[真实全双工运行时]:完成run-agent-live可打断语音闭环,包含持续监听、软件回声抑制和播放取消验证

This commit is contained in:
mkbk
2026-06-18 22:56:01 +08:00
parent e730883c64
commit a489f8eee1
11 changed files with 378 additions and 46 deletions
+77 -7
View File
@@ -10,19 +10,21 @@ from owner_voice_pet.transport import MemoryAudioTransport
from owner_voice_pet.vad import EnergyVadProvider
def speech_frame(idx: int, speaker_id: str, partial: str = "等一下") -> AudioFrame:
def speech_frame(idx: int, speaker_id: str, partial: str | None = "等一下") -> AudioFrame:
metadata = {
"duration_ms": 20,
"speech": True,
"speaker_id": speaker_id,
}
if partial is not None:
metadata["partial_transcript"] = partial
return AudioFrame(
b"\xff\x7f" * 320,
16000,
1,
idx * 20,
idx,
{
"duration_ms": 20,
"speech": True,
"speaker_id": speaker_id,
"partial_transcript": partial,
},
metadata,
)
@@ -127,6 +129,74 @@ class BargeInTests(unittest.TestCase):
self.assertTrue(monitor.interrupted)
def test_async_monitor_interrupts_without_realtime_partial(self) -> None:
gate = BargeInSpeakerGate(
enabled=True,
user_similarity_threshold=0.62,
assistant_reject_threshold=0.72,
min_rms=0.001,
)
gate.remember_user_segment(segment_for_speaker("owner"))
transport = MemoryAudioTransport([speech_frame(1, "owner", None), speech_frame(2, "owner", None)])
transport.start_input()
monitor = AsyncBargeInMonitor(
transport=transport,
vad_provider=EnergyVadProvider(threshold=1),
realtime_stt=None,
speaker_gate=gate,
assistant_profile=gate.assistant_profile(segment_for_speaker("assistant")),
echo_guard_ms=0,
min_speech_ms=40,
listen_interval_ms=1,
)
monitor.vad_provider.load()
monitor.start()
deadline = time.monotonic() + 1
while not monitor.interrupted and time.monotonic() < deadline:
time.sleep(0.005)
monitor.stop()
self.assertTrue(monitor.interrupted)
self.assertEqual(len(monitor.pending_frames()), 2)
def test_render_reference_echo_does_not_interrupt(self) -> None:
gate = BargeInSpeakerGate(
enabled=True,
user_similarity_threshold=0.62,
assistant_reject_threshold=0.72,
min_rms=0.001,
)
assistant = AudioSegment(b"\x01\x20\x02\x20" * 320, 16000, 1, 0, 40)
echo_frame = AudioFrame(
assistant.pcm[: 640 * 2],
16000,
1,
20,
1,
{"duration_ms": 40, "speech": True},
)
transport = MemoryAudioTransport([echo_frame])
transport.start_input()
monitor = AsyncBargeInMonitor(
transport=transport,
vad_provider=EnergyVadProvider(threshold=1),
realtime_stt=None,
speaker_gate=gate,
assistant_profile=gate.assistant_profile(assistant),
assistant_reference=assistant,
echo_guard_ms=0,
min_speech_ms=40,
listen_interval_ms=1,
)
monitor.vad_provider.load()
monitor.start()
time.sleep(0.05)
monitor.stop()
self.assertFalse(monitor.interrupted)
if __name__ == "__main__":
unittest.main()