Fonora is an open-source research project exploring how writing systems and languages can be designed from first principles. It combines experimental software, a phonetic script, and a constructed language to test new ideas about communication.
Our hypothesis: Can a language designed from first principles be learned quickly enough that two people with no shared native language can achieve practical communication after only a short period of study?
Project goals
Push the boundaries of how language can be expressed and understood.
Create a system that is intuitive, accessible, and easy to learn.
Build a language with clarity, consistency, and ease of use at its core.
Make language intuitive by aligning sound and meaning.
Ensure the language works across cultures and linguistic backgrounds.
Empower the community with resources to create, test, and collaborate.
Fonora is an active open-source project and not a finished product.
Fonora combines a phonetic writing system, a constructed language, and software for building and testing both. Every symbol, sound mapping, root, and grammar rule is treated as a hypothesis rather than a conclusion.
Ideas are refined through experimentation and measurement. The language is described by five documents, kept in the Docs tab: the rulebook, three algorithm pages, and the architecture map.
Script
A phonetic writing system built from nine core symbols that represent where and how speech sounds are physically produced.
Language
An experimental language built on Fonora Script to test whether a small shared vocabulary can enable practical communication across native languages.
Tools
The software used to build, analyze, and experiment with both the script and the language.
Reference
Platform overview, grammar notes, and guides for contributors and operators.
Builder and QA tools are available to signed-in @fonora.org accounts.
Sign in with Google to access script debugging tools and language builder utilities. Learn exercises stay public without an account.
Sign in with GoogleFonora separates language from writing. Learn to communicate in Fonoran using roman spelling, or learn the Fonora script to read and write sounds from any language. Neither path depends on the other.
Speak
Practice listening, understanding, and responding in everyday conversations. No Fonora script required.
Read & Write
Read and write speech using Fonora symbols, and transliterate words from your own language. Optional if your goal is simply to speak Fonoran.
See a Fonora symbol and work out the sound it represents.
See a word in your language — type it in Fonora Script.
Read Fonora Script words and type them back in your language.
Recognize Fonoran words and pick the meaning that matches.
Type Fonoran spellings from a meaning prompt.
Listen to a Fonoran word and choose its meaning.
Five-minute rules: order, particles, want+go, then short drills.
Metrics
Community signups, learn progress, engagement, and product health.
Analysis
Check vocabulary coverage and consistency metrics.
Analysis
Track vocabulary growth over time.
Debug & QA
Evaluate IPA → Fonora encoding across languages.
Debug & QA
Validate that Fonora preserves pronunciation through the full encode/decode pipeline.
Demo
Hear Fonora Script read aloud across languages — a universal-script demo.
Community
Community members vote and propose alternates on the public Language dictionary — not in Tools.
Pick an item from the queue
Select an open proposal to review, or a playtest promotion to accept.
Loading analytics…
Loading lab progress…
The ~89 irreducible concepts that everything else is built from. They live in data/fonoran-approved-roots.json and are curated by hand using the campfire test.
Compounds are authored by hand in data/fonoran-compounds.json, then scored deterministically by the four rules. Proposals are queued for review.
npm run fonoran:regen:four-rules -- --apply
Accept or reject each compound proposal. Accepted proposals go into data/fonoran-compounds.json on the next regenerate.
Reload editorial seeds, re-rank compounds, assign phonetic spellings, and rebuild the full dictionary from the canonical JSON files in git.
Run the golden translation suite to confirm previously working concepts still resolve correctly after any changes.
Destructive maintenance only. Prefer Regenerate dictionary above for prod recovery.
Requires typing BUILD. Use only when editorial seeds are already loaded. Blocked on prod when seeds are stale.
Reset lab clears Dictionary, Review queue, and assigned sounds. Concept inventory is kept.
Historical bulk generators are archived in Generator archive.
Nothing selected
Pick a root or compound from the list, or create a new one.
Fonora is an open-source, experimental writing system that represents speech using a small set of symbols based on how sounds are physically produced in the mouth.
Rather than preserving historical spelling conventions, Fonora writes pronunciation directly. Each symbol represents either where a sound is produced or how it is modified, allowing readers to infer pronunciation from the written form itself.
The project explores whether a compact, highly learnable script can represent speech across many languages while remaining visually simple and consistent.
Understand
See how place and manner symbols combine to represent speech sounds.
Reference
Browse the full phonetic inventory and how it maps across languages.
Try it
Type text in a supported language and hear it rendered in Fonora.
Practice
Practice reading symbols, hearing pronunciation, and spelling from sound.
Learn more
For design rationale behind the script, see the language rules.
Type text to transliterate into Fonora, then press Play to hear it with a matching neural or eSpeak voice.
—
—
Press Tab for next word · ↵ check
—
↵ Check · Tab Continue · vowel keys show sound options
Five places of articulation × four manners. The vowel indicator ⚬ is documented under Modifiers and used in the vowel table, not in this grid.
Reversed-order and modifier-pair sounds from the Sound Grid above. Same place and manner symbols, written in a different order, so sounds like th, dh, v, and z can be written.
| Symbols | Sound | IPA | Status | Explanation |
|---|
Simple vowels are ⚬X (2 symbols). Diphthongs are ⚬XᵔY (4 symbols): vowel indicator ⚬, a place or manner glyph, approximant ᵔ, and a destination place.
| Symbols | Key | Lexical set | Recipe | IPA | Example |
|---|
Evaluate IPA → Fonora encoding across languages. Review outputs, tag failures, and export results.
Multilingual test set for validating one IPA pipeline across languages (English, Spanish, French, German, Japanese, Arabic). Use the dialect comparison set to compare en-us, en-gb, en-uk-rp, en-au, en-nz, and en-sc on tomato, water, car, dance, and route.
Pick one or more categories, then generate 20–50 test cards.
Generate fake but readable sound strings to test encoder/decoder consistency.
Paste words, one per line.
Validate whether Fonora preserves pronunciation through the full pipeline: English → IPA → Fonora phonemes → symbols → recovered phonemes → recovered IPA.
Validates the dedicated vowel word set through the full pipeline. Every output must use ⚬X or ⚬XᵔY. No legacy ⚬⚬ marker.
Sound-to-symbol reference for consonants, derived sounds, and vowels. Grid consonants match the ; reserved throat slots have no encoder key yet.
| Phoneme key | Symbol | IPA | Notes |
|---|
| Phoneme key | Symbol | IPA | Notes |
|---|
| Phoneme key | Symbol | IPA | Notes |
|---|
| Phoneme key | Symbol | IPA | Notes |
|---|
Speech-based writing across languages
Fonora is designed to represent speech, not spelling. These samples show how paragraph-length text from different languages may appear when rendered through the Fonora phonetic system.
The goal is to visually demonstrate how Fonora can represent speech across languages and dialects using the same symbol inventory regardless of the source writing system.
Fonora is experimental. Samples may change as the phonetic mapping system improves. Listen uses neural voices and reads recovered Fonora phonetics. Japanese audio is not available yet.
Select a document from the sidebar.
Tap to listen
Tap to listen