feat: add ASR provider support for voice co-creation

backend/tests/test_provider_router.py

@@ -244,8 +244,9 @@ class TestProviderPolicy:
         policies = list_capability_policies()
         capabilities = {item["capability"] for item in policies}
 
-        assert capabilities == {"text", "image", "tts", "storybook"}
+        assert capabilities == {"text", "image", "tts", "storybook", "asr"}
         assert DEFAULT_PROVIDERS["storybook"] == ["storybook_primary"]
+        assert DEFAULT_PROVIDERS["asr"] == ["demo"]
 
     def test_demo_provider_only_added_to_supported_capabilities(self):
         settings = SimpleNamespace(
@@ -253,6 +254,7 @@ class TestProviderPolicy:
             image_providers=["cqtai"],
             tts_providers=["edge_tts"],
             storybook_providers=["storybook_primary"],
+            asr_providers=["openai_asr"],
             enable_demo_providers=True,
         )
 
@@ -263,6 +265,7 @@ class TestProviderPolicy:
             "storybook_primary",
         ]
         assert get_provider_names_from_settings("tts", settings) == ["edge_tts"]
+        assert get_provider_names_from_settings("asr", settings) == ["demo", "openai_asr"]
 
     def test_policy_defaults_when_settings_lists_are_empty(self):
         settings = SimpleNamespace(
@@ -270,6 +273,7 @@ class TestProviderPolicy:
             image_providers=[],
             tts_providers=[],
             storybook_providers=[],
+            asr_providers=[],
             enable_demo_providers=False,
         )
 
@@ -279,6 +283,22 @@ class TestProviderPolicy:
             "elevenlabs",
             "edge_tts",
         ]
+        assert get_provider_names_from_settings("asr", settings) == ["demo"]
+
+    @pytest.mark.asyncio
+    async def test_asr_demo_provider_uses_transcript_hint(self):
+        from app.services import provider_router
+
+        result = await provider_router.transcribe_audio(
+            audio_bytes=b"fake-audio",
+            file_name="turn.webm",
+            mime_type="audio/webm",
+            transcript_hint="我想听一个小熊找星星的故事",
+        )
+
+        assert result.transcript_text == "我想听一个小熊找星星的故事"
+        assert result.confidence == 1.0
+        assert result.provider == "demo"
 
 
 class TestProviderConfigFromDB: