Agora REST Client - Python

Python SDK for Agora REST APIs, including Conversational AI Agent management.

Features

• 🤖 High-level Agent API - Simple interface for managing Agora Conversational AI Agents
• 🔧 Configuration Management - Easy configuration from environment variables
• 🎯 Type-safe - Full type hints support
• 📦 Lightweight - Minimal dependencies
• 🚀 Production-ready - Built for reliability and performance

Installation

From PyPI (Recommended)

pip install agora-rest-client-python

From Git Repository

pip install git+https://github.com/your-org/agora-rest-client-python.git

For Development

git clone https://github.com/your-org/agora-rest-client-python.git
cd agora-rest-client-python
pip install -e .

Quick Start

1. Set up environment variables

Create a .env file:

# Agora Configuration
APP_ID=your_app_id
APP_CERTIFICATE=your_app_certificate

# Agora REST API Credentials
API_KEY=your_customer_id
API_SECRET=your_customer_secret

# LLM - OpenAI
LLM_API_KEY=your_openai_key

# ASR - Deepgram (Optional)
ASR_DEEPGRAM_API_KEY=your_deepgram_key

# TTS - ElevenLabs (Optional)
TTS_ELEVENLABS_API_KEY=your_elevenlabs_key

2. Initialize and use the Agent

import os
import uuid
import random
from agora_rest.agent import (
    AgentClient,
    TokenBuilder,
    DeepgramASRConfig,
    OpenAILLMConfig,
    ElevenLabsTTSConfig
)

# Create agent client
client = AgentClient(
    app_id=os.getenv("APP_ID"),
    app_certificate=os.getenv("APP_CERTIFICATE"),
    customer_id=os.getenv("API_KEY"),
    customer_secret=os.getenv("API_SECRET")
)

# Generate connection configuration
channel_name = f"channel_{uuid.uuid4().hex[:8]}"
user_uid = str(random.randint(100000, 999999))
agent_uid = str(random.randint(100000, 999999))

token = TokenBuilder.generate(
    app_id=os.getenv("APP_ID"),
    app_certificate=os.getenv("APP_CERTIFICATE"),
    channel_name=channel_name,
    uid=agent_uid
)

print(f"Channel: {channel_name}")
print(f"Token: {token}")

# Configure ASR, LLM, TTS - simple, only required fields
asr = DeepgramASRConfig(api_key=os.getenv("ASR_DEEPGRAM_API_KEY"))
llm = OpenAILLMConfig(api_key=os.getenv("LLM_API_KEY"))
tts = ElevenLabsTTSConfig(api_key=os.getenv("TTS_ELEVENLABS_API_KEY"))

# Optional: Customize configurations
# asr.language = "zh-CN"
# llm.model = "gpt-4o"
# tts.voice_id = "custom_voice_id"

# Alternative: Use dictionaries
# asr = {"vendor": "deepgram", "api_key": "xxx", "language": "zh-CN"}
# llm = {"api_key": "xxx", "model": "gpt-4o"}
# tts = {"vendor": "elevenlabs", "api_key": "xxx"}

# Advanced: Use custom vendors (Azure, Google, etc.)
# asr = {"vendor": "azure", "params": {"subscription_key": "xxx", "region": "eastus"}}
# tts = {"vendor": "azure", "params": {"subscription_key": "xxx", "voice_name": "zh-CN-XiaoxiaoNeural"}}

# Start an agent
result = client.start_agent(
    channel_name=channel_name,
    agent_uid=agent_uid,
    user_uid=user_uid,
    asr_config=asr,
    llm_config=llm,
    tts_config=tts
)

print(f"Agent started: {result['agent_id']}")

# Stop the agent
client.stop_agent(result['agent_id'])
print("Agent stopped")

Configuration

Environment Variables

Variable	Description	Required
APP_ID	Agora App ID	Yes
APP_CERTIFICATE	Agora App Certificate	Yes
API_KEY	Agora REST API Customer ID	Yes
API_SECRET	Agora REST API Customer Secret	Yes
LLM_API_KEY	OpenAI LLM API key	Yes
ASR_DEEPGRAM_API_KEY	Deepgram ASR API key	Optional
TTS_ELEVENLABS_API_KEY	ElevenLabs TTS API key	Optional
PORT	Server port (default: 8000)	Optional

API Reference

AgentClient

Core business logic for agent operations.

import os
import uuid
import random
from agora_rest.agent import AgentClient, TokenBuilder

client = AgentClient(
    app_id=os.getenv("APP_ID"),
    app_certificate=os.getenv("APP_CERTIFICATE"),
    customer_id=os.getenv("API_KEY"),
    customer_secret=os.getenv("API_SECRET")
)

# Generate connection configuration
channel_name = f"channel_{uuid.uuid4().hex[:8]}"
user_uid = str(random.randint(100000, 999999))
agent_uid = str(random.randint(100000, 999999))

token = TokenBuilder.generate(
    app_id=os.getenv("APP_ID"),
    app_certificate=os.getenv("APP_CERTIFICATE"),
    channel_name=channel_name,
    uid=agent_uid
)

