Skip to content

Editorial Standard — sound (ambient soundscape and noise colour)

  • Directus reference: https://cmsdocs.naluma.space/sounds
  • Manifest entry: naluma-directus/authoring-docs/reference-manifest.json → sound_card (domain coach-content) -- the coach-facing card that references a sound. The sound catalog itself (content.sounds, content.sound_albums) is a separate Directus collection. The sound_card schema (naluma-directus/schemas/coach-content/sound_card.schema.json) pins the reference shape: ref_kind (track by UUID / album by slug) and ref.
  • Sounds catalog: content.sounds rows in Directus (title, emoji, category, description, duration, audio asset, license fields). The catalog shape is owned by naluma-app migrations; editorial governance here applies to the copy fields that catalog editors author.

Purpose

A sound in the Naluma library is a therapeutic soundscape or noise-colour track that the user deploys for enrichment: to reduce the contrast between tinnitus and the acoustic environment, to support sleep onset, to facilitate concentration, or to create a habituation-friendly background. Sounds are not passive audio wallpaper -- they serve a specific therapeutic function determined by their category and clinical grounding.

The library is organised into five categories: Water (rain, ocean, river -- the most popular tinnitus masking category), Nature (forest, birdsong, fire -- relaxation and attention diversion), Noise Colours (white, pink, brown, blue, violet -- synthesised, omnidirectional), Urban Calm (coffee shop, fan, train -- concentration and ASMR-adjacent), and Tone Therapy (singing bowls, gentle chimes -- sleep and relaxation). Spatial audio is the default for all categories except Noise Colours (which are omnidirectional by nature).

Sound copy governs the title, emoji, short description, and any coach-surfaced card copy. The goal is brief, specific, and honest about what the sound is for.

Voice register

Default register is habituation/still. Sound descriptions are the quietest copy in the app -- the user is about to settle into an audio experience and the description should not over-excite or over-explain. Still, warm, concrete. One sentence about what the sound is; one optional sentence about what it is for.

For spike-adjacent sounds (crisis breathing background, post-spike sequence sounds): the register shifts slightly toward acute -- more directive, less atmospheric. "Steady broadband pink noise for immediate spike support" not "a gentle pink haze to help you relax."

For sleep-category sounds: quieter still. The description should almost be whispering in register -- understated, slow in tempo. "Slow rain on a window. Fifteen minutes, loopable." is correct. "A luxurious rainfall soundscape to help you achieve the perfect night's sleep" is not.

Emoji selection (if used in the title or UI) should be literal and single: one emoji that represents the sound source (raindrop for rain, leaf for forest, wave for ocean). No emoji stacking.

Evidence / IP

Grounded in docs/sound-library-research.md (sound types, clinical evidence, IP/licensing) and docs/session-content-evidence-base.md (vault notes #591 and #375). Key clinical and IP sources:

  • Notched noise/music: strongest clinical evidence of any sound therapy category -- 14 RCTs, n=793, THI improvement -8.6 pts at 3 months and -24.6 pts at 6 months (2025 meta-analysis via PMC8832119). Not in the MVP library (requires tinnitus frequency assessment at onboarding) -- flagged for v1.1.
  • Broadband noise (white/pink/brown): standard of care. Moderate evidence for loudness reduction and habituation support. Synthesised with numpy/scipy -- no licensing required.
  • Violet noise: high-frequency emphasis (power increases 6dB/octave). A differentiator versus most consumer apps; relevant for high-frequency tinnitus masking. Synthesised -- no licensing.
  • Nature sounds: indirect evidence via parasympathetic activation, stress reduction, and indirect tinnitus benefit. ATA recommends dynamic, spatially varied sounds for cortical retraining (ATA Sound Therapy page). Spatial audio is Naluma's differentiator here, matching Kalmeda's marketing advantage.
  • Fractal tones (Widex): weak evidence -- three small studies, none isolating a fractal-tone benefit: open-label pilot n=6 (PMC6197965); Tyler et al. 2017 n=20 (Zen stacked last of 4 treatments -- confounded); Deshpande & Bhatt 2021/22 n=20 RCT (fractal held constant in both arms, tested VR add-on). The foundational Widex Zen patent family (US6816599B2 + DE/CA/EP/JP/AU) is now expired -- so a non-Widex generative engine is buildable, subject to a freedom-to-operate review for any live continuation (research: docs/research/fractal-generative-sound-therapy.md). Still do not claim fractal-tone benefits for any non-fractal sound; position generative sounds as preference/relaxation/anti-monotony only.
  • White noise overuse risk: vault note #591 "cobra effect" -- excessive broadband noise exposure can habituate away and potentially worsen tinnitus. Sound descriptions must not recommend always-on use. "Use as a background while you settle in" not "play all night."

