# Third-party notices

## Midiophilia visual lab

The Midiophilia page is inspired by the MIDI-controlled color and brightness
concept in [Nicholas C. Stanley’s Midiphoria](https://github.com/NicholasCStanley/midiphoria),
reviewed at commit `93b5c4f031933c3372c20b428e3c4160b9485c73` on October 7, 2026.
That repository contains no license file or declared package license at this
revision. No upstream source, artwork, dependencies, or recordings are bundled.
Morphazoid's JavaScript model, Canvas renderer, and interface are independent
implementations. Note trails, source/channel isolation, sustain handling and
disconnect recovery are browser additions. The catalogue icon is a procedural
capture of Morphazoid's own note-trail renderer. This page does not claim Python,
GPU-compute, recording or video-export compatibility.

Browser MIDI-file playback uses SpessaSynth Lib 4.3.14 with Core 4.3.22 and
stb-vorbis 0.0.6 (Apache-2.0). Pinned bundle provenance and the single lifecycle
patch are in `src/instruments/midiphoria/vendor/README.md`; the license is beside
the bundle. AudioWorklet synthesis is independent of the Canvas renderer.

TimGM6mb is GPL-2.0-only; the unmodified editable SF2 and complete copyright
and license records are bundled under `assets/midiphoria/soundfont/`.
Selected complete web MIDI arrangements are free source-hosted downloads.
No public-domain or
Creative Commons status is claimed where the source specifies no license.
Exact download URLs, hashes, embedded credits and source records are in
`assets/midiphoria/collection.json` and `SOURCE.md`.
Creator tracks by Umplix, Alex McCulloch and Komiku retain their CC0 credits in
`assets/midiphoria/GENRE-SOURCES.md`. Original demos and experimental studies
retain reproducible generators and credits in `genre-demos/` and
`experimental/`; they are not transcriptions of existing artists’ songs.

## MIDI Mixer

Midiophilia's two-deck MIDI Mixer view reuses its unchanged MIDI collection
and TimGM6mb SoundFont, with the source and licensing records above. Its owned
audio-clock mixer uses a pinned SpessaSynth Core 4.3.22 bundle with stb-vorbis
0.0.6 under Apache-2.0; see the [bundle provenance and license](../src/instruments/midi-mixer/vendor/README.md).
The mixer, inspector, rectangular renderer and scheduling code are original
Morphazoid implementations.

## I/O setup synthesized voice

The original `assets/audio/midi-received.wav` announcement was generated with
the separate eSpeak NG development tool and processed into a retro-style
sample. It does not contain Apple or Macintosh system recordings. This
announcement plays a rendered sample without loading eSpeak; the separate
browser speech engines are credited below. See
`assets/audio/CREDITS.md` and `scripts/generate-midi-received.py` for provenance
and regeneration.

## Roach Synth cockroach scan

Roach Synth adapts **Cockroach (3D photogrammetry scan)** by
Kathrin&Christian (extracrispi), licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).
[Original model](https://sketchfab.com/3d-models/cockroach-3d-photogrammetry-scan-cff6ea065e2e4029be5ba20f03937b63),
retrieved from the Allen Institute for AI's public Objaverse collection.
Morphazoid preserves the on-disk scan while adding approximate rigid joint
pivots. The mobile derivative pre-splits the textured wing covers, compresses
geometry streams and textures, and retains the source vertex positions and
joint transforms. Runtime adaptations add two authored
veined hindwing fans with independent hinges. The preview image and catalogue
icon are renders of this adapted model. Full attribution, exact hashes, and
modification details are in `assets/roach-synth/SOURCE.LICENSE.txt` and
`assets/roach-synth/rig-manifest.json`.

Roach Synth's word voice also reuses the existing CMU Flite/KAL16 diphone atlas
and CMU pronunciation dictionary credited below.

Roach Synth also includes edited excerpts from **Gromphadorhina_portentosa.aif**
by **nicotep**, released under [CC0 1.0](https://creativecommons.org/publicdomain/zero/1.0/).
The [creator's recording](https://freesound.org/people/nicotep/sounds/547897/)
documents cockroach movements in a vivarium. The bundled mono WAV excerpts are
derived from its public high-quality MP3 preview, with filtering, gain adjustment
and short fades. Exact source offsets, processing, checksums and attribution are
in `assets/roach-synth/audio/CREDITS.md` and its adjacent `manifest.json`.
Other animal recordings in `docs/ROACH_SYNTH_RESEARCH.md` remain linked references.

