diff --git a/apps/web/src/lib/__tests__/transcript-search.test.ts b/apps/web/src/lib/__tests__/transcript-search.test.ts
index f94e35c22..105269a1c 100644
--- a/apps/web/src/lib/__tests__/transcript-search.test.ts
+++ b/apps/web/src/lib/__tests__/transcript-search.test.ts
@@ -54,6 +54,10 @@ describe('filterSegments — empty/null query behavior (#908)', () => {
       segments.length,
     );
   });
+
+  it('returns the original segment array when no filters are active', () => {
+    expect(filterSegments(segments, { search: '' })).toBe(segments);
+  });
 });
 
 describe('filterSegments — null-safe segment text (#908)', () => {
diff --git a/apps/web/src/lib/transcript-search.ts b/apps/web/src/lib/transcript-search.ts
index 7a22c05aa..0c1c12ab9 100644
--- a/apps/web/src/lib/transcript-search.ts
+++ b/apps/web/src/lib/transcript-search.ts
@@ -68,6 +68,8 @@ export function filterSegments(
   segments: readonly TranscriptSegment[],
   { search, speaker }: TranscriptFilter,
 ): TranscriptSegment[] {
+  if (!search && !speaker) return segments as TranscriptSegment[];
+
   // Hoisted out of the loop: removes N toLowerCase() allocations per keystroke
   // update on long transcripts. This is the optimization PR #972 shipped.
   const lowerSearchQuery = search ? search.toLowerCase() : '';