Skip to content

flues-synth

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

flues-synth

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
docs
docs
 
 
include
include
 
 
src
src
 
 
tests
tests
 
 
web-ui
web-ui
 
 
README.md
README.md
 
 
flues-play.service
flues-play.service
 
 
meson.build
meson.build
 
 
meson_options.txt
meson_options.txt
 
 
start-service.sh
start-service.sh
 
 

README.md

Flues-Synth

Headless Hybrid Speech/Physical Modeling Synthesizer for Raspberry Pi

Unified polyphonic (4-voice default) synthesizer combining Disyn distortion algorithms, Chatterbox formant synthesis, and PM-Synth physical modeling. Designed for headless operation on Raspberry Pi 4 with MIDI control.

Features

  • Disyn Oscillators: 17 distortion synthesis algorithms (7 original + 10 extended with param3 control: Dirichlet Pulse, DSF Single/Double, Tanh Square/Saw, PAF, Modified FM, plus extended variants)
  • Chatterbox Formants: 4-formant cascade (F1-F4) with nasal resonance and vocal modes (Sing, Shout, Nasal, Fry)
  • PM-Synth Physical Modeling: 12 interface strategies (Pluck, Hit, Reed, Flute, Brass, Bow, Bell, Drum, Crystal, Vapor, Quantum, Plasma)
  • Dual DC Blocking: Prevents feedback latching with -60dB DC rejection
  • Calibrated Signal Levels: Optimized 0.81 peak output for maximum S16_LE bit depth usage
  • ALSA Audio: Direct PCM output (32 kHz default, 512 samples; configure at build/runtime)
  • ALSA MIDI: Auto-connects to best external MIDI port on startup
  • Headless Control: Comprehensive MIDI CC mapping (29 parameters + 6 control notes)
  • Polyphony: 4 voices by default (set -Dmax_voices= at configure time)

Signal Flow

Signal Flow Diagram

Detailed Signal Path

MIDI Input (Note + 29 CCs)
  ↓
Disyn Oscillators (7 algorithms, ×0.8 level)
  ↓
[DC Blocker 1] (R=0.999, catches DC at source)
  ↓
Sources (Noise 0.15 + DC) → Mix
  ↓
Envelope (AR, exponential time mapping, gate-driven)
  ↓
Chatterbox Formants (F1→F2→F3→F4 cascade, ×2.0 makeup)
  ↓
+ Feedback ← [DC Blocker 2] ← Feedback Mix ← Delay1/2 + Filter
  ↓
Interface Strategy (12 physical models, nonlinear)
  ↓
Dual Delay Lines (pitch-tracked with tuning/ratio)
  ↓
State-Variable Filter (LP/BP/HP morph, Q control)
  ↓
× AM Modulation (LFO bipolar AM↔FM)
  ↓
Global Processing (×0.85 pad → tanh clip → ×0.95 master)
  ↓
ALSA Audio Out (peak ~0.81, optimized for S16_LE bit depth)

Key Architecture Features

  • Dual DC Blocking: R=0.999 (-60dB @ DC, -3dB @ 7.6 Hz)
    • DC Blocker 1 after Disyn: Catches DC at source
    • DC Blocker 2 on feedback path: Prevents loop accumulation
  • Level Management:
    • Disyn level: 0.8 (user adjustable via CC 19)
    • Formant makeup: 2.0× (compensates for Q-induced attenuation)
    • Global pad: 0.85× + soft clip (tanh) + master 0.95×
    • Final peak: ~0.81 (optimized for S16_LE, with tanh safety)
  • Debug Toggles: MIDI notes 36-41 toggle DSP sections (disabled by default; enable with FLUES_CONTROL_NOTES=1)

Requirements

Raspberry Pi 4

  • Raspberry Pi OS (64-bit recommended)
  • 2GB+ RAM

Dependencies

sudo apt update
sudo apt install build-essential meson ninja-build libasound2-dev

Build Instructions

Download Pre-Built Release (Recommended)

Pre-built binaries optimized for Raspberry Pi 4 (aarch64) are available:

Download Latest Release

Download flues-synth-v*.tar.gz, extract, and install:

tar xzf flues-synth-v0.1.0-aarch64.tar.gz
cd flues-synth-v0.1.0
sudo cp flues-synth/flues-synth /usr/local/bin/
sudo chmod +x /usr/local/bin/flues-synth
flues-synth  # Run

Build From Source

On Raspberry Pi (recommended for optimal performance):

cd flues-synth
meson setup builddir
meson compile -C builddir

For multi-platform releases (desktop + Pi builds), see ../scripts/MULTI_PLATFORM_BUILD.md for the complete workflow.

