s-1980s) used primitive sound chips with severe limitations: limited bit depth (4-8 bits), low sample rates (4-22 kHz), simple waveforms. Despite limitations, created iconic sounds that define retro gaming.

This plugin: Emulates these limitations digitally, applying both bit-depth reduction (amplitude quantization) and sample-rate reduction (temporal decimation) to create authentic 8-bit arcade sounds from any audio source.

🔢 Two Processing Approaches

1. Time Domain (Fast - Classic Bit Crusher):

Direct amplitude quantization in time domain
Sample rate reduction via decimation
Fast processing, classic "bit crusher" sound
Creates stair-step waveforms and aliasing

2. Spectral (Slow - Experimental):

Frequency domain processing via FFT
Quantization of spectral components
Experimental metallic/ringing sounds
Frequency-band selective processing

This script implements digital degradation effects — intentionally reducing audio quality to create specific aesthetic results. Bit crushing works by reducing the resolution (bit depth) and/or sampling rate of digital audio, creating quantization noise, aliasing artifacts, and harmonic distortion characteristic of early digital systems and lofi aesthetics.

Key Features:

4 Character Presets — From subtle to extreme degradation
Two Processing Modes — Time domain (fast) and spectral (experimental)
Bit Depth Control — 1-16 bits amplitude resolution
Sample Rate Reduction — Decimation for aliasing artifacts
Spectral Quantization — Frequency-domain processing option
Real-time Preview — Play result immediately after processing
Clean Workflow — Optional intermediate object cleanup

What is bit crushing? Digital audio quality depends on two main parameters: (1) Bit depth: Amplitude resolution (16-bit CD = 65,536 levels, 8-bit = 256 levels). (2) Sample rate: Temporal resolution (44.1 kHz CD = 44,100 samples/second). Bit crushing intentionally reduces these parameters to create: Quantization noise: Granular distortion from limited amplitude steps. Aliasing: Frequency folding from inadequate sampling. Harmonic distortion: Added harmonics from non-linear quantization. Musical uses: Retro video game sounds, lofi hip-hop beats, industrial music, glitch art, creative distortion effect. Psychological effect: Digital degradation creates "gritty," "crunchy," "nostalgic" character that contrasts with modern pristine digital audio.

Technical Implementation: (1) Time domain mode: Apply round() function to quantize amplitude values, then optional down/up sampling for sample rate reduction. (2) Spectral mode: Convert to spectrum via FFT, apply frequency-band quantization, convert back to time domain. (3) Preset application: Pre-configured parameter sets for different character. (4) Post-processing: Peak normalization, optional playback, info display. (5) Cleanup: Optional removal of intermediate spectral objects.

Quick start

In Praat, select exactly one Sound object to process.
Run script… → Bit_Crusher.praat.
Choose Preset or "Default (8-bit)" for classic arcade sound.
Select Mode: Time Domain (fast) or Spectral (experimental).
For Time Domain mode:
- Set bit_depth (1-16, lower = more crushed)
- Set sample_rate_reduction (1 = none, higher = more aliasing)
For Spectral mode:
- Set fast_fourier (yes for speed, no for precision)
- Set lower/upper_frequency for processing band
- Set quantization_steps for spectral resolution
- Set outside_range_multiplier for non-processed frequencies
Adjust scale_peak for output normalization.
Choose play_after_processing for immediate playback.
Set keep_intermediate_objects (usually no for clean workspace).
Click OK — processing completes quickly for Time Domain mode.
Result named "[original]_bitcrushed" appears in Objects window.

Quick tip: Start with presets — "Default (8-bit)" for classic arcade, "Heavy (4-bit)" for aggressive distortion, "Extreme (2-bit)" for severe degradation. Use Time Domain mode for classic bit crusher sounds (fast processing). Try sample_rate_reduction = 2 or 4 for pronounced aliasing effects. For spectral experiments: use narrow frequency bands (500-1500 Hz) with few quantization_steps (2-4). Source material matters: drums and percussion respond well to bit crushing (adds grit). Clean signals (sine waves, speech) show artifacts clearly. Already distorted material may become too noisy. Listen for: stair-step distortion (bit depth), metallic ringing (aliasing), granular noise (severe reduction).

