kubectlA community dictionary of how engineers actually say kubectl, nginx, GIF, JSON, Pydantic, Knative, LaTeX, Postgres… with sources.
git clone https://github.com/anzy-renlab-ai/pronounce.git && cd pronounce && ./install.sh
$ say-it kubectl 🔊 koob control. koob control. koob control. or: cube cuddle. or: kube C T L. $ say-it GIF 🔊 jif. jif. jif. or: gif. # with receipts — Wilhite at the Webby Awards, 2013 $ say-it --why JSON word JSON ipa /ˈdʒeɪsən/ respelling_us jay son source Wikipedia § Pronunciation url https://en.wikipedia.org/wiki/JSON#Pronunciation
Some pronunciations aren't opinions — the creators settled them on record.
Project names, product names, programmer jargon, acronyms. Every entry tagged with confidence (creator-clarified, community-consensus, contested) and linked to a real source where one exists.
When a word is contested — GIF, SQL, GUI, kubectl — the CLI audibly chains the alternates ("…or: gif"), so you hear the debate without watching the terminal.
Ask Claude "how do you pronounce X?" — it replies with audio, IPA, and a source citation, not a phonetic guess.
Wraps the macOS `say` engine you already have. No npm, no sudo, no surprises.
--alt for the rival reading, --all for every variant, --solo to skip the chain, --why shows the dict entry with source URL.
The dictionary is a TSV file you can edit. Open a PR with your favorite mispronounced project — see CONTRIBUTING.md.
IPA is a reference — like a dictionary entry for a foreign word. You don't learn how to pronounce schadenfreude by squinting at /ˈʃɑːdənˌfrɔɪdə/. You learn it by hearing someone say it three times. Pronounce wires your OS's TTS to a one-shot CLI so the answer is sound, not a phonetic transcription.
Click ▶ to hear it in your browser. (Audio quality varies by browser — install the CLI for the macOS Samantha rendering.)
Because you'd have to hunt for the right clip, unmute, wait, and rewind. say-it kubectl plays the right reading three times in 4 seconds. The site is here for when you're not at a terminal.
The CLI uses macOS's built-in say with a tuned respelling. The site uses your browser's Web Speech API, whose voice and pronunciation rules vary by OS and browser. For a contested word like GIF they should agree; for projects with quirky readings the CLI is the canonical one.
This one is for the names engineers actually use — kubectl, nginx, Pydantic, Knative, Cilium. Webster doesn't cover them; this does, with the additional value that each entry is tagged with a confidence level and (where possible) a citable source.
Both readings are real. The dictionary picks the creator's stated reading as primary ("jif", per Steve Wilhite at the 2013 Webby Awards) and surfaces "gif" as the alternate. Run say-it --alt GIF to hear the alternate. Same pattern for SQL, JSON, char, regex, and the other contested ones.
Yes — Windows (PowerShell + System.Speech) and Linux (espeak-ng / cloud TTS) are M2/M3 on the roadmap. The dictionary itself is platform-agnostic; only the playback engine needs the platform-specific backend. PRs welcome.
Open a PR adding a row to data/pronunciations.tsv. See CONTRIBUTING.md for the column format. There's also a pinned issue with a wishlist of words we want next.
No. If no source exists, leave it blank and mark confidence as community-consensus. We'd rather under-claim than fabricate.
Multi-reading words carry context — you should know there's a debate. The audible "or: <alt>" tail makes that perceptible without watching the terminal. Use --solo to skip it once you've internalized the debate.
The dictionary is community-maintained — every star nudges more devs to contribute their favorite mispronounced project name.
★ Star on GitHub