## meshoptimizer

Roach Synth's compressed model uses the unmodified meshoptimizer 0.25 decoder
by Arseny Kapoulkine, under the MIT license. The decoder and its embedded WASM
are in `vendor/meshoptimizer/`; source URLs and checksums are in its `README.md`,
and the full license is in `LICENSE.txt`. The matching encoder under
`scripts/vendor/meshoptimizer/` is used only to prepare assets and is excluded
from published site bundles.

## MakeHuman teeth_base

Dentaphone's optional WebGL jaw adapts the MakeHuman system `teeth_base`
asset. The source contains 32 independent tooth shells and a connected mouth
shell:

- https://static.makehumancommunity.org/assets/assetpacks/makehuman_system_assets.html
- https://static.makehumancommunity.org/about/license.html

MakeHuman released its system assets under CC0 1.0 Universal. Morphazoid
separates and names all 32 teeth, splits the mouth shell into independently
hinged upper and lower sections, recenters and scales the geometry, and converts
it to GLB. Source details and the exact OBJ hash are in
`assets/models/dentaphone-chomper.LICENSE.txt`.

The reproducible converter lives only in the Morphazoid source repository at
`scripts/build-dentaphone-chomper.py`; it is a build and provenance tool, not a
runtime dependency, and is intentionally omitted from published static-site
bundles. References to that path in the model's license notice refer to a
repository checkout.

## Gilbert generalized Hilbert curve

Paths adapts the recursive two-dimensional Gilbert reference algorithm:

- https://github.com/jakubcerveny/gilbert

Copyright (c) 2018 Jakub Červený

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:

1. Redistributions of source code must retain the above copyright notice,
   this list of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice,
   this list of conditions and the following disclaimer in the documentation
   and/or other materials provided with the distribution.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.

## Bioacoustic recordings

The built-in recording menus bundle twelve unchanged Ogg Vorbis or PCM WAV field
recordings from Wikimedia Commons. Full provenance, recording notes, source
links, license links, and file hashes are recorded in
`assets/bioacoustics/SOURCES.md`.

- Thrush nightingale, common blackbird, and chaffinch recordings: Oona
  Räisänen (Wikimedia Commons user Mysid), public-domain dedications.
## WebGPU Chiptune / Chiptune (sound) lineage

WebGPU Chiptune translates the active sound path of **Chiptune (sound)** by
srtuss (2015), Shadertoy shader ID `MljSRt`, from GLSL into WGSL and adds the
Morphazoid streaming runtime, controls, presets, visualization, and host state
adapter.

- Original shader: https://www.shadertoy.com/view/MljSRt
- Author credit in the supplied source: `srtuss, 2015`

The Shadertoy page is not loaded at runtime. The supplied shader text and
archived metadata did not state a reuse license, so this notice records
provenance without asserting license terms.

- House cricket recording: Wikimedia Commons user Morray, CC BY 3.0.
- Field cricket recording: Wikimedia Commons user Thatcher, CC BY-SA 3.0.
- European field cricket recording: Baudewijn Odé, CC BY-SA 4.0.
- Coyote group recording: Wikimedia Commons user Rybkovich, CC BY-SA 4.0.
- Frog chorus recording: Wikimedia Commons user Hughesdarren, CC BY-SA 4.0.
- Dolphin recording: Félix Blume, CC0 1.0.
- Humpback-whale recording: Wikimedia Commons user Spyrogumas, CC0 1.0.
- Killer-whale recording: U.S. National Park Service, public domain in the
  United States.
- Blue-whale recording: NOAA Pacific Marine Environmental Laboratory, public
  domain in the United States.

Morphazoid distributes the Creative Commons recordings unchanged, preserves
attribution and license links, and does not suggest endorsement by the
recordists or licensors.

No Indigenous or other human song recordings are redistributed. Links in the
Acoustic Manifold archive directory are discovery references, not licenses to
copy, train on, rearrange, or resynthesize their holdings. Record-level rights,
community authority, culturally sensitive access protocols, and any Traditional
Knowledge Labels remain applicable.