Configuration Options

Set voice count (default: 4):

meson setup builddir -Dmax_voices=8
meson compile -C builddir

Enable ARM NEON optimizations (Raspberry Pi 4):

meson setup builddir -Denable_simd=true
meson compile -C builddir

Change audio device (run-time):

./builddir/flues-synth hw:1,0

Clean Build

rm -rf builddir
meson setup builddir
meson compile -C builddir

Running

Start the Synthesizer

cd flues-synth
./builddir/flues-synth

Expected output:

=== Flues-Synth v0.1.0 ===
Unified Polyphonic Synthesizer (4-voice default)
Target: Raspberry Pi 4 (ARM Cortex-A72)
Control notes 36-42: disabled (set FLUES_CONTROL_NOTES=1 to enable)

Synth engine initialized
...

Manual MIDI Connection

If no MIDI device is found, connect manually:

# List ALSA MIDI ports
aconnect -l

# Connect your MIDI device to Flues-Synth
aconnect 20:0 128:0

MIDI Feedback

The synth prints CC names on first occurrence:

CC ch1: 20 -> 64  [Noise Level]
CC ch1: 71 -> 45  [F1 (Jaw)]
CC ch1: 20 -> 65
CC ch1: 20 -> 66

After the first move of each CC, subsequent values show only numbers for cleaner console output.

Controller CCs (sliders)

Set your nine sliders to these CC numbers (left→right): 73, 72, 28, 30, 74, 71, 1, 27, 7. These feed the per-program maps in src/midi_mapping.c (Program 0: Disyn Alg, P1, P2, Interface Type, Intensity, Tuning, Delay1 Feedback, Attack, Release). Compatibility aliases exist for common legacy defaults (91/92/93/5/84), but if you see “Unmapped CC…” in the console, reprogram the controller to the canonical list so parameter names and ranges match.

Environment toggles

  • FLUES_CONTROL_NOTES=1 – enable debug control notes 36-41 (default: off to avoid stealing low notes from DAWs).
  • FLUES_MIDI_DEBUG=1 – verbose MIDI event logging (note on/off/CC/All Notes Off with ALSA source port).

Stop the Synthesizer

Press Ctrl+C for clean shutdown.

Autostart on Raspberry Pi

See docs/startup.md for a systemd user service that launches /home/danny/github/flues/play on boot and restarts it if it crashes.

How it works

See docs/how-it-works.md for a quick architecture walkthrough and block descriptions.

Troubleshooting

Low-frequency hum (~50Hz) when idle

  • First isolate the synth: send CC 7 (Master Gain) to 0, or trigger control note 41 (Hard Mute, requires FLUES_CONTROL_NOTES=1). If the hum remains, it is coming from the audio path, not the DSP.
  • Toggle sections to see if noise changes: control notes 36 (Noise) and 38 (Feedback) off/on. If no change, suspect hardware/ALSA.
  • Check the output device: hw:0,0 analog outs can pick up mains/ground noise. Try a USB DAC (./builddir/flues-synth hw:2,0) or plughw to add ALSA conversion.
  • Verify baseline with ALSA: speaker-test -D hw:0,0 -c1 -t s16 -f 0 should be silent; hum here indicates device/grounding, not the synth.
  • Ensure capture paths are muted in alsamixer (mic/boost off) and avoid sharing loopback capture with the playback device.

Audio Device Setup

Using a USB DAC

To use an external USB DAC instead of the built-in audio output:

1. Detect Available Audio Devices

List all playback hardware devices:

aplay -l

Example output:

card 0: Headphones [bcm2835 Headphones], device 0: bcm2835 Headphones [bcm2835 Headphones]
card 2: DAC [USB Audio DAC], device 0: USB Audio [USB Audio]

In this example, the USB DAC is hw:2,0 (card 2, device 0).

2. List Device Names (Including Plugin Devices)

aplay -L

Shows additional options:

  • hw:X,Y - Direct hardware access
  • plughw:X,Y - Hardware with automatic format conversion
  • default:CARD=XXX - Named card reference

3. Test the USB DAC

Generate a test tone to verify the device works:

# Mono test on hw:2,0
speaker-test -D hw:2,0 -c 1 -t sine

# Stereo test
speaker-test -D hw:2,0 -c 2 -t sine

Press Ctrl+C to stop the test.

