Rime IVR Preview Studio

The app generates one preview clip per IVR node and caches prompt/model/voice combinations so repeat previews stay fast.

const response = await fetch(process.env.RIME_API_URL ?? RIME_ENDPOINT, {
  method: "POST",
  headers: {
    Accept: "audio/webm;codecs=opus",
    Authorization: `Bearer ${apiKey}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: prompt,
    modelId: flow.modelId,
    speaker: flow.speaker,
    lang: flow.lang,
    samplingRate: 24000,
    timeScaleFactor: 1.12,
  }),
});

Voice demos break trust when brand names, medications, products, or domain terms sound wrong. The coverage check makes those risks visible early.

const response = await fetch("https://users.rime.ai/oov", {
  method: "POST",
  headers: {
    Accept: "application/json",
    Authorization: `Bearer ${apiKey}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ text }),
});

The handoff is not just a pretty demo. It exports files that help a team wire the flow into telephony, Vapi, LiveKit, Twilio, or a contact-center runtime.

const files: ExportFile[] = [
  { path: "ivr-flow.json", language: "json", content: JSON.stringify(flow, null, 2) },
  { path: "curl-examples.md", language: "markdown", content: buildCurlExamples(flow) },
  { path: "node-example.ts", language: "typescript", content: buildNodeExample(flow) },
  { path: "python-example.py", language: "python", content: buildPythonExample(flow) },
  { path: "pronunciation-checklist.md", language: "markdown", content: buildPronunciationChecklist(flow) },
];