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 { loadAudioManager } = require("./harness/audioManager");
const LOCAL_WHISPER_SETTINGS = {
useLocalWhisper: true,
localTranscriptionProvider: "whisper",
whisperModel: "base",
cloudTranscriptionMode: "byok",
isSignedIn: false,
};
async function loadManagerClass(t, settings = LOCAL_WHISPER_SETTINGS) {
const { AudioManager } = await loadAudioManager(t, {
cachePrefix: "openwhispr-no-audio-lifecycle-test-",
settingsKey: "__noAudioLifecycleSettings",
settings,
});
return AudioManager;
}
function createManager(AudioManager, failure) {
const order = [];
const saved = [];
const manager = Object.assign(Object.create(AudioManager.prototype), {
isProcessing: true,
_localSpeechGateState: null,
pendingAssistantConversation: null,
pendingSelectionEdit: null,
lastAudioBlob: {},
processWithLocalWhisper: async () => {
throw failure;
},
processWithOpenWhisprCloud: async () => {
throw failure;
},
onStateChange: (state) => order.push(state.isProcessing ? "processing" : "idle"),
onNoAudio: () => order.push("no-audio"),
onError: () => order.push("error"),
saveFailedTranscription: (message, code) => saved.push({ message, code }),
});
return { manager, order, saved };
}
test("local silence becomes one no-audio outcome after processing is idle", async (t) => {
const AudioManager = await loadManagerClass(t);
const { manager, order, saved } = createManager(AudioManager, new Error("No audio detected"));
await manager.processAudio({ size: 256, type: "audio/webm" });
assert.equal(manager.isProcessing, false);
assert.deepEqual(order, ["idle", "no-audio"]);
assert.deepEqual(saved, []);
});
test("dictionary-echo silence keeps the recording but shares the settled outcome", async (t) => {
const AudioManager = await loadManagerClass(t);
const { DICTIONARY_ECHO_CODE } = await import("../../src/utils/dictionaryEchoFilter.js");
const failure = new Error("No audio detected");
failure.code = DICTIONARY_ECHO_CODE;
const { manager, order, saved } = createManager(AudioManager, failure);
await manager.processAudio({ size: 256, type: "audio/webm" });
assert.deepEqual(order, ["idle", "no-audio"]);
assert.deepEqual(saved, [{ message: "No audio detected", code: DICTIONARY_ECHO_CODE }]);
});
test("Cloud finding no speech is the no-audio outcome and keeps the recording", async (t) => {
const AudioManager = await loadManagerClass(t, {
useLocalWhisper: false,
cloudTranscriptionMode: "openwhispr",
isSignedIn: true,
});
// cloud-transcribe's shape for the API's 422 (ipcHandlers.js).
const failure = Object.assign(new Error("No speech detected in audio"), {
code: "NO_SPEECH_DETECTED",
});
const { manager, order, saved } = createManager(AudioManager, failure);
await manager.processAudio({ size: 256, type: "audio/webm" });
assert.deepEqual(order, ["idle", "no-audio"]);
assert.deepEqual(saved, [{ message: "No speech detected in audio", code: "NO_SPEECH_DETECTED" }]);
});