4. Run Flues-Synth with USB DAC

Pass the device as a command-line argument:

./builddir/flues-synth hw:2,0

5. Set USB DAC as Default (Optional)

To make the USB DAC the system default, create or edit ~/.asoundrc:

nano ~/.asoundrc

Add:

pcm.!default {
    type hw
    card 2
    device 0
}

Then run flues-synth without arguments:

./builddir/flues-synth

6. Check Active Audio

See which devices are currently in use:

cat /proc/asound/pcm

Audio Device Auto-Detection

If no device is specified, flues-synth tries these in order:

  1. Command-line argument (e.g., hw:2,0)
  2. hw:Headphones (Raspberry Pi headphone jack)
  3. plughw:Headphones
  4. hw:2,0 (common USB DAC location)
  5. plughw:2,0
  6. hw:1,0
  7. default

Override with: ./builddir/flues-synth <device-name>

Audio Format and Quality

Flues-synth supports multiple ALSA PCM formats with automatic fallback:

Preferred Formats (tried in order):

  1. S16_LE (16-bit signed little-endian) - Most widely supported
  2. FLOAT_LE (32-bit float) - Best quality, less common

S16_LE Signal Optimization:

  • Bit depth usage: 81% of full scale (0.85 × 0.95 master gain)
    • Effective range: ±26,542 of ±32,768 possible values
    • Uses ~15.4 of 16 available bits for optimal SNR
  • TPDF Dithering: Triangular PDF dither (±1 LSB) linearizes quantization
    • Eliminates low-level distortion and quantization artifacts
    • Decorrelates quantization noise from signal
    • Improves perceived SNR by ~3-4 dB
  • Proper Rounding: Uses round-to-nearest instead of truncation to prevent DC bias

Why This Matters:

  • Without dithering, low-level signals (quiet passages, reverb tails) produce audible "digital grunge"
  • TPDF dither trades a tiny amount of noise floor (~-96 dBFS) for much cleaner low-level reproduction
  • Signal levels optimized to use maximum bit depth without clipping

Format Detection:

# Check what format your device negotiated
cat /proc/asound/card*/pcm*p/sub*/hw_params

Example output:

access: RW_INTERLEAVED
format: S16_LE
subformat: STD
channels: 1
rate: 48000 (48000/1)

MIDI Control

Full MIDI specification: See docs/midi.md for complete reference including:

  • 29 MIDI Programs (0-28) with signal chain configurations
  • 29 MIDI CCs with exponential/bipolar scaling formulas
  • 6 Control Notes (36-41) for debug toggles
  • Complete parameter mapping and ranges

Quick reference for common controls:

Control Notes (36-42) - Debug Toggles

Special MIDI notes toggle DSP sections for diagnostics and isolation:

Note Function Description
36 Toggle Noise Enable/disable white noise source
37 Toggle Disyn Enable/disable Disyn oscillator
38 Toggle Feedback Enable/disable feedback loop
39 Toggle Formants Enable/disable formant cascade
40 Toggle Filter Enable/disable SVF filter
41 Hard Mute Emergency mute (clears delays, DC blockers, filters)
42 Reset Reset all parameters to defaults and clear all notes
48+ Musical Notes Standard MIDI note range (C3 = 48)

Usage: Press any of notes 36-42 to toggle or reset. Console shows:

Ctl Note 36: Noise ENABLED
Ctl Note 37: Disyn DISABLED

MIDI CC Mapping

Standard Controls

CC Parameter Range Description
1 Intensity 0-1 Interface intensity (mod wheel)
7 Master Gain 0-1 Output volume
10 F2 (Tongue) 500-3000 Hz Second formant frequency
71 F1 (Jaw) 200-1000 Hz First formant frequency
72 Release 0-1 Envelope release time
73 Attack 0-1 Envelope attack time
74 F3 (Lips) 1500-4000 Hz Third formant frequency
75 F4 (Quality) 2500-4500 Hz Fourth formant frequency

Vocal Modes (Toggle: ≥64 = ON)

CC Parameter Description
80 Nasal Enable nasal formant (250 Hz)
81 Sing Vibrato (5.5 Hz ±1.5%)
82 Shout 15% formant frequency boost
83 Fry Vocal fry (f₀/2 subharmonic)