Important: DESTRUCTIVE PROCESSING — creates new sound, doesn't modify original. Bit depth below 4 bits creates extreme distortion (may be unpleasant). Sample rate reduction > 4x causes severe aliasing (non-musical frequencies). Spectral mode is slow — large files may take time. Fast Fourier option speeds spectral processing but reduces frequency resolution. Frequency range in spectral mode should be within Nyquist (sample_rate/2). Outside_range_multiplier < 1.0 attenuates non-processed frequencies. Peak normalization (scale_peak) prevents clipping but may increase noise floor. Keep intermediate objects only for debugging/learning. Result may be much louder/quieter than original due to quantization characteristics.

Digital Audio Theory

Digital Audio Fundamentals

🔢 Sampling & Quantization

Analog to digital conversion:

Sampling: Measure amplitude at regular intervals (sample rate)
Quantization: Round amplitude to nearest discrete level (bit depth)

Key parameters:

Sample rate (Hz): Temporal resolution. Must be > 2× highest frequency (Nyquist theorem)
Bit depth (bits): Amplitude resolution. Determines dynamic range: DR ≈ 6.02×bits + 1.76 dB
Nyquist frequency: Highest reproducible frequency = sample_rate/2

Standard formats:

CD audio: 16-bit, 44.1 kHz (65,536 levels, 20-20,000 Hz)
8-bit arcade: 8-bit, 4-22 kHz (256 levels, limited bandwidth)
Telephone: 8-bit, 8 kHz (narrow bandwidth for speech)

QUANTIZATION MATHEMATICS: For N-bit quantization: Number of amplitude levels: L = 2^N Level spacing: Δ = 2 / L (assuming amplitude range -1 to +1) Example (8-bit): L = 2^8 = 256 levels Δ = 2 / 256 = 0.0078125 Any amplitude rounded to nearest multiple of 0.0078125 Quantization error (noise): Error per sample: e = actual - quantized Error range: -Δ/2 ≤ e ≤ Δ/2 Assuming uniform distribution: noise power = Δ²/12 Signal-to-noise ratio: SNR = (signal power) / (Δ²/12) Dynamic range: DR (dB) = 20×log₁₀(2^N) ≈ 6.02×N + 1.76 Examples: 16-bit: 98 dB 8-bit: 50 dB 4-bit: 26 dB 1-bit: 8 dB SAMPLING THEORY: Nyquist-Shannon theorem: To reproduce frequency f, need sample rate > 2f Aliasing: Frequencies above Nyquist fold back as f_alias = |f - k×sample_rate| Where k = integer that places f_alias below Nyquist

Quantization Noise & Distortion

📊 Error Characteristics

Ideal quantization: Error uniformly distributed between -Δ/2 and +Δ/2

Real quantization: Error correlated with signal (distortion)

Types of quantization distortion:

Granular noise: Random-like error from rounding
Idle channel noise: Noise at low signal levels
Overload distortion: Clipping when signal exceeds maximum level
Harmonic distortion: Added harmonics from non-linear quantization

Perceptual effects:

Low bit depth (≤8-bit): Gritty, crunchy texture
Very low bit depth (≤4-bit): Extreme distortion, "digital fuzz"
Sample rate reduction: Metallic, ring-mod-like aliasing
Combined reduction: Classic 8-bit video game sound

Dithering (not implemented): Adding noise before quantization to decorrelate error, making it less perceptible as distortion. This plugin intentionally doesn't dither to preserve the "crunchy" character.