# Start an agent
result = client.start_agent(
    channel_name=channel_name,
    agent_uid=agent_uid,
    user_uid=user_uid,
    asr_config={...},
    llm_config={...},
    tts_config={...}
)

# Stop an agent
client.stop_agent(agent_id)

Configuration Components

from agora_rest.agent import DeepgramASRConfig, OpenAILLMConfig, ElevenLabsTTSConfig

# ASR Configuration (Deepgram) - simple, only api_key required
asr = DeepgramASRConfig(api_key="your_deepgram_key")
# Optional: customize with defaults
# asr.language = "zh-CN"
# asr.model = "nova-3"

# LLM Configuration (OpenAI) - simple, only api_key required
llm = OpenAILLMConfig(api_key="your_openai_key")
# Optional: customize
# llm.model = "gpt-4o"
# llm.system_message = "You are a helpful assistant"
# llm.max_tokens = 512

# TTS Configuration (ElevenLabs) - simple, only api_key required
tts = ElevenLabsTTSConfig(api_key="your_elevenlabs_key")
# Optional: customize
# tts.voice_id = "custom_voice_id"
# tts.stability = 0.5

Supported Vendors

Built-in Vendors (Configuration Classes Provided)

All built-in vendors have dedicated configuration classes for easy use.

ASR (Automatic Speech Recognition)

• ✅ Deepgram - DeepgramASRConfig (Recommended for English)
• ✅ Fengming - FengmingASRConfig
• ✅ Tencent - TencentASRConfig
• ✅ Microsoft - MicrosoftASRConfig (Azure Speech)
• ✅ Ares - AresASRConfig

LLM (Large Language Model)

• ✅ OpenAI - OpenAILLMConfig (GPT-4, GPT-3.5, GPT-4o)
• ✅ Azure OpenAI - Use OpenAILLMConfig (Compatible with OpenAI API)
• ✅ Any OpenAI-compatible API - Use OpenAILLMConfig

TTS (Text-to-Speech)

• ✅ ElevenLabs - ElevenLabsTTSConfig (Recommended for quality)
• ✅ Minimax - MinimaxTTSConfig
• ✅ Tencent - TencentTTSConfig
• ✅ Bytedance - BytedanceTTSConfig
• ✅ Microsoft - MicrosoftTTSConfig (Azure TTS)
• ✅ Cartesia - CartesiaTTSConfig
• ✅ OpenAI - OpenAITTSConfig (OpenAI TTS)

Usage Examples

Using Built-in Vendors (Simple)

from agora_rest.agent import (
    DeepgramASRConfig,
    MicrosoftASRConfig,
    TencentASRConfig,
    OpenAILLMConfig,
    ElevenLabsTTSConfig,
    MicrosoftTTSConfig,
    OpenAITTSConfig,
)

# ASR - Deepgram (English) - simple, only api_key required
asr = DeepgramASRConfig(api_key="xxx")
# Optional: customize
# asr.language = "zh-CN"
# asr.model = "nova-3"

# ASR - Microsoft (Chinese)
asr = MicrosoftASRConfig(key="xxx", language="zh-CN")

# ASR - Tencent
asr = TencentASRConfig(
    key="xxx",
    app_id="xxx",
    secret="xxx"
)

# LLM - OpenAI
llm = OpenAILLMConfig(api_key="xxx")
# Optional: customize
# llm.model = "gpt-4o"

# TTS - ElevenLabs - simple, only api_key required
tts = ElevenLabsTTSConfig(api_key="xxx")
# Optional: customize
# tts.voice_id = "custom_voice_id"

# TTS - Microsoft (Chinese voice)
tts = MicrosoftTTSConfig(
    key="xxx",
    voice_name="zh-CN-XiaoxiaoNeural"
)

# TTS - OpenAI
tts = OpenAITTSConfig(api_key="xxx")
# Optional: customize
# tts.model = "tts-1-hd"
# tts.voice = "nova"

Using Custom Vendors (Advanced)

For vendors not listed above, use dictionary with params:

# Example: Custom ASR vendor
asr = {
    "vendor": "custom_asr",
    "params": {
        "api_key": "xxx",
        "custom_param": "value"
    }
}

# Example: Custom TTS vendor
tts = {
    "vendor": "custom_tts",
    "params": {
        "api_key": "xxx",
        "custom_param": "value"
    }
}

client.start_agent(..., asr_config=asr, tts_config=tts)

Examples

See the examples directory for more usage examples.

Requirements

• Python >= 3.8
• Dependencies listed in requirements.txt

Development

Setup Development Environment

# Clone repository
git clone https://github.com/your-org/agora-rest-client-python.git
cd agora-rest-client-python

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Install in editable mode
pip install -e .

Running Tests

pytest

Code Style

# Format code
black .

# Lint code
flake8 .

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Support

• 📧 Email: support@agora.io
• 📖 Documentation: Agora Docs
• 💬 Community: Agora Community

Changelog

See CHANGELOG.md for version history.