Voice library support dashscope

engine/agents/default.yaml

@@ -0,0 +1,62 @@
+# Agent behavior configuration (safe to edit per profile)
+# This file only controls agent-side behavior (VAD/LLM/TTS/ASR providers).
+# Infra/server/network settings should stay in .env.
+
+agent:
+  vad:
+    type: silero
+    model_path: data/vad/silero_vad.onnx
+    threshold: 0.5
+    min_speech_duration_ms: 100
+    eou_threshold_ms: 800
+
+  llm:
+    # provider: openai | openai_compatible | siliconflow
+    provider: openai_compatible
+    model: deepseek-v3
+    temperature: 0.7
+    # Required: no fallback. You can still reference env explicitly.
+    api_key: sk-fc4d59b360475f53401a864db8ce0985010acc4e696723d20a90d6569f38d80a
+    # Optional for OpenAI-compatible endpoints:
+    api_url: https://api.qnaigc.com/v1
+
+  tts:
+    # provider: edge | openai_compatible | siliconflow | dashscope
+    # dashscope defaults (if omitted):
+    provider: dashscope
+    api_url: wss://dashscope.aliyuncs.com/api-ws/v1/realtime
+    model: qwen3-tts-flash-realtime
+    api_key: sk-391f5126d18345d497c6e8717c8c9ad7
+    mode: commit
+    voice: Cherry
+    speed: 1.0
+    # provider: openai_compatible
+    # api_key: sk-thmzysdpqqmhqxxshyqoxvjeiflexjdgaftyufrsgrhpjnyx
+    # api_url: https://api.siliconflow.cn/v1/audio/speech
+    # model: FunAudioLLM/CosyVoice2-0.5B
+    # voice: anna
+    # speed: 1.0
+
+  asr:
+    # provider: buffered | openai_compatible | siliconflow
+    provider: openai_compatible
+    api_key: sk-thmzysdpqqmhqxxshyqoxvjeiflexjdgaftyufrsgrhpjnyx
+    api_url: https://api.siliconflow.cn/v1/audio/transcriptions
+    model: FunAudioLLM/SenseVoiceSmall
+    interim_interval_ms: 500
+    min_audio_ms: 300
+    start_min_speech_ms: 160
+    pre_speech_ms: 240
+    final_tail_ms: 120
+
+  duplex:
+    enabled: true
+    system_prompt: You are a helpful, friendly voice assistant. Keep your responses concise and conversational.
+
+  barge_in:
+    min_duration_ms: 200
+    silence_tolerance_ms: 60
+  
+  tools:
+    - calculator
+    - current_time

engine/agents/example.yaml

@@ -0,0 +1,55 @@
+# Agent behavior configuration (safe to edit per profile)
+# This file only controls agent-side behavior (VAD/LLM/TTS/ASR providers).
+# Infra/server/network settings should stay in .env.
+
+agent:
+  vad:
+    type: silero
+    model_path: data/vad/silero_vad.onnx
+    threshold: 0.5
+    min_speech_duration_ms: 100
+    eou_threshold_ms: 800
+
+  llm:
+    # provider: openai | openai_compatible | siliconflow
+    provider: openai_compatible
+    model: deepseek-v3
+    temperature: 0.7
+    # Required: no fallback. You can still reference env explicitly.
+    api_key: your_llm_api_key
+    # Optional for OpenAI-compatible endpoints:
+    api_url: https://api.qnaigc.com/v1
+
+  tts:
+    # provider: edge | openai_compatible | siliconflow | dashscope
+    # dashscope defaults (if omitted):
+    #   api_url: wss://dashscope.aliyuncs.com/api-ws/v1/realtime
+    #   model: qwen3-tts-flash-realtime
+    #   dashscope_mode: commit (engine splits) | server_commit (dashscope splits)
+    #   note: dashscope_mode/mode is ONLY used when provider=dashscope.
+    provider: openai_compatible
+    api_key: your_tts_api_key
+    api_url: https://api.siliconflow.cn/v1/audio/speech
+    model: FunAudioLLM/CosyVoice2-0.5B
+    voice: anna
+    speed: 1.0
+
+  asr:
+    # provider: buffered | openai_compatible | siliconflow
+    provider: openai_compatible
+    api_key: you_asr_api_key
+    api_url: https://api.siliconflow.cn/v1/audio/transcriptions
+    model: FunAudioLLM/SenseVoiceSmall
+    interim_interval_ms: 500
+    min_audio_ms: 300
+    start_min_speech_ms: 160
+    pre_speech_ms: 240
+    final_tail_ms: 120
+
+  duplex:
+    enabled: true
+    system_prompt: You are a helpful, friendly voice assistant. Keep your responses concise and conversational.
+
+  barge_in:
+    min_duration_ms: 200
+    silence_tolerance_ms: 60

engine/agents/tools.yaml

@@ -0,0 +1,78 @@
+# Agent behavior configuration with tool declarations.
+# This profile is an example only.
+
+agent:
+  vad:
+    type: silero
+    model_path: data/vad/silero_vad.onnx
+    threshold: 0.5
+    min_speech_duration_ms: 100
+    eou_threshold_ms: 800
+
+  llm:
+    # provider: openai | openai_compatible | siliconflow
+    provider: openai_compatible
+    model: deepseek-v3
+    temperature: 0.7
+    api_key: your_llm_api_key
+    api_url: https://api.qnaigc.com/v1
+
+  tts:
+    # provider: edge | openai_compatible | siliconflow | dashscope
+    # dashscope defaults (if omitted):
+    #   api_url: wss://dashscope.aliyuncs.com/api-ws/v1/realtime
+    #   model: qwen3-tts-flash-realtime
+    #   dashscope_mode: commit (engine splits) | server_commit (dashscope splits)
+    #   note: dashscope_mode/mode is ONLY used when provider=dashscope.
+    provider: openai_compatible
+    api_key: your_tts_api_key
+    api_url: https://api.siliconflow.cn/v1/audio/speech
+    model: FunAudioLLM/CosyVoice2-0.5B
+    voice: anna
+    speed: 1.0
+
+  asr:
+    # provider: buffered | openai_compatible | siliconflow
+    provider: openai_compatible
+    api_key: your_asr_api_key
+    api_url: https://api.siliconflow.cn/v1/audio/transcriptions
+    model: FunAudioLLM/SenseVoiceSmall
+    interim_interval_ms: 500
+    min_audio_ms: 300
+    start_min_speech_ms: 160
+    pre_speech_ms: 240
+    final_tail_ms: 120
+
+  duplex:
+    enabled: true
+    system_prompt: You are a helpful voice assistant with tool-calling support.
+
+  barge_in:
+    min_duration_ms: 200
+    silence_tolerance_ms: 60
+
+  # Tool declarations consumed by the engine at startup.
+  # - String form enables built-in/default tool schema when available.
+  # - Object form provides OpenAI function schema + executor hint.
+  tools:
+    - current_time
+    - calculator
+    - name: weather
+      description: Get weather by city name.
+      parameters:
+        type: object
+        properties:
+          city:
+            type: string
+            description: City name, for example "San Francisco".
+        required: [city]
+      executor: server
+    - name: open_map
+      description: Open map app on the client device.
+      parameters:
+        type: object
+        properties:
+          query:
+            type: string
+        required: [query]
+      executor: client