Disyn Source

CC Parameter Range Description
16 Algorithm 0-6 Disyn synthesis algorithm
17 Param1 0-1 Algorithm parameter 1
18 Param2 0-1 Algorithm parameter 2
19 Disyn Level 0-1 Oscillator mix level
20 Noise Level 0-1 White noise mix level
21 DC Level 0-1 DC offset mix level

Interface & Delay

CC Parameter Range Description
24 Interface Type 0-11 Physical model (see below)
26 Tuning -12 to +12 Delay line pitch offset
27 Ratio 0.5 to 2.0 Delay2/Delay1 length ratio

Feedback

CC Parameter Range Description
28 Delay1 Feedback 0-1 First delay return level
29 Delay2 Feedback 0-1 Second delay return level
30 Filter Feedback 0-1 Filter output return level

Filter

CC Parameter Range Description
32 Frequency 20-20000 Hz Filter cutoff frequency
33 Q 0.1-10 Filter resonance
34 Shape 0-1 LP (0) → BP (0.5) → HP (1)

Modulation

CC Parameter Range Description
36 LFO Freq 0.1-20 Hz Modulation LFO frequency
37 AM ↔ FM -1 to +1 AM (-1) ↔ None (0) ↔ FM (+1)

Interface Types

ID Name Description Character
0 Pluck Karplus-Strong plucked string Bright, metallic attack
1 Hit Struck/mallet percussion Sharp transient, decay
2 Reed Clarinet-style reed instrument Woody, breathy
3 Flute Jet-edge aerophone Airy, hollow
4 Brass Lip-buzz brass instrument Buzzy, brassy
5 Bow Bowed string (friction) Sustained, singing
6 Bell Metallic resonator Bright, ringing
7 Drum Membrane percussion Thuddy, resonant
8 Crystal Golden-ratio coupled oscillators Glassy, shimmering
9 Vapor Chaotic turbulent flow Unstable, evolving
10 Quantum Phase-locked harmonic coupling Metallic, interference
11 Plasma Amplitude-driven energy feedback Aggressive, growling

Disyn Algorithms

17 Total Algorithms (7 original + 10 extended with param3 control):

ID Name Description
0 Dirichlet Pulse Sum of harmonics (Param1: harmonics)
1 DSF Single Discrete summation formula
2 DSF Double Two DSF oscillators (Param2: detune)
3 Tanh Square Waveshaped square wave
4 Tanh Saw Waveshaped sawtooth wave
5 PAF Phase-aligned formant
6 Modified FM Phase-modulated FM synthesis
7-16 Extended Variants Enhanced algorithms with param3 control

Note: Algorithms 7-16 are extended variants that use param3 for additional spectral control. See docs/algorithms.md for complete algorithm documentation.

Troubleshooting

No Audio Output

  1. Check Disyn level: Default 0.8, boost via CC 19 if needed
  2. Check Master gain: Default 0.5, boost via CC 7 if needed
  3. Verify parts not toggled off: Use notes 36-41 to check status
  4. Test audio device: 
    # List audio devices
aplay -l

# Test device
speaker-test -D default -c 1

# Try different device
./builddir/flues-synth hw:1,0

Low Output Level

If output is too quiet with only Disyn enabled (formants/feedback off):

  1. Boost Disyn level: CC 19 to 100-127 (0.78-1.0)
  2. Boost Master gain: CC 7 to 100-127
  3. Reason: Default levels (0.8 Disyn, 0.5 master) are calibrated for full signal chain

Clipping/Distortion

  1. Reduce Disyn level: CC 19 to 30-60 (0.24-0.47) when using formants
  2. Turn off Shout mode: CC 82 < 64 (15% boost can cause clipping)
  3. Lower feedback: Reduce CC 28-30 (delay/filter returns)
  4. Check formant tuning: Extreme F1-F4 values can cause resonance buildup

DC Offset / Audio Lockup

If audio locks up or produces continuous tone:

  1. Hard mute: Press MIDI note 41 (clears delays, DC blockers, filters)
  2. Disable feedback: Press note 38 to isolate feedback loop
  3. Check DC level: CC 21 should be 0 (no DC injection)
  4. Restart synth: DC blockers reset on note on, but hard mute is faster

DC Blocking Protection:

  • Two-stage DC blocking (R=0.999, -60dB @ DC)
  • DC Blocker 1 after Disyn catches source DC
  • DC Blocker 2 on feedback prevents loop accumulation
  • Hard mute (note 41) resets all DC blockers

No MIDI Connection

# List MIDI ports
aconnect -l

# Check if device is detected
cat /proc/asound/cards

# Manual connection
aconnect <source-port> <flues-synth-port>

Buffer Underruns (ALSA XRUN)