Historical Context: 8-Bit Sound Chips

🎮 Classic Arcade & Console Sound

Early systems (late 1970s):

Atari 2600 (1977): 2-channel mono, simple waveforms
Space Invaders (1978): Simple descending bassline
Pac-Man (1980): Iconic waka-waka sound

8-bit era (early 1980s):

NES (1983): 5-channel PSG (Programmable Sound Generator)
Commodore 64 (1982): SID chip, 3-voice synth with filters
ZX Spectrum (1982): 1-bit "beeper" speaker

Technical limitations:

Limited polyphony: 3-5 simultaneous voices
Simple waveforms: Square, triangle, sawtooth, noise
No samples: All sounds synthesized in real-time
Low resolution: 4-8 bit DACs, 4-22 kHz sample rates

Creative solutions:

Pulse width modulation: Varying square wave duty cycle for timbre
Arpeggiation: Rapid note cycling to simulate chords
Noise channel: For percussion and effects
Programmable filters: On more advanced chips like SID

This plugin emulates: The DAC limitations (bit depth, sample rate) rather than the synthesis methods.

Aliasing Artifacts

ALIASING MATHEMATICS: When sampling at rate f_s: • Frequencies below f_s/2 (Nyquist) reproduced correctly • Frequencies above f_s/2 fold back (alias) Aliasing formula: f_alias = |f - k × f_s| Where k = integer that places f_alias in 0 to f_s/2 range Example: f_s = 8000 Hz, input frequency f = 5000 Hz Nyquist = 4000 Hz k = 1: f_alias = |5000 - 1×8000| = 3000 Hz Result: 5000 Hz input appears as 3000 Hz output Sample rate reduction process: 1. Original sample rate: f_original 2. Reduction factor: R 3. New sample rate: f_new = f_original / R 4. New Nyquist: f_new / 2 5. Any content above f_new/2 will alias Implementation in plugin: 1. Resample down to f_new (decimation) 2. Resample back up to f_original (interpolation) 3. Result: Aliased frequencies preserved at original sample rate Musical use of aliasing: • Creates metallic, bell-like tones • Adds inharmonic content • Creates "digital" character • Used intentionally in chiptune music

Processing Modes

Mode 1: Time Domain (Fast - Classic Bit Crusher)

⏱️ Direct Amplitude Quantization

Process: Operates directly on sample values in time domain. Fast, efficient, creates classic bit crusher sound.

Two-stage processing:

Bit depth reduction: Quantize amplitude values to discrete levels
Sample rate reduction (optional): Downsample and upsample to create aliasing

Characteristics: Creates stair-step waveforms, adds quantization noise, can create aliasing artifacts.

TIME DOMAIN ALGORITHM: Step 1: Bit depth reduction quant_steps = 2 ^ bit_depth For each sample value s in range [-1, 1]: s_quantized = round(s × quant_steps) / quant_steps Example (4-bit, quant_steps = 16): s = 0.73 s × quant_steps = 0.73 × 16 = 11.68 round(11.68) = 12 s_quantized = 12 / 16 = 0.75 Error = 0.02 Step 2: Sample rate reduction (if sample_rate_reduction > 1) original_sr = Get sampling frequency reduced_sr = original_sr / sample_rate_reduction # Downsample Resample: reduced_sr, 50 (50-point interpolation) # Upsample back to original rate Resample: original_sr, 50 Effect: Creates aliasing for frequencies above reduced_sr/2 Also creates step-like interpolation for waveforms Praat implementation: selectObject: originalSound result = Copy: originalName$ + "_bitcrushed" Formula: "round(self * quant_steps) / quant_steps" if sample_rate_reduction > 1 Resample: reduced_sr, 50 Resample: original_sr, 50 endif

Mode 2: Spectral (Slow - Experimental)

📊 Frequency Domain Quantization

Process: Convert to frequency domain via FFT, quantize spectral components, convert back to time domain.

Unique characteristics: Creates metallic, ringing sounds. Can process specific frequency bands differently.

Experimental nature: Results less predictable than time domain. Can create unusual textures not typical of classic bit crushing.

