Update: デフォルトモデルの追加に伴い、E2E テストで StyleBertVITS2TTSEngine を用いるよう変更

これで CI により最低限音声合成できることが保証されるはず

test/e2e/__snapshots__/test_tts.ambr

@@ -1,4 +1,4 @@
 # serializer version: 1
 # name: test_テキストとキャラクターIDから音声を合成できる
-  'MD5:8f7ddc461c68542d4d8ef4cd5c54ca82'
+  'MD5:015a2f6482f2a893c4c61e63c10993b3'
 # ---

test/e2e/conftest.py

@@ -9,12 +9,15 @@
 
 from voicevox_engine.aivm_manager import AivmManager
 from voicevox_engine.app.application import generate_app
-from voicevox_engine.core.core_initializer import initialize_cores
+from voicevox_engine.core.core_initializer import MOCK_VER, initialize_cores
 from voicevox_engine.engine_manifest import load_manifest
 from voicevox_engine.library.library_manager import LibraryManager
 from voicevox_engine.preset.preset_manager import PresetManager
 from voicevox_engine.setting.setting_manager import SettingHandler
-from voicevox_engine.tts_pipeline.tts_engine import make_tts_engines_from_cores
+from voicevox_engine.tts_pipeline.style_bert_vits2_tts_engine import (
+    StyleBertVITS2TTSEngine,
+)
+from voicevox_engine.tts_pipeline.tts_engine import TTSEngineManager
 from voicevox_engine.user_dict.user_dict_manager import (
     DEFAULT_DICT_DIR_PATH,
     UserDictionary,
@@ -36,7 +39,11 @@ def _copy_under_dir(file_path: Path, dir_path: Path) -> Path:
 def app_params(tmp_path: Path) -> dict[str, Any]:
     aivm_manager = AivmManager(get_save_dir() / "Models")
     core_manager = initialize_cores(use_gpu=False, enable_mock=True)
-    tts_engines = make_tts_engines_from_cores(core_manager)
+    tts_engines = TTSEngineManager()
+    tts_engines.register_engine(
+        StyleBertVITS2TTSEngine(aivm_manager, False, False),
+        MOCK_VER,
+    )
     setting_loader = SettingHandler(tmp_path / "not_exist.yaml")
 
     # テスト用に隔離されたプリセットを生成する

test/e2e/single_api/tts_pipeline/__snapshots__/test_synthesis.ambr

@@ -1,7 +1,7 @@
 # serializer version: 1
 # name: test_post_synthesis_200
-  'MD5:f7d42ce5787856549abc3d2d7561c06f'
+  'MD5:bb372629df13cbec280585d3d34dc217'
 # ---
 # name: test_post_synthesis_old_audio_query_200
-  'MD5:f7d42ce5787856549abc3d2d7561c06f'
+  'MD5:f76a5639ff8327ccee59844624be2ef5'
 # ---

test/e2e/single_api/tts_pipeline/test_accent_phrases.py

@@ -12,7 +12,7 @@ def test_post_accent_phrases_200(
     client: TestClient, snapshot_json: SnapshotAssertion
 ) -> None:
     response = client.post(
-        "/accent_phrases", params={"text": "テストです", "speaker": 0}
+        "/accent_phrases", params={"text": "テストです", "speaker": 888753760}
     )
     assert response.status_code == 200
     assert snapshot_json == round_floats(response.json(), 2)

test/e2e/single_api/tts_pipeline/test_audio_query.py

@@ -11,6 +11,8 @@
 def test_post_audio_query_200(
     client: TestClient, snapshot_json: SnapshotAssertion
 ) -> None:
-    response = client.post("/audio_query", params={"text": "テストです", "speaker": 0})
+    response = client.post(
+        "/audio_query", params={"text": "テストです", "speaker": 888753760}
+    )
     assert response.status_code == 200
     assert snapshot_json == round_floats(response.json(), round_value=2)

test/e2e/single_api/tts_pipeline/test_initialize_speaker.py

@@ -9,6 +9,6 @@
 def test_post_initialize_speaker_204(
     client: TestClient, snapshot: SnapshotAssertion
 ) -> None:
-    response = client.post("/initialize_speaker", params={"speaker": 0})
+    response = client.post("/initialize_speaker", params={"speaker": 888753760})
     assert response.status_code == 204
     assert snapshot == response.content

test/e2e/single_api/tts_pipeline/test_is_initialized_speaker.py

@@ -9,6 +9,6 @@
 def test_get_is_initialized_speaker_200(
     client: TestClient, snapshot_json: SnapshotAssertion
 ) -> None:
-    response = client.get("/is_initialized_speaker", params={"speaker": 0})
+    response = client.get("/is_initialized_speaker", params={"speaker": 888753760})
     assert response.status_code == 200
     assert snapshot_json == response.json()