Add ASR interim results support in Assistant model and API

- Introduced `asr_interim_enabled` field in the Assistant model to control interim ASR results.
- Updated AssistantBase and AssistantUpdate schemas to include the new field.
- Modified the database schema to add the `asr_interim_enabled` column.
- Enhanced runtime metadata to reflect interim ASR settings.
- Updated API endpoints and tests to validate the new functionality.
- Adjusted documentation to include details about interim ASR results configuration.
This commit is contained in:
Xin Wang
2026-03-06 12:58:54 +08:00
parent e11c3abb9e
commit da38157638
19 changed files with 183 additions and 5 deletions

View File

@@ -32,6 +32,7 @@ def test_create_asr_service_openai_compatible_returns_offline_provider():
)
assert isinstance(service, OpenAICompatibleASRService)
assert service.mode == "offline"
assert service.enable_interim is False
def test_create_asr_service_fallback_buffered_for_unsupported_provider():

View File

@@ -282,7 +282,7 @@ async def test_local_yaml_adapter_rejects_path_traversal_like_assistant_id(tmp_p
@pytest.mark.asyncio
async def test_local_yaml_translates_agent_schema_to_runtime_services(tmp_path):
async def test_local_yaml_translates_agent_schema_with_asr_interim_flag(tmp_path):
config_dir = tmp_path / "assistants"
config_dir.mkdir(parents=True, exist_ok=True)
(config_dir / "default.yaml").write_text(
@@ -305,6 +305,7 @@ async def test_local_yaml_translates_agent_schema_to_runtime_services(tmp_path):
" model: asr-model",
" api_key: sk-asr",
" api_url: https://asr.example.com/v1/audio/transcriptions",
" enable_interim: false",
" duplex:",
" system_prompt: You are test assistant",
]
@@ -321,4 +322,5 @@ async def test_local_yaml_translates_agent_schema_to_runtime_services(tmp_path):
assert services.get("llm", {}).get("apiKey") == "sk-llm"
assert services.get("tts", {}).get("apiKey") == "sk-tts"
assert services.get("asr", {}).get("apiKey") == "sk-asr"
assert services.get("asr", {}).get("enableInterim") is False
assert assistant.get("systemPrompt") == "You are test assistant"

View File

@@ -145,10 +145,11 @@ async def test_start_asr_capture_uses_streaming_begin(monkeypatch):
@pytest.mark.asyncio
async def test_start_asr_capture_uses_offline_interim_control(monkeypatch):
async def test_start_asr_capture_uses_offline_interim_control_when_enabled(monkeypatch):
asr = _FakeOfflineASR()
pipeline = _build_pipeline(monkeypatch, asr)
pipeline._asr_mode = "offline"
pipeline._runtime_asr["enableInterim"] = True
pipeline._pending_speech_audio = b"\x00" * 320
pipeline._pre_speech_buffer = b"\x00" * 640
@@ -159,6 +160,69 @@ async def test_start_asr_capture_uses_offline_interim_control(monkeypatch):
assert pipeline._asr_capture_active is True
@pytest.mark.asyncio
async def test_start_asr_capture_skips_offline_interim_control_when_disabled(monkeypatch):
asr = _FakeOfflineASR()
pipeline = _build_pipeline(monkeypatch, asr)
pipeline._asr_mode = "offline"
pipeline._runtime_asr["enableInterim"] = False
pipeline._pending_speech_audio = b"\x00" * 320
pipeline._pre_speech_buffer = b"\x00" * 640
await pipeline._start_asr_capture()
assert asr.start_interim_calls == 0
assert asr.sent_audio
assert pipeline._asr_capture_active is True
@pytest.mark.asyncio
async def test_offline_interim_callback_ignored_when_disabled(monkeypatch):
asr = _FakeOfflineASR()
pipeline = _build_pipeline(monkeypatch, asr)
pipeline._asr_mode = "offline"
pipeline._runtime_asr["enableInterim"] = False
captured_events = []
captured_deltas = []
async def _capture_event(event: Dict[str, Any], priority: int = 20):
_ = priority
captured_events.append(event)
async def _capture_delta(text: str):
captured_deltas.append(text)
monkeypatch.setattr(pipeline, "_send_event", _capture_event)
monkeypatch.setattr(pipeline, "_emit_transcript_delta", _capture_delta)
await pipeline._on_transcript_callback("ignored interim", is_final=False)
assert captured_events == []
assert captured_deltas == []
assert pipeline._latest_asr_interim_text == ""
@pytest.mark.asyncio
async def test_offline_final_callback_emits_when_interim_disabled(monkeypatch):
asr = _FakeOfflineASR()
pipeline = _build_pipeline(monkeypatch, asr)
pipeline._asr_mode = "offline"
pipeline._runtime_asr["enableInterim"] = False
captured_events = []
async def _capture_event(event: Dict[str, Any], priority: int = 20):
_ = priority
captured_events.append(event)
monkeypatch.setattr(pipeline, "_send_event", _capture_event)
await pipeline._on_transcript_callback("final only", is_final=True)
assert any(event.get("type") == "transcript.final" for event in captured_events)
@pytest.mark.asyncio
async def test_streaming_eou_falls_back_to_latest_interim(monkeypatch):
asr = _FakeStreamingASR()