SPECTRAL PROCESSING ALGORITHM: Step 1: Convert to spectrum spectrum = To Spectrum: fast_fourier fast_fourier option: • "yes": Faster but lower frequency resolution • "no": Slower but higher frequency resolution Step 2: Apply spectral quantization Formula applied to each frequency bin: if x >= lower_frequency and x <= upper_frequency # Within processing band normalized = (x - lower_frequency) / (upper_frequency - lower_frequency) quantized = round(quantization_steps × normalized) / quantization_steps output = self × quantized else # Outside processing band output = self × outside_range_multiplier endif Where: x = frequency of current bin (Hz) self = original spectral amplitude at that frequency quantization_steps = number of discrete levels in band outside_range_multiplier = attenuation for non-processed frequencies Step 3: Convert back to time domain result = To Sound This inverse FFT creates time-domain signal with quantized frequency characteristics. EFFECTS: • Within band: Reduces frequency resolution, creates harmonic distortion • Outside band: Attenuates (if multiplier < 1) or passes through • Creates comb-filter-like effects • Can sound metallic or ring-mod-like

Mode Comparison

Aspect	Time Domain Mode	Spectral Mode
Speed	Fast (real-time for short files)	Slow (FFT processing intensive)
Character	Classic bit crusher, gritty, crunchy	Metallic, ringing, experimental
Processing	Direct amplitude quantization	Frequency domain quantization
Controls	Bit depth, sample rate reduction	Frequency band, quantization steps
Artifacts	Quantization noise, aliasing	Spectral distortion, comb filtering
Use case	8-bit emulation, lofi beats	Sound design, experimental textures
Predictability	High (consistent results)	Lower (depends on spectral content)

Parameter Details

TIME DOMAIN PARAMETERS: bit_depth (1-16): • Lower = more severe quantization • 16-bit: CD quality (subtle effect) • 12-bit: Early digital recorders • 8-bit: Classic arcade/telephone • 4-bit: Severe distortion • 2-bit: Extreme degradation • 1-bit: Essentially square waves sample_rate_reduction (1, 2, 4, 8...): • 1: No reduction (preserves bandwidth) • 2: Halves bandwidth, creates aliasing above Nyquist/2 • 4: Quarters bandwidth, severe aliasing • Higher: More aliasing, lower effective bandwidth Example (44.1 kHz original): Reduction 2 → 22.05 kHz effective Reduction 4 → 11.025 kHz effective Reduction 8 → 5.5125 kHz effective SPECTRAL PARAMETERS: lower_frequency / upper_frequency: • Define frequency band for processing • Outside band attenuated by outside_range_multiplier • Typical: 200-3000 Hz (speech range) • Narrow bands: 500-1500 Hz (focus on mids) quantization_steps (1+): • Number of discrete levels within band • 1: All frequencies in band reduced to single value • 2: Binary quantization (on/off) • 4-8: Moderate quantization • Higher: Finer spectral resolution outside_range_multiplier (0.0-1.0+): • 0.0: Complete attenuation outside band • 0.5: -6 dB outside band • 1.0: No attenuation outside band • >1.0: Amplification outside band (use carefully) fast_fourier (yes/no): • "yes": Faster FFT, lower frequency resolution • "no": Slower FFT, higher frequency resolution • For most uses: "yes" is sufficient

Built-in Presets

🎨 Curated Degradation Levels

Four presets covering common bit-crushed aesthetics from subtle coloration to extreme distortion.

Preset Details

Preset	Bit Depth	Sample Rate Reduction	Quantization Steps	Character	Use Case
Default (8-bit)	8 bits	1x (none)	2	Classic arcade, clear but crunchy	General 8-bit emulation
Subtle (12-bit)	12 bits	1x (none)	4	Warm lofi, subtle grit	Lofi beats, subtle coloration
Heavy (4-bit)	4 bits	2x	1	Aggressive distortion, noisy	Industrial, noise music
Extreme (2-bit)	2 bits	4x	1	Severe degradation, artifact-heavy	Experimental, glitch art

Preset Analysis

🎮 Default (8-bit) - Classic Arcade

Design philosophy: Authentic 8-bit video game sound.