## Three.js

Dentaphone, Nightingale Manifold, and Acoustic Manifold vendor Three.js 0.185.1
for local, offline-safe WebGL rendering. Dentaphone also vendors the GLTFLoader
utilities needed by its anatomical jaw:

- https://threejs.org/
- https://github.com/mrdoob/three.js

Copyright © 2010-2026 three.js authors

Three.js is MIT-licensed. The complete license text is in
`vendor/three/LICENSE.txt`. Dentaphone's vendored loader utility imports were
changed only to point at the adjacent local Three.js ESM build.

## Fluid Music Open Drums / Hyperreal TR-808 and TR-909 samples

Sample Drums references version-pinned WAV files from the Fluid Music
open-drums npm packages at runtime:

- https://www.npmjs.com/package/%40fluid-music/tr-808
- https://www.npmjs.com/package/%40fluid-music/tr-909
- https://github.com/fluid-music/open-drums

The TR-808 package is ISC-licensed and its README says the original Hyperreal
samples were recorded by Michael Fischer with no licensing restrictions noted.

The TR-909 package README says the samples were recorded by Jason Baker / Rob
Roy Recordings and reproduces the original Hyperreal text allowing free copying
and distribution while prohibiting distribution of the samples for profit.
Because of that condition, Morphazoid does not vendor the TR-909 WAV files in
this repository; it references the pinned package URLs from the browser and
decodes them into memory during use.

## WebGPU 303 / Acid Synth lineage

WebGPU 303 adapts the local WebGPU Audio Acid Synth shader and streaming
architecture. That local demo credits its acid voice lineage to:

- sound - acid jam by srtuss on Shadertoy
- https://www.shadertoy.com/view/ldfSW2

Morphazoid keeps that attribution visible on the WebGPU 303 page and does not
load the Shadertoy page as a runtime dependency.

## srtuss WebGPU master synth and Sound archive

The sound-only srtuss instrument contains WGSL translations of ten original
Shadertoy Sound passes by srtuss:

- MljSRt — Chiptune (sound)
- XdSGz1 — noir et blanc (sound)
- Xd2GW3 — Industry II (sound)
- 4tsGD8 — Boulder Dash title (sound)
- ldlfRS — Shift (sound)
- lldGDM — Noise Bands (sound)
- ltKSRc — Gravity Shielding (sound)
- 4tdSDB — DnB (sound)
- MslBR4 — Cipher (sound)
- ldfSW2 — sound - acid jam

Original project pages are under:

- https://www.shadertoy.com/user/srtuss
- https://www.shadertoy.com/view/SHADER_ID

Nine exact Sound-pass source strings were recovered from the public
GabeRundlett/shadertoy-api-shaders JSON archive pinned at commit
f6d538adf936215ccf2d11ba9b4a6c79ccb448c5. Acid Jam was recovered from an
author-attributed SimpleShader copy and independently corroborated against an
audiojs copy whose executable GLSL matches after normalizing its wrapper-only
entry-point and helper rename:

- https://raw.githubusercontent.com/mike-seger/simpleshader/3d0593e14bbd585b3eb88d042a3bf251af6c4b63/shaders/shadertoy-gpu-audio/sound-acid-jam-sound.glsl
- Primary source SHA-256: 1f6c0e5cfb9577681ca79843907ab7f061427fc1a8e35b5d292d3ee8a01b456a
- https://raw.githubusercontent.com/audiojs/audio-shader/f8de3cd2ddf231198ae7d7aa367bae25d01ed453/test/sounds/acid.glsl
- Corroborating source SHA-256: f889afecc6270358227964d7fe6f2ac2ea714822ef1eefdf9751da4320cf4cd2

Each WGSL module records its original source length and SHA-256 digest, pinned
source URL, translation notes, and project URL.

The exact translations remain separately selectable. Morphazoid also derives
parameterized Master variants from XdSGz1, Xd2GW3, ldlfRS, 4tsGD8, lldGDM,
ltKSRc, 4tdSDB, and MslBR4. Those variants expose source-native component
levels and mapped performance controls; they retain the same attribution.
Chiptune (MljSRt) and Acid Jam (ldfSW2) remain exact-only on this page because
Morphazoid provides dedicated WebGPU instruments for those two systems.

Morphazoid treats these Shadertoy adaptations under the site's
Creative Commons Attribution-NonCommercial-ShareAlike 3.0 publication terms
and preserves the noncommercial/share-alike restriction:

- https://creativecommons.org/licenses/by-nc-sa/3.0/

The standalone Chiptune notice above separately records that its supplied code
and mirrored JSON do not embed a per-file license declaration; this archive
does not represent that absence as a more permissive license.

