@speakableio/core 1.0.64 → 1.0.65

package/dist/index.native.mjs

@@ -2217,6 +2217,10 @@ function detectTranscriptHallucination(transcript) {
     return false;
   }
   const text = transcript.trim();
+  const wordCount = text.split(/\s+/).filter(Boolean).length;
+  if (text.length < 120 || wordCount < 20) {
+    return false;
+  }
   const shortRepeats = detectShortRepeats(text);
   if (shortRepeats) return true;
   const phraseRepeats = detectPhraseRepeats(text);
@@ -2232,14 +2236,14 @@ function detectShortRepeats(text) {
   for (let i = 1; i < words.length; i++) {
     if (words[i] === words[i - 1]) {
       repeatCount++;
-      if (repeatCount >= 3) return true;
+      if (repeatCount >= 4) return true;
     } else {
       repeatCount = 1;
     }
   }
   const uniqueWords = new Set(words);
   const repetitionRatio = words.length / uniqueWords.size;
-  if (words.length >= 10 && uniqueWords.size <= 3 && repetitionRatio >= 3) {
+  if (words.length >= 12 && uniqueWords.size <= 2 && repetitionRatio >= 5) {
     return true;
   }
   return false;
@@ -2256,12 +2260,12 @@ function detectPhraseRepeats(text) {
         break;
       }
     }
-    if (consecutiveRepeats >= 2) {
+    if (consecutiveRepeats >= 3) {
       return true;
     }
   }
   const uniqueSentences = new Set(sentences);
-  if (sentences.length >= 3 && uniqueSentences.size === 1) {
+  if (sentences.length >= 4 && uniqueSentences.size === 1) {
     return true;
   }
   return false;
@@ -2278,7 +2282,7 @@ function isSimilarSentence(s1, s2) {
   const set2 = new Set(words2);
   const intersection = new Set([...set1].filter((w) => set2.has(w)));
   const similarity = intersection.size * 2 / (set1.size + set2.size);
-  return similarity >= 0.8;
+  return similarity >= 0.9;
 }
 function detectCyclicPattern(text) {
   const normalized = text.toLowerCase().replace(/\s+/g, " ").trim();
@@ -2298,7 +2302,7 @@ function detectCyclicPattern(text) {
         break;
       }
     }
-    if (matchCount >= 3) {
+    if (matchCount >= 4) {
       return true;
     }
   }
@@ -2319,7 +2323,7 @@ async function getTranscript(model, args, cleanHallucinations = true) {
   const getAssemblyAITranscript = (_d = (_c = api).httpsCallable) == null ? void 0 : _d.call(_c, "transcribeAssemblyAIAudio");
   const getWhisper3Transcript = (_f = (_e = api).httpsCallable) == null ? void 0 : _f.call(_e, "generateGroqTranscript");
   const getWhisper1Transcript = (_h = (_g = api).httpsCallable) == null ? void 0 : _h.call(_g, "transcribeAudio");
-  console.log("Getting transcript from", model);
+  console.log("Getting transcript from", model, " cleanHallucinations", cleanHallucinations);
   if (model === "whisper-3") {
     try {
       const { data } = await (getWhisper3Transcript == null ? void 0 : getWhisper3Transcript({
@@ -2372,13 +2376,19 @@ async function getTranscript(model, args, cleanHallucinations = true) {
   return null;
 }
 async function getTranscriptCycle(args) {
+  var _a;
   const models = ["whisper-3", "whisper", "gemini", "assemblyai"];
   let transcript = "";
   let lastError = null;
   for (const model of models) {
     try {
-      console.log("Getting transcript from", model);
-      const transcriptResult = await getTranscript(model, args, false);
+      console.log(
+        "Getting transcript from",
+        model,
+        " cleanHallucinations",
+        args.cleanHallucinations
+      );
+      const transcriptResult = await getTranscript(model, args, (_a = args.cleanHallucinations) != null ? _a : true);
       const rawTranscript = transcriptResult || "";
       transcript = cleanHallucinatedTranscript(rawTranscript);
       if (transcript !== "") {
@@ -3137,9 +3147,10 @@ function useSpeakableTranscript() {
       model,
       audioUrl,
       language,
-      prompt
+      prompt,
+      cleanHallucinations = true
     }) => {
-      return getTranscript(model, { audioUrl, language, prompt });
+      return getTranscript(model, { audioUrl, language, prompt }, cleanHallucinations);
     },
     retry: false
   });
@@ -3150,7 +3161,7 @@ function useSpeakableTranscript() {
 function useSpeakableTranscriptCycle() {
   const mutation = useMutation3({
     mutationFn: async (args) => {
-      return getTranscriptCycle(args);
+      return getTranscriptCycle({ ...args, cleanHallucinations: args.cleanHallucinations });
     },
     retry: false
   });