Parameters:

8-bit depth: 256 amplitude levels, characteristic crunchy texture
No sample rate reduction: Preserves bandwidth for clarity
Spectral mode: 2 quantization steps (minimal processing)

Technical details:

8-bit characteristics: Amplitude levels: 2^8 = 256 Level spacing: 2/256 = 0.0078125 Dynamic range: ~50 dB Quantization noise: Moderate, "gritty" Perceptual effect: • Clear but noticeably digital • Added high-frequency "fizz" • Sounds like early digital samplers • Nostalgic video game character

Best source material: Simple waveforms (sine, square), speech, percussion, synth sounds.

Musical result: Clean but crunchy digital sound, reminiscent of NES/arcade games.

🎵 Subtle (12-bit) - Warm Lofi

Design philosophy: Subtle coloration, warm lofi character without extreme distortion.

Parameters:

12-bit depth: 4096 amplitude levels, subtle quantization
No sample rate reduction: Preserves frequency content
Spectral mode: 4 quantization steps (gentle spectral shaping)

Technical details:

12-bit characteristics: Amplitude levels: 2^12 = 4096 Level spacing: 2/4096 = 0.000488 Dynamic range: ~74 dB Quantization noise: Subtle, "warm" Comparison: • CD audio: 16-bit (65,536 levels) • 12-bit: Early digital samplers (Akai S900, etc.) • Perceived as "warm" or "vintage" digital Use in music production: • Adds subtle grit to clean sources • Creates "sampled" character • Works well on drums, vocals • Popular in lofi hip-hop

Best source material: Drums, vocals, acoustic instruments, full mixes.

Musical result: Warm, vintage digital character without severe distortion.

⚡ Heavy (4-bit) - Aggressive Distortion

Design philosophy: Pronounced digital distortion for aggressive sounds.

Parameters:

4-bit depth: Only 16 amplitude levels, severe quantization
2x sample rate reduction: Adds aliasing artifacts
Spectral mode: 1 quantization step (extreme spectral reduction)

Technical details:

4-bit characteristics: Amplitude levels: 2^4 = 16 Level spacing: 2/16 = 0.125 Dynamic range: ~26 dB Quantization noise: Severe, "crunchy" Sample rate reduction (2x): Original 44.1 kHz → effective 22.05 kHz Aliasing for frequencies above 11.025 kHz Creates metallic ringing Combined effect: • Extreme stair-step waveforms • Added aliasing artifacts • Very noisy, distorted • Industrial/experimental character

Best source material: Drums, distorted guitars, noise textures, industrial sounds.

Musical result: Aggressive digital distortion with pronounced artifacts.

💥 Extreme (2-bit) - Digital Destruction

Design philosophy: Maximum degradation for experimental/glitch sounds.

Parameters:

2-bit depth: Only 4 amplitude levels, extreme quantization
4x sample rate reduction: Severe aliasing and bandwidth reduction
Spectral mode: 1 quantization step (maximum spectral reduction)

Technical details:

2-bit characteristics: Amplitude levels: 2^2 = 4 Level spacing: 2/4 = 0.5 Dynamic range: ~14 dB Essentially converts to square waves Sample rate reduction (4x): Original 44.1 kHz → effective 11.025 kHz Aliasing for frequencies above 5.5125 kHz Severe bandwidth limitation Result: • Near-binary amplitude (4 levels) • Extreme aliasing artifacts • Very low effective bandwidth • Sounds "broken" or "glitched"

Best source material: Experimental sources, glitch material, sound design elements.

Musical result: Extreme digital destruction, barely recognizable source material.

Parameter Reference

Parameter	Type	Default	Range	Description
Preset	option	Default (8-bit)	1-4	Pre-configured degradation levels
Mode	option	Time Domain	1-2	Processing algorithm
bit_depth	positive	8	1-16	Amplitude resolution in bits
sample_rate_reduction	positive	1	1-16	Sampling rate divider
fast_fourier	boolean	yes	yes/no	FFT speed vs resolution tradeoff
lower_frequency	positive	200	20-20000	Low frequency bound for spectral processing
upper_frequency	positive	3000	20-20000	High frequency bound for spectral processing
quantization_steps	positive	2	1-100	Discrete levels in spectral band
outside_range_multiplier	positive	0.5	0.0-2.0	Attenuation for non-processed frequencies
scale_peak	positive	0.99	0.1-1.0	Output peak normalization level
play_after_processing	boolean	1	0/1	Auto-play processed sound
keep_intermediate_objects	boolean	0	0/1	Keep FFT spectrum objects