The texture sampled by Noise Bands, Gravity Shielding, and DnB is the exact
512 × 512 Shadertoy “Pebbles” channel asset, recovered from mrRay/VVISF-GL
at commit 7306b1abdd0b5536f0f230c0519453e79184e08b:

- https://raw.githubusercontent.com/mrRay/VVISF-GL/7306b1abdd0b5536f0f230c0519453e79184e08b/examples/Qt/ISFEditor/ISFEditor_app/ISFConverter/shadertoy_textures/ad56fba948dfba9ae698198c109e71f118a54d209c0ea50d77ea546abad89c57.png
- Vendored file SHA-256: 4d0c9886f97a362824c1e21d26ff40a16469564c4ad7e0ce73aab5b392541b04

Three further Sound projects (MdXXW2, Xd2XDm, and 4ddfWX) stay
marked source-pending because accessible copies are absent, incomplete, or
conflicting. A separate VEDA acid shader describes itself as heavily inspired
by ldfSW2; it is not used or represented as the 1:1 source.
The fourteen image-only srtuss projects are inventoried but are neither bundled
nor rendered.

## XYFlow comparison renderer

The optional `?graph=xyflow` renderer for Modular WebGPU Synth statically
bundles React Flow and its browser dependencies. It is loaded only for that
comparison route; the original Morphazoid graph renderer remains the default.

- React Flow / XYFlow: https://github.com/xyflow/xyflow (MIT)
- React and React DOM: https://github.com/facebook/react (MIT)
- Zustand and classcat (MIT)
- D3 selection, drag, zoom, transition, interpolation, dispatch, timer, and
  color modules (ISC), plus d3-ease (BSD-3-Clause)

Every production build generates the complete dependency license texts from
the installed, lockfile-pinned packages at
`assets/xyflow/THIRD_PARTY_LICENSES.txt`.

## Pink Trombone

The shared classic 44-section vocal-tract geometry used by Throatazoid, Pink
Trombonazoid, Spelling Synthesizer, and Hiccup Head, including its tongue-rest curve,
reflection coefficients, and two-step waveguide behavior, is adapted from Neil
Thapen's Pink Trombone. Gesticules' Choir also adapts its Liljencrants–Fant
glottal source through Morphazoid's Hiccup Head and Throatazoid implementations:

- https://dood.al/pinktrombone/
- https://github.com/IMAGINARY/pink-trombone

Copyright 2017 Neil Thapen

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

## CMU Flite / KAL16 diphone voice

The Spelling Synthesizer diphone audio sprite was generated from the
`cmu_us_kal16` diphone voice in CMU Flite 2.2, then silence-trimmed, faded,
level-normalized, and packed by Morphazoid. These processing and packaging
changes are not endorsed by Carnegie Mellon University.

The additional Spelling LPC coefficient atlas is also a modified derivative of
that same KAL16 recording. Morphazoid's original build script low-pass filters,
downsamples and performs ten-order LPC analysis with parameter quantization.
It contains no Texas Instruments ROM data, original Speak & Spell recordings,
or third-party emulator code. Carnegie Mellon and Texas Instruments do not
endorse this instrument.

- https://github.com/festvox/flite

Language Technologies Institute
Carnegie Mellon University
Copyright (c) 1999-2017
All Rights Reserved.

Permission is hereby granted, free of charge, to use and distribute this
software and its documentation without restriction, including without
limitation the rights to use, copy, modify, merge, publish, distribute,
sublicense, and/or sell copies of this work, and to permit persons to whom this
work is furnished to do so, subject to the following conditions:

1. The code must retain the above copyright notice, this list of conditions
   and the following disclaimer.
2. Any modifications must be clearly marked as such.
3. Original authors' names are not deleted.
4. The authors' names are not used to endorse or promote products derived from
   this software without specific prior written permission.

CARNEGIE MELLON UNIVERSITY AND THE CONTRIBUTORS TO THIS WORK DISCLAIM ALL
WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL CARNEGIE MELLON UNIVERSITY NOR
THE CONTRIBUTORS BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES
OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER
IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.

