Files

Xin Wang 00b88c5afa Add manual opener tool calls to Assistant model and API

- Introduced `manual_opener_tool_calls` field in the Assistant model to support custom tool calls.
- Updated AssistantBase and AssistantUpdate schemas to include the new field.
- Implemented normalization and migration logic for handling manual opener tool calls in the API.
- Enhanced runtime metadata to include manual opener tool calls in responses.
- Updated tests to validate the new functionality and ensure proper handling of tool calls.
- Refactored tool ID normalization to support legacy tool names for backward compatibility.

2026-03-02 12:34:42 +08:00

app

Enhance WebSocket session configuration by introducing an optional config.resolved event, which provides a public snapshot of the session's configuration. Update the API reference documentation to clarify the conditions under which this event is emitted and the details it includes. Modify session management to respect the new setting for emitting configuration details, ensuring sensitive information remains secure. Update tests to validate the new behavior and ensure compliance with the updated configuration schema.

2026-03-01 23:08:44 +08:00

config/agents

add dashscope tts

2026-02-26 03:02:48 +08:00

core

Add manual opener tool calls to Assistant model and API

2026-03-02 12:34:42 +08:00

data

Add backend api and engine

2026-02-06 14:01:34 +08:00

docs

2026-03-01 23:08:44 +08:00

examples

Enhance WebSocket session management by requiring assistant_id as a query parameter for connection. Update API reference documentation to reflect changes in message flow and metadata validation rules, including the introduction of whitelists for allowed metadata fields and restrictions on sensitive keys. Refactor client examples to align with the new session initiation process.

2026-03-01 14:10:38 +08:00

models

2026-03-01 14:10:38 +08:00

processors

Add ASR capture timeout handling in DuplexPipeline and enhance EOU detection logic. Introduce _ASR_CAPTURE_MAX_MS constant and manage capture state timing to ensure timely end of utterance processing, even during silence. Update EouDetector to allow silence-only EOU when VAD state is lost.

2026-02-27 09:59:54 +08:00

scripts

Add backend api and engine

2026-02-06 14:01:34 +08:00

services

Implement API URL resolution in OpenAICompatibleASRService to ensure correct endpoint handling for transcription requests.

2026-02-26 12:04:59 +08:00

tests

Add manual opener tool calls to Assistant model and API

2026-03-02 12:34:42 +08:00

utils

Add backend api and engine

2026-02-06 14:01:34 +08:00

.env.example

Refactor WebSocket authentication handling by removing auth requirements from the hello message. Update related documentation and schemas to reflect the changes in authentication strategy, simplifying the connection process.

2026-02-28 17:33:40 +08:00

.gitignore

Unify db api

2026-02-26 01:58:39 +08:00

pyproject.toml

Add backend api and engine

2026-02-06 14:01:34 +08:00

README.md

add dashscope tts

2026-02-26 03:02:48 +08:00

requirements.txt

add dashscope tts

2026-02-26 03:02:48 +08:00

README.md

py-active-call-cc

Python Active-Call: real-time audio streaming with WebSocket and WebRTC.

This repo contains a Python 3.11+ codebase for building low-latency voice pipelines (capture, stream, and process audio) using WebRTC and WebSockets. It is currently in an early, experimental stage.

Usage

启动

uvicorn app.main:app --reload --host 0.0.0.0 --port 8000

使用 agent profile（推荐）

python -m app.main --agent-profile default

使用指定 YAML

python -m app.main --agent-config config/agents/default.yaml

Agent 配置路径优先级

--agent-config
--agent-profile（映射到 config/agents/<profile>.yaml）
AGENT_CONFIG_PATH
AGENT_PROFILE
config/agents/default.yaml（若存在）

说明

Agent 相关配置是严格模式：YAML 缺少必须项会直接报错，不会回退到 .env 或代码默认值。
如果要引用环境变量，请在 YAML 显式写 ${ENV_VAR}。
siliconflow 独立 section 已移除；请在 agent.llm / agent.tts / agent.asr 内通过 provider、api_key、api_url、model 配置。
agent.tts.provider 现支持 dashscope（Realtime 协议，非 OpenAI-compatible）；默认 URL 为 wss://dashscope.aliyuncs.com/api-ws/v1/realtime，默认模型为 qwen3-tts-flash-realtime。
agent.tts.dashscope_mode（兼容旧写法 agent.tts.mode）支持 commit | server_commit，且仅在 provider=dashscope 时生效：
- commit：Engine 先按句切分，再逐句提交给 DashScope。
- server_commit：Engine 不再逐句切分，由 DashScope 对整段文本自行切分。
现在支持在 Agent YAML 中配置 agent.tools（列表），用于声明运行时可调用工具。
工具配置示例见 config/agents/tools.yaml。

Backend Integration

Engine runtime now supports adapter-based backend integration:

BACKEND_MODE=auto|http|disabled
BACKEND_URL + BACKEND_TIMEOUT_SEC
HISTORY_ENABLED=true|false

Behavior:

auto: use HTTP backend only when BACKEND_URL is set, otherwise engine-only mode.
http: force HTTP backend; falls back to engine-only mode when URL is missing.
disabled: force engine-only mode (no backend calls).

History write path is now asynchronous and buffered per session:

HISTORY_QUEUE_MAX_SIZE
HISTORY_RETRY_MAX_ATTEMPTS
HISTORY_RETRY_BACKOFF_SEC
HISTORY_FINALIZE_DRAIN_TIMEOUT_SEC

This keeps turn processing responsive even when backend history APIs are slow/failing.

Detailed notes: docs/backend_integration.md.

测试

python examples/test_websocket.py

python mic_client.py

WS Protocol

/ws uses a strict v1 JSON control protocol with binary PCM audio frames.

See docs/ws_v1_schema.md.

README.md Unescape Escape

py-active-call-cc

Usage

Backend Integration

WS Protocol

README.md