monokeys browser representative samples — source and license credits # Recorded keyboard audio Author: Thomas Lai Project: https://github.com/tplai/kbsim Pinned revision: ba103f3b0afa9dab80447aa2e7e2ed80b6bd80e4 Upstream path: src/assets/audio/{cream,boxnavy,topre,holypanda}/ License: MIT, as declared by the upstream repository. Full notice is preserved in LICENSE-kbsim.txt. 48 original MP3 files, 114,436 bytes total. No AI-generated audio is used. Originals are bundled unchanged; runtime decode to 48kHz mono, silence trimming, boundary fades, and energy-based gain alignment are applied in memory. The player uses each physical key row's press recording and separate Space, Enter, Backspace and release recordings. Subtle playback gain variation is decorative, not force sensing. Artwork is a generated material metaphor, not a photograph of the recorded keyboard. The upstream recordings describe the named switches/keyboards and do not reproduce every physical configuration or exact sound of the user's device. manifest.json records the original path, byte size, upstream Git object hash and SHA-256 of each file. scripts/fetch-recordings.py reproduces the download with TLS verification and validates Git object hashes before saving. This build does not download audio at runtime. Kitten calls: lolamadeus, "Kitten Meowing.wav", Freesound #196251. Source description: small kitten, 6–7 weeks old. Real field recording; not a human imitation. https://freesound.org/people/lolamadeus/sounds/196251/ License: CC0 1.0 (as explicitly declared on the author page, checked 2026-10-29). https://creativecommons.org/publicdomain/zero/1.0/ Source used: the public HQ MP3 preview linked on the author page, not the login-only original WAV. https://cdn.freesound.org/previews/196/196251_544580-hq.mp3 Four separate complete calls from the same kitten, 660–780ms, mono 48kHz / PCM16. Only room-rumble filtering, gentle high-frequency softening, gain matching and edge fades. No pitch shifting, artificial purr, syllable looping, added music or synthetic vocal. Original source, source-page snapshot and rejected/previous candidates are kept in Design/Audio/Kittens. scripts/prepare-kittens.py reproduces the clips offline after afconvert decoding. manifest.json records offsets, gains and SHA-256 for each shipped WAV. # Mud Steps — edited field recording - Source: **Steps in the Mud**, sound 0495, Joseph Sardin / BigSoundBank. - Source page: https://bigsoundbank.com/steps-in-the-mud-s0495.html - Original WAV: https://bigsoundbank.com/UPLOAD/bwf-en/0495.wav - License: **CC0 1.0**, as explicitly stated by the source page and https://bigsoundbank.com/licenses.html (checked 2026-10-29). Commercial use, editing and redistribution are permitted; attribution is optional and retained here. - CC0 legal text: https://creativecommons.org/publicdomain/zero/1.0/legalcode - No AI-generated audio is used for this pack. This is a derivative of a real human walking in liquid mud, not an unprocessed recording of tiny animal feet. The original WAV and source/license HTML snapshots are retained in `Design/Audio/Mud/` in the project. Its SHA-256 is `1fc9a1686e1441cb17c317943208aba4fa2e084f6281b9759d2da64e6fe6de25`. `scripts/prepare-mud.py` reproduces all 48 bundled clips offline using numpy: six distinct steps, four key families, and separate impact/after-motion excerpts. Processing removes rumble (110 Hz high-pass), softens treble (4.8–6.2 kHz low-pass), rounds sharp transients, aligns levels within peak/gain limits, and adds 2 ms attack / 25–45 ms end fades. Down clips are 125–225 ms; release clips are 65 ms and at most 22% of their matching down clip's RMS. No synthesized thump, bubble or music is mixed in. Only the edited mono 48 kHz / 16-bit WAV files and provenance ship in the app. No runtime network request or model download is required. `manifest.json` records source offsets, gains, durations and output hashes. The project review file `Design/Audio/Mud/mud-typing-preview.wav` plays isolated ordinary keys, Space/Enter/Delete, then dense typing; it is a dry montage before the app's output EQ, spatial pan and volume controls. Listening preference still requires human review. # Recorded material hybrids CC0-1.0, retrieved 2026-10-29. - https://bigsoundbank.com/drops-of-water-1-s1384.html — Joseph Sardin - https://bigsoundbank.com/pencil-2-s3233.html — Joseph Sardin & Axeline T. - https://bigsoundbank.com/ball-paper-1-s1220.html — Joseph Sardin License: https://bigsoundbank.com/licenses.html 144 mono PCM16 / 48kHz clips. Six source regions per material; four key-family treatments and independent release tails. Paper source is crumpled paper, edited to a short fold-like gesture. These are processed hybrids, not unaltered recordings. Original bytes, hashes, edit offsets and output hashes are in manifest.json. Originals and page/license snapshots are retained under Design/Audio/Layered/Originals. No runtime network. # Typewriter · sources and edits Joseph Sardin / BigSoundBank, Hermes Precisa 305 recordings, CC0-1.0. - Key: https://bigsoundbank.com/typewriter-key-s2842.html - Space: https://bigsoundbank.com/machine-a-ecrire-espace-s2843.html - Bell: https://bigsoundbank.com/machine-ecrire-cloche-s2844.html - License: https://bigsoundbank.com/licenses.html Downloaded 2026-10-29, 659,836 bytes total. Original hashes and per-file hashes in manifest.json. Original recordings are retained in Design/Audio/Expansion, not required at runtime. 48 mono 48kHz PCM16 edits. Six variants use subtle rate/gain changes, not six recorded takes. Ordinary key return is cut from the mechanical release; Enter return is a quiet bell tail. Bell on Enter is shortened and attenuated. No recorded text or user input is included. Puppy play bark recording by cribbler, Dog play.WAV, Freesound #396307. https://freesound.org/people/cribbler/sounds/396307/ Public HQ MP3 preview: https://cdn.freesound.org/previews/396/396307_3728489-hq.mp3 License: CC0 1.0, declared on the author page, checked 2026-10-29. https://creativecommons.org/publicdomain/zero/1.0/ Two separate short barks from a puppy playing in a backyard. No breed is specified by the author; these are not claimed as French Bulldog recordings. Edits: mono 48 kHz, remove room rumble, soften high frequencies, align gain, short edge fades. No pitch shift, voice synthesis or imitation. Source/page and offsets retained in Design/Puppy-v01 and manifest.json.