## OddVoices CC0 voice recordings

Vocalzoid includes compact audio sprites derived from the Air, Cicada, and
Quake voice recordings in OddVoices. Morphazoid extracts only eight labelled
English units per voice, trims long vowel bodies, removes DC offset, normalizes
level, adds short edge fades, and packs the units into browser-sized WAV files.
These packaging changes are not endorsed by the OddVoices authors. The
reproducible builder is `scripts/generate-vocalzoid-oddvoices.py`.

- https://gitlab.com/oddvoices/oddvoices/
- https://gitlab.com/oddvoices/oddvoices/-/tree/develop/voices
- Source revision: 33a248af8df88edf5166593bf36b7e24e7bc1f94

Public Domain.

OddVoices vocal data by Nathan Ho is marked CC0 1.0. To view a copy of this
mark, visit https://creativecommons.org/publicdomain/zero/1.0/

## CMU ARCTIC voice recordings

Vocalzoid includes five compact audio sprites made from the BDL, CLB, JMK,
KSP, and SLT voices in the CMU ARCTIC 0.95 releases. These files are modified
excerpts, not the original voices: Morphazoid selects eight automatically
aligned phone units per speaker, trims them to the “vocalzoid” demo inventory,
removes DC offset, normalizes level, adds edge fades, finds phase-matched vowel
loops, and packs the results into browser-sized mono WAV files. The original
authors do not endorse these modifications. The reproducible builder is
`scripts/generate-vocalzoid-arctic.py`.

- http://www.festvox.org/cmu_arctic/
- https://www.cs.cmu.edu/~awb/papers/ssw5/arctic.pdf
- Source releases: `cmu_us_{bdl,clb,jmk,ksp,slt}_arctic-0.95-release`

BDL, JMK, and SLT: Copyright (c) 2003 Carnegie Mellon University.
CLB and KSP: Copyright (c) 2005 Carnegie Mellon University.
All Rights Reserved.

Permission to use, copy, modify, and license the voices and their documentation
for any purpose is granted without fee, subject to the conditions and warranty
disclaimers reproduced in `vendor/cmu-arctic/COPYING` (2003 voices) and
`vendor/cmu-arctic/COPYING-2005` (2005 voices).

## PocketSphinx English pronunciation dictionary

Spelling Synthesizer uses the PocketSphinx English CMU pronunciation
dictionary to translate complete words into ARPABET phone sequences for local
readback:

- https://github.com/cmusphinx/pocketsphinx

Copyright (c) 1995-2014 Carnegie Mellon University. All rights reserved.
Copyright (c) 2014-2015 Alpha Cephei Inc.

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:

1. Redistributions of source code must retain the above copyright notice,
   this list of conditions and the following disclaimer.

2. Redistributions in binary form must reproduce the above copyright notice,
   this list of conditions and the following disclaimer in the documentation
   and/or other materials provided with the distribution.

This work was supported in part by funding from the Defense Advanced Research
Projects Agency and the National Science Foundation of the United States of
America, and the CMU Sphinx Speech Consortium.

THIS SOFTWARE IS PROVIDED BY CARNEGIE MELLON UNIVERSITY "AS IS" AND ANY
EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL CARNEGIE MELLON UNIVERSITY NOR ITS EMPLOYEES BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.

## Tactile

The Lattice, Spiral, and Escher instruments use Tactile's
isohedral tiling data and utilities:

- https://isohedral.ca/software/tactile/
- https://github.com/isohedral/tactile-js

BSD 3-Clause License

Copyright (c) 2018, Craig S. Kaplan
All rights reserved.

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:

* Redistributions of source code must retain the above copyright notice, this
  list of conditions and the following disclaimer.

* Redistributions in binary form must reproduce the above copyright notice,
  this list of conditions and the following disclaimer in the documentation
  and/or other materials provided with the distribution.

* Neither the name of the copyright holder nor the names of its
  contributors may be used to endorse or promote products derived from
  this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

## Signalsmith Stretch

L-system Delay's optional Silky renderer and Trackazoid's optional key lock use the Web Audio release of Signalsmith
Stretch:

- https://signalsmith-audio.co.uk/code/stretch/
- https://github.com/Signalsmith-Audio/signalsmith-stretch

MIT License

Copyright (c) 2022 Geraint Luff / Signalsmith Audio Ltd.