Applications

Music Production & Sound Design

Use case: Add character, grit, and vintage digital aesthetic to productions

Techniques:

Lofi hip-hop: 8-12 bit on drums, samples, full mixes
Chiptune/bitpop: Authentic 8-bit arcade sounds
Industrial music: Heavy 4-bit distortion on percussion
Glitch/experimental: Extreme 2-bit destruction
Vocal processing: Subtle bit reduction for vintage character
Drum processing: Add punch and grit to drums

Film, Game & Media Sound Design

Use case: Create retro, futuristic, or degraded sound effects

Techniques:

Retro game sounds: Authentic 8-bit effects
Futuristic UI: Clean digital artifacts for sci-fi interfaces
Robot voices: Bit-crushed speech
Radio effects: Combined with filtering for vintage radio
Glitch transitions: Extreme reduction for scene changes
Memory/dream sequences: Degraded audio for flashbacks

Educational Demonstrations

Use case: Teach digital audio concepts through auditory examples

Techniques:

Quantization demonstration: Show bit depth effects on clean signals
Aliasing examples: Demonstrate Nyquist theorem violations
Historical audio formats: Recreate early digital sound quality
Comparative listening: A/B different bit depths and sample rates
Artifact identification: Train ears to recognize digital artifacts

Practical Workflow Examples

🎵 Lofi Hip-Hop Beat Processing

Goal: Add warm, vintage character to drum loop

Settings:

Preset: Subtle (12-bit)
Mode: Time Domain
Adjust: bit_depth = 10 (custom between 8 and 12)
Adjust: sample_rate_reduction = 2 (adds subtle aliasing)
Process: Drum loop only

Result: Drums have added grit and vintage character

Mix technique: Parallel processing: blend crushed version with clean

Additional processing: Add vinyl crackle, tape saturation, light compression

🎮 Retro Game Sound Effect Creation

Goal: Create authentic 8-bit game sounds from modern recordings

Settings:

Preset: Default (8-bit)
Mode: Time Domain
bit_depth: 8 (keep default)
sample_rate_reduction: 2 (emulates lower sample rates of old systems)
Source: Various sound effects (explosions, jumps, collectibles)

Processing chain:

Start with clean, simple sound effect
Apply bit crusher with 8-bit/2x reduction
Optional: Add simple synthesizer elements (square waves, noise)
Optional: Limit to short duration (old systems had memory limits)
Export as WAV for game engine

Result: Authentic 8-bit game sounds

🔬 Educational Demo: Nyquist Theorem

Goal: Demonstrate aliasing through audible examples

Method:

Create sine wave sweep 20 Hz to 20 kHz (10 seconds)
Apply bit crusher with sample_rate_reduction = 4
Original: 44.1 kHz → Effective: 11.025 kHz → Nyquist: 5.5125 kHz
Listen: Frequencies above 5.5 kHz "fold back" as descending tones
Repeat with different reduction factors (2, 8, 16)

Teaching points:

Hear aliasing as metallic, bell-like tones
Understand frequency folding mathematically
See importance of anti-aliasing filters
Historical context: Early digital systems without proper filtering

Advanced Techniques

Creative parameter combinations:

Progressive degradation: Script that gradually reduces bit depth over time
Frequency-dependent bit depth: High frequencies more crushed than lows (simulates old codecs)
Multi-stage processing: Apply different presets to different frequency bands
Dynamic bit depth: Link bit depth to amplitude (louder = more bits)
Stereo imaging: Different settings for left/right channels

Source material optimization:

Clean sources: Show artifacts most clearly (good for demonstrations)
Already distorted material: May become too noisy (use subtler settings)
Complex signals: Hide artifacts in complexity (can use more extreme settings)
Simple waveforms: Clearly show stair-step quantization
Transient-rich material: Drums respond well to bit crushing
Sustained tones: Show aliasing effects clearly

