Voice-to-text dictation app with local (Nvidia Parakeet/Whisper) and cloud models (BYOK). Privacy-first and available cross-platform.
JavaScript aianthropiccross-platformgeminigroqlinuxmacosnvidiaopen-sourceopenaiparakeetspeech-to-texttranscribewhisperwindows
const test = require("node:test");
const assert = require("node:assert/strict");
const load = () => import("../../src/utils/liveTranscriptPresentation.ts");
test("short live transcript keeps the whole active phrase shimmering", async () => {
const { splitTranscriptForShimmer } = await load();
assert.deepEqual(splitTranscriptForShimmer("A short live phrase", 6), {
settled: "",
active: "A short live phrase",
});
});
test("long live transcript shimmers only its trailing phrase", async () => {
const { splitTranscriptForShimmer } = await load();
const text = "one two three four five six seven eight nine ten";
assert.deepEqual(splitTranscriptForShimmer(text, 4), {
settled: "one two three four five six ",
active: "seven eight nine ten",
});
});
test("empty live transcript has no shimmer parts", async () => {
const { splitTranscriptForShimmer } = await load();
assert.deepEqual(splitTranscriptForShimmer(" "), { settled: "", active: "" });
});
test("default shimmer work stays bounded to the newest six words", async () => {
const { splitTranscriptForShimmer } = await load();
assert.deepEqual(splitTranscriptForShimmer("one two three four five six seven eight"), {
settled: "one two ",
active: "three four five six seven eight",
});
});