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

## Puggler acoustic drums and audience voices

`assets/puggler/` bundles eight CC0 1.0 Universal recordings: five Karoryfer
Samples acoustic drums, rhink’s “Oi.wav,” NeoSpica’s “Booing Crowd,” and jayfrosting’s “Woo 2.wav.”
See [recording sources and processing](../assets/puggler/CREDITS.md) and the
[included CC0 text](../assets/puggler/CC0-1.0.txt). Guitar and bass are original repository synthesis. No episode or commercial music
audio is bundled.

## Spider Synth scanned specimen

`assets/spider-synth/spider-mobile.glb`, its specimen image and catalogue icon
are derived from **Argiope bruennichi**, a CC0 scan by **Yuichi Kano / ffish.asia
and floraZia.com**, published on Sketchfab on 2025-03-18. Morphazoid removes the
calibration cube and adds authored skin weights and 38 joint controls. Source:
https://sketchfab.com/3d-models/cc0-orb-weaver-spider-a-bruennichi-bb646be39dad44948a403366b0ebc977

See [exact provenance and license](../assets/spider-synth/SOURCE.LICENSE.txt).
Voice uses the existing KAL16 diphone atlas and CMU pronunciation resources;
Spider Synth does not bundle new animal audio recordings.

## Gesticules — Elena FF

[“Rigged hand” by Elena FF](https://sketchfab.com/3d-models/rigged-hand-eae97cc2a742413cb5338ab942b12c1e), [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/). `assets/gesticulating-hand/hand.glb` retains the original mesh, textures, weighted rig and Open/Close animation, losslessly repacked from an attributed public distribution. `src/instruments/gesticulating-hand/hand-source-motion-data.js` contains derived quaternion animation data under the same license. The catalogue image is a rendered view of this model and also retains CC BY-SA 4.0. Original attribution, pinned source, changes and rebuild steps are in `assets/gesticulating-hand/README.md`. The separate runtime sampler and instrument implementation are MIT.

## eSpeak NG browser speech

Spelling Synthesizer and Voicesaurus include a WebAssembly build of eSpeak NG
1.52.0.1 with Echogarden's `eSpeakNGWorker` wrapper, from pinned fork commit
`9a550bef455f03b459f51796e3482833aab7fbc0`. GPL-3.0-or-later applies to the engine;
its Apache, BSD and Unicode component notices are also retained.

The engine generates English phoneme sounds locally in a browser worker after
explicit Audio enable. Morphazoid builds a separate WASM binary and repacks
English data; the synthesis code is unchanged. The Klatt mode is eSpeak's
implementation, not DECtalk. [Source, changes and build instructions](../vendor/espeak-ng/README.md),
[build manifest](../vendor/espeak-ng/build.json), and [GPL license](../vendor/espeak-ng/COPYING)
accompany the engine. `scripts/build-espeak-wasm.py` rebuilds pinned source and
reproduces the checked-in JavaScript, WebAssembly and voice-data hashes.

## Flite browser speech

The speech pages include Flite's WASI build from `@echogarden/flite-wasi` 0.1.1,
which reports Flite 2.3-current (March 2022). Its embedded KAL, KAL16, AWB, RMS,
SLT and limited-domain AWB-time voice definitions retain the terms in
[Flite COPYING](../vendor/flite/COPYING) and [voice notices](../vendor/flite/VOICE_NOTICES.md).
The pages expose the general SLT/AWB/RMS Clustergen voices; other embedded models
are not presented as additional live methods.

Morphazoid removes only WebAssembly custom debug/name/producers sections;
executable and voice-data sections are unchanged. The small in-memory WASI
adapter and phoneme-atlas integration are original Morphazoid code.
[Provenance and reproduction](../vendor/flite/README.md) identify the pinned package
and source links. `scripts/strip-flite.py` reproduces the stripped binary.

The repository's MIT license does not replace these third-party terms.

## GnuspeechSA / Trillium browser articulation

Voicesaurus and the shared speech reader include GnuspeechSA 0.1.9, compiled to
WASM from `mym-br/gnuspeech_sa` revision
`f62e8b88eeeb3fa7e51ac138d561c3061a4a415f`. The engine and English linguistic/
articulatory data are GPL-3.0-or-later, with Trillium and original authors'
notices preserved. RapidXML's Boost/MIT terms are retained separately.