  • Increase buffer size in include/config.h: 
    #define DEFAULT_BUFFER_SIZE 1024  // Increase from 512
  • Rebuild: meson compile -C builddir

High CPU Usage

  • Disable ARM optimizations: meson setup builddir -Denable_simd=false
  • Reduce voice count: meson setup builddir -Dmax_voices=1
  • Lower sample rate to 44100 Hz in config.h

Build Errors

# Clean build
rm -rf builddir
meson setup builddir
meson compile -C builddir

# Check dependencies
dpkg -l | grep -E 'meson|ninja|alsa'

Performance

Raspberry Pi 4 (Cortex-A72):

  • Single voice: ~5-10% CPU @ 48kHz
  • Target (4 voices): ~20-30% CPU
  • Latency: ~5.3ms (256 samples)

Testing

Three test suites verify correct operation:

# Run all tests
meson test -C builddir

# Individual tests
./builddir/engine-smoke    # Basic signal generation (RMS check)
./builddir/envelope-test   # Envelope attack/sustain/release
./builddir/disyn-levels    # Disyn algorithm level analysis

Expected output:

1/3 engine-smoke  OK    (RMS: 0.019+, verifies non-zero output)
2/3 envelope-test OK    (attack/sustain/release phases detected)
3/3 disyn-levels  OK    (all 7 algorithms within safe bounds)

Smoke Test Details:

  • Renders 1 second of audio with default settings
  • Measures RMS level of last 0.5 seconds
  • Verifies RMS > 0.01 (signal present, not zero)
  • Catches DC blocker issues, envelope failures, signal path breaks

Envelope Test Details:

  • Tests note on triggers envelope attack
  • Verifies sustained tone has stable output
  • Tests note off triggers release decay
  • Tests rapid note retriggering (5 cycles)

Disyn Levels Test Details:

  • Measures peak/RMS for all 7 algorithms
  • Sweeps Param1/Param2 across full ranges
  • Verifies peaks within safe bounds (<2.2)
  • Reports clipping warnings if any

Notes for Raspberry Pi Deployment

  • The app auto-falls back through common Pi audio devices if no CLI device given: hw:Headphones, plughw:Headphones, hw:2,0, plughw:2,0, hw:1,0, then default
  • Override with: ./builddir/flues-synth hw:2,0
  • Quick rebuild on Pi: meson setup builddir --reconfigure && meson compile -C builddir && meson test -C builddir
  • Enable MIDI debug: FLUES_MIDI_DEBUG=1 ./builddir/flues-synth hw:2,0

Architecture

flues-synth/
├── include/
│   ├── config.h                    # Build-time configuration
│   ├── dsp_modules.h               # Module interfaces
│   ├── dsp_utils.h                 # DSP utilities
│   ├── synth_engine.h              # Main coordinator
│   ├── audio_backend_alsa.h        # ALSA audio
│   └── midi_backend_alsa.h         # ALSA MIDI
├── src/
│   ├── main.c                      # Entry point + CC mapping
│   ├── synth_engine.c              # Voice management + signal flow
│   └── audio/
│       ├── audio_backend_alsa.c    # ALSA PCM output
│       ├── midi_backend_alsa.c     # ALSA sequencer input
│       └── modules/
│           ├── disyn_wrapper.cpp   # C++ Disyn → C interface
│           ├── sources_module.c    # Noise + DC
│           ├── envelope_module.c   # Attack/Release
│           ├── formant_module.c    # Biquad bandpass filter
│           ├── formant_bank_module.c # 4-formant cascade
│           ├── interface_module.c  # Strategy pattern
│           ├── delay_lines_module.c # Dual delays
│           ├── feedback_module.c   # 3-way mixer
│           ├── filter_module.c     # State-variable filter
│           ├── modulation_module.c # LFO
│           └── strategies/         # 12 interface implementations
├── meson.build                     # Build configuration
├── meson_options.txt               # Build options
└── README.md                       # This file

Documentation

Development Roadmap

Phase 1 (Current) ✅

  • ✅ Single-voice operation
  • ✅ Disyn + Chatterbox + PM-Synth integration
  • ✅ ALSA audio/MIDI backends with auto-connect
  • ✅ 29 MIDI CC controls with name printing
  • ✅ 6 control notes for debug toggles
  • ✅ Dual DC blocking protection
  • ✅ Calibrated signal levels (0.81 peak, optimized for S16_LE)
  • ✅ TPDF dithering for clean 16-bit output
  • ✅ Test suite (engine-smoke, envelope-test, disyn-levels)

