diff --git a/pecha_api/plans/audio/tts_service.py b/pecha_api/plans/audio/tts_service.py
index 7ddd06e7..a656db0e 100644
--- a/pecha_api/plans/audio/tts_service.py
+++ b/pecha_api/plans/audio/tts_service.py
@@ -5,9 +5,6 @@
 from pecha_api.plans.audio.monlam_tts_service import generate_monlam_tts_audio
 from pecha_api.plans.plans_enums import PlanAudioType
 
-SUPPORTED_TTS_LANGUAGES = {"en", "bo"}
-
-
 def _normalize_language(language: str) -> str:
     return (language or "en").strip().lower()
 
@@ -22,10 +19,6 @@ def generate_tts_audio(
         raise ValueError("Content cannot be empty")
 
     normalized_language = _normalize_language(language)
-    if normalized_language not in SUPPORTED_TTS_LANGUAGES:
-        raise ValueError(
-            f"Unsupported language for TTS: {language}. Supported: {', '.join(sorted(SUPPORTED_TTS_LANGUAGES))}"
-        )
 
     if normalized_language == "bo":
         return generate_monlam_tts_audio(content, voice_name=voice_name)
diff --git a/tests/plans/audio/test_tts_service.py b/tests/plans/audio/test_tts_service.py
index e6a7c0ff..ec3884b1 100644
--- a/tests/plans/audio/test_tts_service.py
+++ b/tests/plans/audio/test_tts_service.py
@@ -68,13 +68,21 @@ def test_generate_tts_audio_rejects_empty_content():
         generate_tts_audio(content="   ", audio_type=PlanAudioType.RECITATION)
 
 
-def test_generate_tts_audio_rejects_unsupported_language():
-    with pytest.raises(ValueError, match="Unsupported language for TTS"):
-        generate_tts_audio(
-            content="Hello",
-            audio_type=PlanAudioType.RECITATION,
-            language="zh",
-        )
+@patch("google.genai.Client")
+@patch("pecha_api.plans.audio.tts_service.get", return_value="test-api-key")
+def test_generate_tts_audio_routes_non_tibetan_languages_to_gemini(mock_get, mock_client_cls):
+    audio_data = b"\x00\x01\x02\x03"
+    _configure_gemini_client(mock_client_cls, _build_gemini_audio_response(audio_data=audio_data))
+
+    result = generate_tts_audio(
+        content="你好",
+        audio_type=PlanAudioType.RECITATION,
+        language="zh",
+    )
+
+    assert result[:4] == b"RIFF"
+    assert result.endswith(audio_data)
+    mock_client_cls.assert_called_once()
 
 
 @patch("pecha_api.plans.audio.tts_service.generate_monlam_tts_audio")
diff --git a/tests/plans/audio/test_tts_test_views.py b/tests/plans/audio/test_tts_test_views.py
index 4941fbd7..aedfd45e 100644
--- a/tests/plans/audio/test_tts_test_views.py
+++ b/tests/plans/audio/test_tts_test_views.py
@@ -44,13 +44,25 @@ def test_preview_tts_returns_502_for_runtime_error(mock_generate):
 
 @patch(
     "pecha_api.plans.audio.tts_test_views.generate_tts_audio",
-    side_effect=ValueError("Unsupported language for TTS: zh"),
+    return_value=b"RIFF" + b"\x00" * 40,
+)
+def test_preview_tts_returns_wav_for_non_tibetan_language(mock_generate):
+    response = _post_tts_preview(language="zh", type="TEXT_READING")
+
+    assert response.status_code == 200
+    assert response.headers["content-type"] == "audio/wav"
+    mock_generate.assert_called_once()
+
+
+@patch(
+    "pecha_api.plans.audio.tts_test_views.generate_tts_audio",
+    side_effect=ValueError("Content cannot be empty"),
 )
 def test_preview_tts_returns_400_for_validation_error(mock_generate):
-    response = _post_tts_preview(language="zh")
+    response = client.post("/preview", json={"text": "   ", "language": "en"})
 
     assert response.status_code == 400
-    assert "Unsupported language" in response.json()["detail"]
+    assert "Content cannot be empty" in response.json()["detail"]
 
 
 @patch(