[Source, changes and reproduction](../vendor/gnuspeech/README.md),
[GPL](../vendor/gnuspeech/COPYING.txt), [data grant](../vendor/gnuspeech/DATA_LICENSE.txt),
and [RapidXML license](../vendor/gnuspeech/RAPIDXML_LICENSE.txt) accompany the build.
`scripts/build-gnuspeech-wasm.py` and `scripts/gnuspeech-wasm-adapter.cpp` rebuild
all three runtime files from pinned source. The five physical voice profiles
share one articulatory method. Their atlas playback is distinct from native
whole-sentence synthesis.

## Vizsn browser voice

The Vizsn WASM port preserves Ville-Matias Heikkilä's formant/resonator engine
and Jari Komppa's SoLoud integration, both explicitly WTFPL, from SoLoud revision
`e82fd32c1f62183922f08c14c814a02b58db1873`. [License](../vendor/vizsn/COPYING) and
[source/changes](../vendor/vizsn/README.md) are included. Modified DSP, the host shim
and a pinned-source rebuild script are in `scripts/vendor/vizsn/`.

The reader's 43-gesture mapping is an original approximation to Vizsn's smaller
native phone vocabulary. It is not a recovered English voice database.

## MEA8000 parameter-frame voice

The MEA8000 WASM generator extracts Antoine Miné's BSD-3-Clause DSP and tables
from MAME revision `c8588c15c78215a0ce36ac573aa5e8da03185e4c`.
[Copyright/license](../vendor/mea8000/COPYING) and
[extraction, modifications and rebuild instructions](../vendor/mea8000/README.md)
are included; build/bridge sources live in `scripts/vendor/mea8000/`.
The offline frame host excludes MAME's device bus/timers and is not cycle-accurate
chip emulation. English-like gesture parameters are original Morphazoid examples;
no historical phoneme ROM or speech recording is bundled.

## SVOX Pico browser speech

Pico uses the unmodified runtime from `@echogarden/svoxpico-wasm@0.2.0`, package
source revision `4b1d64dc4c69cec088141d8653c526f09b76ea62`, plus English resources
from `naggety/picotts` revision `21089d223e177ba3cb7e385db8613a093dff74b5`.
Core/package Apache-2.0 terms and SVOX/Android resource notices are retained in
[vendor/pico](../vendor/pico/README.md). `DEBIAN_COPYRIGHT` documents that the model
binaries are redistributable but lack preferred source; they are not presented
as fully reproducible open training data. The independent public-C-API bridge
and phoneme-atlas integration are original Morphazoid code.

## Csound FOF and VOSIM voice methods

Voicesaurus plays the original Csound FOF/VOSIM opcodes through the LGPL Csound
WASM library from `@csound/wasm-bin@6.18.7`. The unmodified binary reports Csound
6.18.1. Its codec/runtime dependencies remain embedded: libsndfile, FLAC, Ogg,
Vorbis, LAME, mpg123, wasi-libc and LLVM runtime components. Their notices and
source links accompany the [Csound provenance](../vendor/csound/README.md).

The local WASI host and phonetic demonstration orchestras are original code.
They do not bundle the original CHANT program, its voice database, or Csound's
example plugins. `scripts/vendor-csound-wasm.py` verifies and restores the exact
published binary; the pinned upstream synthesis/build sources remain linked
for modification and recompilation.

## Compiler runtime notices for browser speech

The newly compiled GnuspeechSA, Vizsn and MEA8000 engines include Emscripten
runtime glue and C/C++ library code. Relevant Emscripten MIT/NCSA, musl MIT, and
LLVM Apache-with-exceptions notices are preserved under
[vendor/wasm-runtime-notices](../vendor/wasm-runtime-notices/README.md).
These component licenses do not replace the separate speech-engine licenses.

## HTS English HMM voice

The HTS Working Group’s hts_engine API1.10 and Flite+hts_engine1.07 are
modified-BSD software; their English Flite frontend has separate CMU notices.
The official HTS Voice CMU ARCTIC SLT1.06 is CC-BY-3.0, trained from Carnegie
Mellon University’s CMU ARCTIC database. The voice model is unmodified.
Morphazoid extracts, balances and loops generated phoneme audio for playback.
See [vendor/hts/README.md](../vendor/hts/README.md), the four preserved license
files beside it, and `scripts/vendor/hts/` for pinned source/rebuild instructions.