Troubleshooting Common Issues

Problem: Output too quiet/noisy
Causes: Severe quantization reduces average amplitude, increases noise floor
Solutions: Increase scale_peak, use parallel processing (blend with clean), use less extreme settings

Problem: Unpleasant aliasing artifacts
Causes: Sample rate reduction too high for source material
Solutions: Reduce sample_rate_reduction, use spectral mode with narrower band, filter before processing

Problem: Spectral mode very slow
Causes: Large files, high sample rates, fast_fourier = no
Solutions: Use Time Domain mode, set fast_fourier = yes, process shorter segments

Problem: No audible effect
Causes: Bit depth too high (≥14), sample rate reduction = 1, subtle source material
Solutions: Reduce bit depth (≤12), increase sample_rate_reduction (≥2), use more distinctive source

Problem: Extreme distortion/clipping
Causes: Very low bit depth (≤3), high input levels
Solutions: Increase bit depth (≥4), reduce input amplitude before processing, use scale_peak < 1.0

Problem: Praat becomes unresponsive during spectral processing
Causes: Very large files, insufficient memory
Solutions: Process shorter segments, increase Praat memory allocation, use Time Domain mode

Performance Considerations

PROCESSING SPEED: TIME DOMAIN MODE: • Very fast: O(N) where N = number of samples • Real-time for typical audio lengths • Sample rate reduction adds resampling cost but still fast Example (1-minute 44.1 kHz mono): Samples = 60 × 44100 = 2,646,000 Processing time: < 1 second SPECTRAL MODE: • Slow: O(N log N) for FFT • Fast Fourier option reduces time but also frequency resolution • Large files may take noticeable time Example (1-minute 44.1 kHz mono): FFT size determined by Praat Processing time: 5-30 seconds depending on settings MEMORY USAGE: Time Domain: Minimal (input + output buffers) Spectral: Additional FFT buffers (temporary, cleaned up if keep_intermediate_objects = no) OPTIMIZATION TIPS: • For long files: Use Time Domain mode • For experimentation: Process short segments first • For spectral mode: Use fast_fourier = yes • Clean workspace: keep_intermediate_objects = no (default) • Batch processing: Script multiple files with same settings QUALITY CONSIDERATIONS: Time Domain: Mathematical precision (rounding) Spectral: FFT artifacts possible (windowing, spectral leakage) Resampling: 50-point interpolation (good quality)

Historical & Technical Context

Evolution of Digital Audio Quality

📅 Timeline of Digital Audio

1970s: Early digital

PCM adapters using video tape (1970s)
First commercial digital recordings (1971)
13-bit resolution, ~32 kHz sample rates

1980s: Consumer digital

CD format introduced (1982): 16-bit, 44.1 kHz
Early digital samplers: 8-12 bit (Fairlight, Synclavier)
Video game sound chips: 4-8 bit PSG

1990s: Prosumer digital

ADAT, DA-88: 16-bit digital tape
Early computer audio: 8-16 bit sound cards
MP3 compression introduced (1995)

2000s+: High-resolution digital

24-bit, 96+ kHz common in production
DSP power enables complex processing
Ironically: plugins to emulate earlier limitations

This plugin emulates: 1970s-1980s digital audio limitations that are now considered "character" rather than "defect."

Classic Hardware Bit Crushers

🎛️ Hardware Predecessors

Early digital samplers with low resolution:

Fairlight CMI (1979): 8-bit, 24 kHz
Synclavier (1975): 8-bit, variable rates
E-mu Emulator (1981): 8-bit, 27.7 kHz
Akai S612 (1985): 12-bit, 32 kHz

Video game sound chips:

AY-3-8910 (1978): 3-voice PSG, used in many arcade games
Texas Instruments SN76489 (1979): 4-voice PSG, used in Sega Master System
Ricoh 2A03 (1983): 5-channel PSG, Nintendo Entertainment System
MOS Technology 6581 (1982): SID chip, Commodore 64

Modern hardware bit crushers:

Electro-Harmonix Mainframe: Dedicated bit crusher pedal
Malekko Scrutator: Bit crusher with filter
WMD Geiger Counter: Bit crusher with wavetable distortion
Industrialectric Incinerator: Extreme digital distortion

This plugin approach: Software emulation of the DAC limitations rather than the synthesis methods of these chips.

