Detect repetition while decoding

decoder_worker.js

@@ -73,6 +73,46 @@ async function fetchModelWithProgress(url, modelName) {
 let cfg = null;
 let tailLatency = 0;
 
+// Decoding config
+const TOKENS_PER_SECOND = 6.5;  // Max tokens per second of audio
+const FRAME_DURATION_MS = 20;   // Each encoder frame is 20ms
+
+// Check for repetitive token patterns that indicate decoding should stop
+function hasRepetition(tokens) {
+    const len = tokens.length;
+    if (len < 5) return false;
+
+    // Check if last 5 tokens are the same
+    const last5 = tokens.slice(-5);
+    if (last5.every(t => t === last5[0])) {
+        return true;
+    }
+
+    // Check for 3 repeated same pairs (e.g., [A,B,A,B,A,B])
+    if (len >= 6) {
+        const pair1 = [tokens[len - 6], tokens[len - 5]];
+        const pair2 = [tokens[len - 4], tokens[len - 3]];
+        const pair3 = [tokens[len - 2], tokens[len - 1]];
+        if (pair1[0] === pair2[0] && pair2[0] === pair3[0] &&
+            pair1[1] === pair2[1] && pair2[1] === pair3[1]) {
+            return true;
+        }
+    }
+
+    // Check for 2 repeated same triples (e.g., [A,B,C,A,B,C])
+    if (len >= 6) {
+        const triple1 = [tokens[len - 6], tokens[len - 5], tokens[len - 4]];
+        const triple2 = [tokens[len - 3], tokens[len - 2], tokens[len - 1]];
+        if (triple1[0] === triple2[0] &&
+            triple1[1] === triple2[1] &&
+            triple1[2] === triple2[2]) {
+            return true;
+        }
+    }
+
+    return false;
+}
+
 // Sessions
 let adapterSession = null;
 let decoderInitSession = null;
@@ -203,7 +243,9 @@ async function decodeAccumulated() {
         await initDecoderCache(context);
 
         const numFrames = accumulatedFeatures.dims[1];
-        const maxTokens = Math.max(10, Math.floor(numFrames * 1.5));
+        // Calculate duration in seconds and max tokens based on that
+        const durationSeconds = (numFrames * FRAME_DURATION_MS) / 1000;
+        const maxTokens = Math.max(10, Math.floor(durationSeconds * TOKENS_PER_SECOND));
 
         const tokens = [1];  // BOS
         for (let step = 0; step < maxTokens; step++) {
@@ -219,7 +261,15 @@ async function decodeAccumulated() {
             }
 
             tokens.push(maxIdx);
-            if (maxIdx === 2) break;  // EOS
+
+            // Stop on EOS
+            if (maxIdx === 2) break;
+
+            // Stop on repetitive patterns
+            if (hasRepetition(tokens)) {
+                console.log('Stopping decode due to repetition detected');
+                break;
+            }
         }
 
         return tokenizer.decode(tokens, true);

index.html

@@ -207,8 +207,8 @@
             display: flex;
             flex-direction: column;
             justify-content: center;
-            align-items: center;
-            text-align: center;
+            align-items: flex-start;
+            text-align: left;
             backdrop-filter: blur(10px);
             border: 1px solid rgba(255, 255, 255, 0.1);
             box-shadow: 0 8px 32px rgba(0, 0, 0, 0.3);
@@ -404,7 +404,7 @@
                 background: rgba(0, 0, 0, 0.9);
                 border-radius: 0 0 12px 12px;
                 min-height: 80px;
-                text-align: center;
+                text-align: left;
                 backdrop-filter: blur(10px);
                 box-shadow: 0 4px 20px rgba(0, 0, 0, 0.5);
             }