## Sinsy native Japanese singing

Voicesaurus includes Sinsy 0.92 (2015-12-25), the original Japanese MusicXML
frontend and hts_engine API 1.10, under their Modified BSD notices. The
NIT SONG070 F001 0.90 trained singing voice (2013-12-25) is credited to the
Sinsy Working Group / Nagoya Institute of Technology, Department of Computer
Science, under [CC BY 3.0](https://creativecommons.org/licenses/by/3.0/).
It is synthesized locally from the editable score; it is not a bundled song.
Notices and exact source/model hashes are in `vendor/sinsy/`; rebuild code and
native API forwarding changes are in `scripts/vendor/sinsy/`.
Official source: <https://sinsy.sourceforge.net/>.

## Midiophilia black MIDI and full orchestra collections

BLEEDING EDGE MONOTONE's complete Black-MIDI-Archive (35 MIDI files, 19 works)
and its 16 creator AUDIO editions are CC0 1.0. The nine rin-w geometric MIDI
patterns use the repository's public-use/modification contribution terms.
Original bytes, credit requests, exact license/README and hashes are retained;
see [black MIDI sources](../assets/midiphoria/BLACK-MIDI-SOURCES.md).

The complete Eroica score MIDIs prepared by Jay Anderson are CC BY-SA 3.0.
J.F. Lucarelli's Dvořák 7 edition is expressly Public Domain at Mutopia.
TheOuterLinux's Night on Bald Mountain conversion is CC0 1.0. Source links and
license texts accompany [orchestral sources](../assets/midiphoria/ORCHESTRAL-SOURCES.md).
Copyrighted private-use Carlos and Mahler downloads are excluded from runtime assets.

## Sound Fonts collection

The Kalimba, Ocarina, Jaw Harp and Lately Bass banks are unmodified FreePats
releases under CC0 1.0. Their original README files and the full CC0 legal text
accompany each SF2 in `assets/sound-fonts/`. Kalimba and Ocarina credit Xavimart
(Javier), Gonzalo and Roberto; Jaw Harp credits Eugene Vlaskin, Michail Chayka,
N. K. Chayka and Roberto; Lately Bass credits Yingchun Soul and FreePats.
Exact source/download URLs, versions and hashes are recorded in
[`assets/sound-fonts/provenance.json`](../assets/sound-fonts/provenance.json).

The Sound Fonts page also exposes the existing TimGM6mb bank, copyright Tim
Brechbill and David Bolton, under GPL version 2. Its source record, copyright
notice and complete GPL text remain in `assets/midiphoria/soundfont/`.
These asset licenses are separate from Morphazoid's application license.

GeneralUser GS 2.0.3 by S. Christian Collins is distributed unmodified under
its GeneralUser GS License v2.0, which permits software integration and music
creation and records uncertainty about some historical sample origins. The
original license, release changelog and source record accompany the bank in
`assets/sound-fonts/generaluser-gs/`.

FluidR3Mono GM 2.312 retains the original MuseScore 2.1 SF3 and MIT license,
crediting Frank Wen, Michael Cowgill, Ethan Winer and Michael Schorsch. Its
accompanying license explicitly permits redistribution despite a stale
restrictive comment embedded in the upstream asset; both are preserved.
FluidR3 GM 3.1 by Frank Wen and contributors is distributed under its original
MIT license as a Vorbis-compressed SF3, with all 189 presets and 1,418 sample
mappings preserved. Its original SF2/archive hashes and reproducible conversion
script are retained in `assets/sound-fonts/fluidr3-gm/` and the shared provenance
record. SF3 compression is lossy. The original Fluid notices and contributors
remain beside both banks; the application license does not replace them.

## Rig Lab model downloads

Rig Lab bundles 11 Sketchfab 3D models under CC BY 4.0 by rayray, Msassasa,
adarshktk, Osian CG, Dirk.z, volvor, Cas_Models, Emma Leyers, SCANIMAT,
ayvlasov and David Hall. Attribution, original model links, license,
SHA-256 checksums, texture reductions, the new mosquito/lattice bones,
and runtime modifications are recorded beside the files in
[`assets/rig-lab/SOURCE.LICENSE.txt`](../assets/rig-lab/SOURCE.LICENSE.txt).
The original model meshes, textures and their derivatives are not covered by
Morphazoid's MIT application license. Foam and lattice render independently
generated editable geometry; the two credited GLBs remain reference downloads.