plugnmeet-server

Plug-N-Meet is an open-source, scalable web conferencing system built on the LiveKit WebRTC infrastructure. It is designed to be integrated into existing websites or applications via APIs and SDKs.

Key capabilities include:

• Adaptive Streaming: Supports H264, VP8, VP9, and AV1 codecs with Simulcast and Dynacast.
• Collaboration: Whiteboard (with office file support), shared notepad, polls, and breakout rooms.
• AI Intelligence: Real-time transcription, spoken translations, and automated summaries via the Artifacts API.
• Broadcasting: MP4 recording and RTMP/RTMPS/WHIP support.
• Security: WebRTC encryption by default with optional End-to-End Encryption (E2EE).

The system is composed of three primary architectural pieces:

1. plugNmeet-server: The Go-based backend responsible for all API requests and business logic.
2. plugNmeet-client: The React and Redux-based frontend user interface.
3. plugNmeet-recorder: A dedicated Go application used to manage MP4 recordings and RTMP broadcasting.

To install Plug-N-Meet manually, you must satisfy the following dependencies:

• LiveKit: A properly configured instance.
• Redis: Used for caching and messaging.
• MariaDB or MySQL: For persistent data storage.
• Optional: libreoffice and mupdf-tools if you require office file support (PDF, DOCX, PPTX) within the collaborative whiteboard.

You can run the server using Docker by providing a config.yaml file. Ensure you have created your config.yaml from the config_sample.yaml provided in the repository before running the command.

For a multi-container setup, refer to the docker-compose_sample.yaml file in the repository.

docker run --rm -p 8080:8080 \
    -v $PWD/config.yaml:/config.yaml \
    mynaparrot/plugnmeet-server \
    --config /config.yaml

The following official images are available on Docker Hub:

• mynaparrot/plugnmeet-server
• mynaparrot/plugnmeet-etherpad
• mynaparrot/plugnmeet-recorder

Plug-N-Meet provides several ways to integrate conferencing into your ecosystem:

Official SDKs

• PHP SDK
• JavaScript SDK (Supports NodeJS and Deno)

Ready-to-Use Plugins

• WordPress Plugin
• Moodle Plugin
• Joomla Component
• LTI (Learning Tools Interoperability) for compatible LMS

Mobile and Desktop

• Native Mobile App Integration: A hybrid model that allows you to combine a native media layer with the full plugNmeet web client UI for iOS, Android, Flutter, React Native, or desktop applications.

CreateTranscription initializes a real-time transcription stream using OpenAI's realtime API. It returns an insights.TranscriptionStream.

Parameters:

• ctx: The execution context.
• roomId: The unique identifier for the meeting room.
• userId: The unique identifier for the user.
• options: A JSON-encoded byte slice containing insights.TranscriptionOptions.

stream, err := provider.CreateTranscription(ctx, "room-123", "user-456", transcriptionOptionsJSON)
if err != nil {
	// handle error
}

AITextChatStream sends a prompt with conversation history and returns a channel that streams back the AI's response. This is useful for building interactive AI chat interfaces.

Parameters:

• ctx: The execution context.
• chatModel: The name of the OpenAI model to use (e.g., gpt-4o).
• history: A slice of *plugnmeet.InsightsAITextChatContent representing the conversation history.

resultsChan, err := provider.AITextChatStream(ctx, "gpt-4o", history)
if err != nil {
	// handle error
}
for result := range resultsChan {
	// Process each chunk of the streamed response
	fmt.Print(result.Content)
}

GetSupportedLanguages returns a list of supported languages for a given insights.ServiceType. This is used to populate UI elements like language selectors.

Parameters:

• serviceType: The type of service (e.g., translation or transcription) for which to retrieve supported languages.

SynthesizeText performs stateless text-to-speech (TTS) synthesis. It returns an io.ReadCloser which can be read to obtain the audio data.

Parameters:

• options: A JSON-encoded byte slice containing insights.SynthesisTaskOptions. The options object must include:
  • Text: The string to synthesize.
  • Language: The target language.
  • Voice: The specific OpenAI voice to use.

// options should be a JSON representation of insights.SynthesisTaskOptions
audioStream, err := provider.SynthesizeText(ctx, synthesisOptionsJSON)
if err != nil {
	// handle error
}
defer audioStream.Close()
// Read from audioStream to get the audio content

To use Azure services for transcription, translation, and text synthesis within Plug-N-Meet, you must initialize an AzureProvider using the NewProvider function. This requires a config.ProviderAccount (containing credentials like API Key and Region) and a config.ServiceConfig (containing service-specific options like the model name).

// Example initialization
provider, err := azure.NewProvider(providerAccount, serviceConfig, logger)
if err != nil {
    // handle error
}

Use NewProvider to create an instance of OpenAIProvider which implements the insights.Provider interface. This provider enables real-time transcription, text translation, text-to-speech synthesis, and AI-driven chat/summarization using OpenAI's services.

Requirements:

• A config.ProviderAccount containing a valid APIKey in its credentials.
• An optional endpoint option in the provider account to override the default OpenAI base URL (https://api.openai.com/v1).

Parameters:

• ctx: The execution context.
• providerAccount: Configuration containing the API key and optional endpoint.
• serviceConfig: The service-level configuration.
• log: A logrus.Entry for logging.
• redis: A *redis.Client for state management.

provider, err := openai.NewProvider(ctx, providerAccount, serviceConfig, logger, redisClient)
if err != nil {
	// handle error
}