Related Digital Audio Effects

OTHER DIGITAL DEGRADATION EFFECTS: 1. SAMPLE RATE REDUCTION (DECIMATION): • Reduces temporal resolution • Creates aliasing artifacts • Often combined with bit crushing • This plugin includes this as sample_rate_reduction 2. WAVEFOLDING/DIGITAL CLIPPING: • Different from analog clipping • Creates harsh, digital distortion • Not implemented in this plugin (focus on quantization) 3. DATA COMPRESSION ARTIFACTS: • MP3, AAC, etc. artifacts • Pre-echo, quantization noise, spectral holes • Different character from pure bit crushing 4. GRANULAR SYNTHESIS/DEGRADATION: • Time-based segmentation and scrambling • Creates stutter, glitch effects • Complementary to bit crushing 5. DIGITAL DELAY DEGRADATION: • Reducing bit depth in delay feedback • Creates decaying digital artifacts • Classic "degrading delay" effect THIS PLUGIN'S FOCUS: Pure amplitude quantization (bit depth reduction) Optional temporal decimation (sample rate reduction) Spectral domain variant (experimental) COMPLEMENTARY PRAAT PLUGINS: Could combine with: • Polynomial Envelope Shaper (dynamics before/after crushing) • XMod (ring modulation after bit crushing) • GENDYN Synthesis (bit-crushed stochastic textures) • Visual Game of Life (crushed sonification)

Aesthetic Considerations

🎨 The Art of Digital Degradation

Why degrade pristine digital audio?

Nostalgia:

Sounds of childhood (video games, early computers)
Historical recording quality
"Vintage" digital as aesthetic category

Texture & character:

Adds grit, noise, imperfection
Contrasts with clean digital production
Creates "human" element in digital domain

Creative constraint:

Limitations inspire creativity
Forces focus on essential elements
Historical practice: composers working within technical limits

Cultural commentary:

Critique of "perfect" digital audio
Embrace of technological artifacts
Post-digital aesthetics

This plugin philosophy: Provide tools for intentional degradation as creative choice, not as "fix" for poor quality.

What this does

🎮 8-Bit Arcade Sound Aesthetic

🔢 Two Processing Approaches

Quick start

Digital Audio Theory

Digital Audio Fundamentals

🔢 Sampling & Quantization

Quantization Noise & Distortion

📊 Error Characteristics

Historical Context: 8-Bit Sound Chips

🎮 Classic Arcade & Console Sound

Aliasing Artifacts

Processing Modes

Mode 1: Time Domain (Fast - Classic Bit Crusher)

⏱️ Direct Amplitude Quantization

Mode 2: Spectral (Slow - Experimental)

📊 Frequency Domain Quantization

Mode Comparison

Parameter Details

Built-in Presets

🎨 Curated Degradation Levels

Preset Details

Preset Analysis

🎮 Default (8-bit) - Classic Arcade

🎵 Subtle (12-bit) - Warm Lofi

⚡ Heavy (4-bit) - Aggressive Distortion

💥 Extreme (2-bit) - Digital Destruction

Parameter Reference

Applications

Music Production & Sound Design

Film, Game & Media Sound Design

Educational Demonstrations

Practical Workflow Examples

🎵 Lofi Hip-Hop Beat Processing

🎮 Retro Game Sound Effect Creation

🔬 Educational Demo: Nyquist Theorem

Advanced Techniques

Troubleshooting Common Issues

Performance Considerations

Historical & Technical Context

Evolution of Digital Audio Quality

📅 Timeline of Digital Audio

Classic Hardware Bit Crushers

🎛️ Hardware Predecessors

Related Digital Audio Effects

Aesthetic Considerations

🎨 The Art of Digital Degradation

Further Reading & Resources