wx44wx/AI-VideoAssistant
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Use openai compatible as vendor

Browse Source
This commit is contained in:
Xin Wang
2026-02-12 18:44:55 +08:00
parent 260ff621bf
commit ff3a03b1ad
23 changed files with 822 additions and 905 deletions
Download Patch File Download Diff File Expand all files Collapse all files

117
engine/core/duplex_pipeline.py
View File

@@ -30,8 +30,8 @@ from processors.vad import SileroVAD, VADProcessor
from services.asr import BufferedASRService
from services.base import BaseASRService, BaseLLMService, BaseTTSService, LLMMessage, LLMStreamEvent
from services.llm import MockLLMService, OpenAILLMService
from services.siliconflow_asr import SiliconFlowASRService
from services.siliconflow_tts import SiliconFlowTTSService
from services.openai_compatible_asr import OpenAICompatibleASRService
from services.openai_compatible_tts import OpenAICompatibleTTSService
from services.streaming_text import extract_tts_sentence, has_spoken_content
from services.tts import EdgeTTSService, MockTTSService
@@ -60,57 +60,6 @@ class DuplexPipeline:
_TOOL_WAIT_TIMEOUT_SECONDS = 15.0
_SERVER_TOOL_TIMEOUT_SECONDS = 15.0
_DEFAULT_TOOL_SCHEMAS: Dict[str, Dict[str, Any]] = {
"search": {
"name": "search",
"description": "Search the internet for recent information",
"parameters": {
"type": "object",
"properties": {"query": {"type": "string"}},
"required": ["query"],
},
},
"calculator": {
"name": "calculator",
"description": "Evaluate a math expression",
"parameters": {
"type": "object",
"properties": {"expression": {"type": "string"}},
"required": ["expression"],
},
},
"weather": {
"name": "weather",
"description": "Get weather by city name",
"parameters": {
"type": "object",
"properties": {"city": {"type": "string"}},
"required": ["city"],
},
},
"translate": {
"name": "translate",
"description": "Translate text to target language",
"parameters": {
"type": "object",
"properties": {
"text": {"type": "string"},
"target_lang": {"type": "string"},
},
"required": ["text", "target_lang"],
},
},
"knowledge": {
"name": "knowledge",
"description": "Query knowledge base by question",
"parameters": {
"type": "object",
"properties": {
"query": {"type": "string"},
"kb_id": {"type": "string"},
},
"required": ["query"],
},
},
"current_time": {
"name": "current_time",
"description": "Get current local time",
@@ -120,51 +69,6 @@ class DuplexPipeline:
"required": [],
},
},
"code_interpreter": {
"name": "code_interpreter",
"description": "Execute Python code in a controlled environment",
"parameters": {
"type": "object",
"properties": {"code": {"type": "string"}},
"required": ["code"],
},
},
"turn_on_camera": {
"name": "turn_on_camera",
"description": "Turn on camera on client device",
"parameters": {
"type": "object",
"properties": {},
"required": [],
},
},
"turn_off_camera": {
"name": "turn_off_camera",
"description": "Turn off camera on client device",
"parameters": {
"type": "object",
"properties": {},
"required": [],
},
},
"increase_volume": {
"name": "increase_volume",
"description": "Increase speaker volume",
"parameters": {
"type": "object",
"properties": {"step": {"type": "integer"}},
"required": [],
},
},
"decrease_volume": {
"name": "decrease_volume",
"description": "Decrease speaker volume",
"parameters": {
"type": "object",
"properties": {"step": {"type": "integer"}},
"required": [],
},
},
}
def __init__(
@@ -386,6 +290,11 @@ class DuplexPipeline:
return False
return None
@staticmethod
def _is_openai_compatible_provider(provider: Any) -> bool:
normalized = str(provider or "").strip().lower()
return normalized in {"openai_compatible", "openai-compatible", "siliconflow"}
def _tts_output_enabled(self) -> bool:
enabled = self._coerce_bool(self._runtime_tts.get("enabled"))
if enabled is not None:
@@ -495,15 +404,15 @@ class DuplexPipeline:
tts_model = self._runtime_tts.get("model") or settings.siliconflow_tts_model
tts_speed = float(self._runtime_tts.get("speed") or settings.tts_speed)
if tts_provider == "siliconflow" and tts_api_key:
self.tts_service = SiliconFlowTTSService(
if self._is_openai_compatible_provider(tts_provider) and tts_api_key:
self.tts_service = OpenAICompatibleTTSService(
api_key=tts_api_key,
voice=tts_voice,
model=tts_model,
sample_rate=settings.sample_rate,
speed=tts_speed
)
logger.info("Using SiliconFlow TTS service")
logger.info("Using OpenAI-compatible TTS service (SiliconFlow implementation)")
else:
self.tts_service = EdgeTTSService(
voice=tts_voice,
@@ -531,8 +440,8 @@ class DuplexPipeline:
asr_interim_interval = int(self._runtime_asr.get("interimIntervalMs") or settings.asr_interim_interval_ms)
asr_min_audio_ms = int(self._runtime_asr.get("minAudioMs") or settings.asr_min_audio_ms)
if asr_provider == "siliconflow" and asr_api_key:
self.asr_service = SiliconFlowASRService(
if self._is_openai_compatible_provider(asr_provider) and asr_api_key:
self.asr_service = OpenAICompatibleASRService(
api_key=asr_api_key,
model=asr_model,
sample_rate=settings.sample_rate,
@@ -540,7 +449,7 @@ class DuplexPipeline:
min_audio_for_interim_ms=asr_min_audio_ms,
on_transcript=self._on_transcript_callback
)
logger.info("Using SiliconFlow ASR service")
logger.info("Using OpenAI-compatible ASR service (SiliconFlow implementation)")
else:
self.asr_service = BufferedASRService(
sample_rate=settings.sample_rate