Phase 2 (Next)

  • ☐ 4-voice polyphony
  • ☐ Voice allocation and stealing
  • ☐ Per-voice parameter management
  • ☐ Polyphonic aftertouch

Phase 3 (Future)

  • ☐ GTK4 UI (optional)
  • ☐ MIDI learn
  • ☐ Preset system
  • ☐ Performance profiling

Recent Changes

v0.1.0 Release (December 2025)

First official release with 29 MIDI programs (expanded from 8), 17 Disyn algorithms (expanded from 7), and multi-platform build support.

Download: GitHub Releases

Key Features:

  • 29 MIDI Programs: Comprehensive preset library covering physical modeling, distortion synthesis, hybrid chains, and experimental combinations
  • 17 Disyn Algorithms: 7 original algorithms + 10 extended variants with param3 control for enhanced spectral manipulation
  • Program 6 Re-Enabled: Full Hybrid mode now working (race condition fixed in December 2025)
  • Backwards Compatibility: All existing code works without modification; param3 is optional with default value
  • Multi-Platform Builds: Optimized binaries for Ubuntu desktop (x86_64) and Raspberry Pi (aarch64)
  • Complete Documentation: Algorithm reference, MIDI spec, program guide, signal flow diagrams

Program Categories:

  • Programs 0-7: Original physical modeling presets (Classic Reed, Pluck, Brass, etc.)
  • Programs 8-17: Distortion synthesis explorations (ModFM Formant, DSF Bells, PAF Vowels, Hybrid chains)
  • Programs 18-28: Experimental combinations (Morphing engines, feedback networks, spectral animators)

Each program includes optimized 9-slider mappings and level safety verification. See docs/PROGRAM_CHANGE.md for complete program guide.

Technical Improvements:

  • Dual DC blocking (R=0.999) prevents feedback latching
  • Calibrated signal levels (0.81 peak, optimized for S16_LE)
  • TPDF dithering for clean 16-bit output
  • Test suite: engine-smoke, envelope-test, disyn-levels (all passing)

2025-12-11: S16_LE Audio Quality Improvements

  • Increased signal levels for better bit depth usage (56% → 81% of full scale)
    • Global pad: 0.7 → 0.85
    • Master gain: 0.8 → 0.95
    • Peak output: 0.28 → 0.81 (uses ~15.4 effective bits instead of ~13 bits)
  • Added TPDF dithering for S16_LE conversion (±1 LSB triangular PDF)
    • Eliminates quantization distortion on low-level signals
    • Improves perceived SNR by ~3-4 dB
  • Fixed truncation bias using proper round-to-nearest conversion
  • Result: Much cleaner quiet passages, reverb tails, and overall sound quality

2025-12-05: Signal Level & DC Blocking Fixes

  • Removed aggressive output DC blocker (was killing envelope attack)
  • Optimized to dual DC blocking (source + feedback loop, R=0.999)
  • Increased default noise level (0.02 → 0.15) for formant excitation
  • Reduced formant makeup gain (3.0× → 2.0×) to prevent Disyn clipping
  • Boosted master gain (0.35 → 0.5) and global pad (0.5 → 0.7)
  • Increased Disyn level (0.2 → 0.8) for better audibility
  • Added CC name printing on first occurrence for each controller
  • All 3 test suites passing (engine-smoke, envelope-test, disyn-levels)

2025-12-04: Initial Implementation

  • Merged Disyn + Chatterbox + PM-Synth architectures
  • Implemented triple DC blocking (later reduced to dual)
  • Created test suite for signal verification
  • Calibrated signal levels to prevent clipping

License

Part of the Flues project. See main repository for license information.

Related Projects

  • experiments/disyn: JavaScript distortion synthesizer
  • experiments/chatterbox: JavaScript formant speech synthesizer
  • experiments/pm-synth: JavaScript physical modeling synthesizer
  • lv2/disyn: Disyn LV2 plugin
  • lv2/chatterbox: Chatterbox LV2 plugin
  • lv2/floozy: Hybrid Disyn+PM LV2 plugin
  • gtk-synth: GTK4 desktop PM-Synth

Credits

  • Disyn: Distortion synthesis algorithms
  • Chatterbox: Formant synthesis and vocal modes
  • PM-Synth: Physical modeling framework
  • Built with: Meson, ALSA, C11/C++17