IP and sourcing rules for audio assets (per docs/sound-library-research.md): - Primary CC0 source: Freesound.org with license:cc0 filter only. Mixed licences per-sound -- CC0 is the only safe filter for commercial app embedding without attribution requirements. - Also safe: Free To Use Sounds (proprietary royalty-free, explicitly permits app development), Pixabay Audio, OpenGameArt CC0 packs. - Synthesised noise colours: no licence needed. - Do NOT use: BBC Sound Effects Rewind (RemArc -- non-commercial only), Zapsplat standard tier ("primary value" clause), Meta AudioCraft/AudioGen (CC-BY-NC). - The license field in content.sounds must be populated for every track. CC0 tracks require no attribution in-app but the field should still record the source URL for audit purposes.

Length / reading level

  • Title: 2-5 words. Plain-language. Describes the source, not the experience. "Light rain", "Brown noise", "Forest dawn", "Coffee shop background."
  • Emoji (if used): one emoji, literal. Placed before or after the title at the discretion of the UI style guide.
  • Short description (catalog view): 1-2 sentences maximum. States what the sound is and optionally what it is clinically suited for. Total: 15-35 words.
  • Album description: one sentence, ≤120 characters (hard-capped by validate.py).
  • Coach-surfaced sound card: reference-resolved from the catalog -- no additional copy authored in the card itself. The sound_card schema contains only ref_kind and ref. Title, image, and duration are resolved at compose time from content.sounds.
  • Reading level: Grade 6 or below. Sound descriptions are read at a glance during browsing.

Editorial-required elements

  1. Title names the source, not the experience. "Ocean waves" not "Serene coastal soundscape." "Pink noise" not "Calming pink haze." The user is choosing a sound; the title should tell them what they are choosing.

  2. Description is honest about the clinical role. If a sound has evidence for a specific use (pink noise for speech-in-noise improvement, vault #375; violet noise for high-frequency masking), say so. If it is primarily preference-based (nature sounds for parasympathetic activation), do not overclaim. Do not attribute fractal-tone benefits to sounds that are not fractal tones.

  3. No always-on recommendations. Sound descriptions must not recommend continuous or overnight use of broadband noise. The cobra-effect risk (vault note #591) is real. "Use as a background" or "for settling in" is correct. "Play overnight for best results" is not.

  4. License field must be accurate. Every content.sounds row must have the license field populated before publication. The source URL and license type (CC0, Free To Use Sounds, synthesised) are required. A sound without a license record is not ready to publish.

  5. Spatial audio flag must match reality. Noise colours are omnidirectional (no spatial positioning). Nature, Water, Urban Calm, and Tone Therapy sounds have spatial audio by default. The spatial flag in the catalog must accurately reflect the delivered audio file -- a mono file marked as spatial is a quality error.

  6. Obeys ai-patterns-en.md. Sound titles and descriptions must obey ai-patterns-en.md. Sound descriptions are the most likely place for meditation-app breathiness to appear ("allow the gentle rain to wash away your worries"). That fails on multiple counts.

Examples

Good -- nature sound description:

Title: Forest dawn

Description: Layered morning birdsong with light wind through trees. Spatially positioned -- the birds move across the stereo field. Suited for morning listening and attention diversion.

Why this works: plain title, specific description (layered, spatially positioned, birds move -- distinguishing Naluma's spatial audio), and a direct statement of the clinical use case (morning listening, attention diversion). 30 words. No breathiness.

Bad -- sound description:

❌ Title: Healing Rainfall Journey

❌ Description: Allow the gentle, healing sounds of rain to wash away your tinnitus distress and find your inner peace on this calming journey to tranquillity and restoration.

Why it fails: "Healing" in the title is a medical hype term (banned under ai-patterns-en.md). "Wash away your tinnitus distress", "find your inner peace", and "journey to tranquillity" are wellness clichés and/or banned constructions. The description contains no clinical information, no honest description of what the sound is, and no concrete statement of what it is suited for. It also implies the sound can reduce tinnitus distress directly -- an overclaim not supported by the evidence base for nature sounds.