Now, in addition to checking coverage in the Rime dashboard, you can determine if a word is in our dictionary via API.
Why word coverage matters for production TTS
Voice agents earn trust on the hard words: brand names, product SKUs, medication names, street addresses, and the surnames of the customers they’re talking to. A text-to-speech model that handles everyday vocabulary but stumbles on your company’s own product name will undermine an otherwise excellent call. Checking coverage ahead of time turns pronunciation into something you verify before launch instead of discovering in production.
How the coverage endpoint works
Send a word to the endpoint and the response tells you whether it exists in Rime’s pronunciation dictionary. Because it’s a simple API call, you can build the check directly into your workflow: validate an IVR script as it’s written, sweep your product catalog before a release, or run new prompt content through a CI step so nothing unverified reaches callers.
Check it out in our API documentation (includes code snippets).
What to do when a word isn’t covered
If a word isn’t covered, you have options:
- Request a dictionary update (we can add new words in ~24 hours). Just email us.
- Instantly generate a custom pronunciation using the Rime phonetic alphabet (documentation).
Even if a word isn’t in our dictionary, our TTS model will still attempt a pronunciation, often with great results, especially for words that follow common linguistic patterns. English spelling maps to sound in mostly regular ways, and our models learn those patterns the way human readers do, so an unusual but well-behaved word usually comes out right on the first try.
Beyond single words: numbers, dates, and spelled-out strings
Dictionary coverage is one layer of pronunciation control. Rime’s text normalization decides how numbers, dates, currency amounts, and street addresses are read aloud, so “1600 Pennsylvania Ave” is read as an address and “$1,299.00” as a price. The spell() function forces letter-by-letter delivery for confirmation codes and account numbers. And when you need exact phonetics inside a single request, Mist v2 supports inline pronunciation overrides written in the Rime phonetic alphabet.
Checking pronunciation from your editor
If you build with Claude, Codex, or another MCP-compatible tool, the Rime MCP server at mcp.rime.ai brings these checks into your development loop. Your coding assistant can check pronunciation, browse the voice catalog, and generate audio samples on the spot, so you can hear how a tricky surname sounds while the prompt that contains it is still open in your editor.
A pre-launch pronunciation checklist
Teams shipping a production voice agent tend to converge on the same four steps:
- Export every name, product, and term your agent will say: catalog entries, provider directories, branch addresses, drug names.
- Run the list through the coverage endpoint and collect the misses.
- Fix gaps with a custom pronunciation or a dictionary request, whichever fits your timeline.
- Re-run the check whenever the underlying content changes, ideally as an automated step in your release process.
An hour of coverage work before launch beats a week of bug reports about a mispronounced brand name after it.
Pronunciation control at enterprise scale
Enterprises in healthcare, financial services, and telecom typically maintain their own vocabulary of names and terms that must be spoken correctly every single time. The coverage endpoint pairs with Rime’s custom pronunciation tools to make that manageable at scale: audit your full vocabulary programmatically, fix any gaps with custom pronunciations or dictionary updates, and re-verify whenever your content changes.
This tooling reflects how Rime was built. Our team of linguists treats pronunciation as a first-class product surface, which is why coverage tools ship in both the dashboard and the API.
Learn more about our pronunciation tools in our original announcement, or start testing coverage today with your Rime API key.
More from the lab
Writing for the ear: Prompting your TTS to sound human
Rime was founded by linguists. We spent years studying how people actually talk before we trained models on it, so our prompting guide teaches something different: how to transcribe speech the way it’s spoken. Below is the short version of that guide and gives some context behind why we do it this way.
.png)
Text to Speech for Business: How to Choose a TTS App That Scales
This guide shows you how to pick a TTS app that holds up in production.
What Is Back-Channeling in Conversational AI? A Plain-English Guide
This guide explains back-channeling in conversational AI in plain English. You'll learn what it is and how a voice agent decides when to use it. You'll also see how it produces a natural cue and why timing makes or breaks the effect.