Pipeline

The Pipeline component manages the AI model and its configuration. It provides a standardized interface for different model providers.

Key Features:

• Flexible model selection
• Configurable model parameters
• Streaming audio support
• Multi-modal capabilities

Example Implementation:

from videosdk.agents import RealTimePipeline
from openai.types.beta.realtime.session import InputAudioTranscription, TurnDetection
from videosdk.plugins.openai import OpenAIRealtime, OpenAIRealtimeConfig

# For Gemini
# from videosdk.plugins.gemini import GeminiRealtime, GeminiLiveConfig
# from google.genai.types import AudioTranscriptionConfig

# Option 1: Using OpenAI's real-time models
openai_model = OpenAIRealtime(
    model="gpt-4o-realtime-preview",
    api_key="your-openai-api-key"  # Or use environment variable
)

# Option 2: Using Google's models
# gemini_model = GeminiRealtime(
#     model="gemini-pro-1.5",
#     api_key="your-google-api-key"
# )

# Configure the pipeline with your selected model
    model = OpenAIRealtime(
        model="gpt-4o-realtime-preview",
        config=OpenAIRealtimeConfig(
            modalities=["text", "audio"],
            input_audio_transcription=InputAudioTranscription(
                model="whisper-1"
            ),
            turn_detection=TurnDetection(
                type="server_vad",
                threshold=0.5,
                prefix_padding_ms=300,
                silence_duration_ms=200,
            ),
            tool_choice="auto"
        )
    )

    # For Gemini
    #     model = GeminiRealtime(
    #     model="gemini-2.0-flash-live-001",
    #     config=GeminiLiveConfig(
    #         response_modalities=["AUDIO"],
    #         output_audio_transcription=AudioTranscriptionConfig(
    #         )
    #     )
    # )

# Create the pipeline with the model and configuration
pipeline = RealTimePipeline(model=openai_model)