Interfaces

Configuration types, provider contracts, and event structures.

• AgentAudioConfig — Audio encoding and sample rate settings for Agent API input and output.
• AgentFunctionDefinition — Defines a function the agent can invoke during a conversation.
• AgentListenConfig — STT (speech-to-text) configuration for the Agent API listen stage.
• AgentSpeakConfig — TTS configuration for the Agent API speak stage.
• AgentThinkConfig — LLM configuration for the Agent API think stage.
• AnthropicLLMConfig — Configuration for the Anthropic LLM provider.
• AssemblyAISTTConfig — Configuration options for the AssemblyAISTT provider.
• AttachableInputProvider— An audio input provider that connects to its platform through a per-call handle passed to attach().
• AudioBufferQueueConfig — Configuration for an AudioBufferQueue instance.
• AudioChunk — A chunk of audio data flowing through the pipeline.
• AudioInputConfig — Configuration for audio input (microphone capture).
• AudioInputProvider — Audio input provider interface for the ‘input’ pipeline role.
• AudioMetadata — Metadata describing the format of audio data received from a provider.
• AudioOutputConfig — Configuration for audio output (playback).
• AudioOutputProvider — Audio output provider interface for the ‘output’ pipeline role.
• AwsCredentials — A static set of AWS credentials.
• AzureSTTConfig — Configuration options for the AzureSTT provider.
• AzureTTSConfig — Configuration for the AzureTTS provider.
• BaseProvider — Base interface that all providers must implement.
• BaseProviderConfig — Base configuration shared by all provider types.
• BlocklistOptions — Options for createBlocklistGuardrail.
• BrowserAudioOutputConfig — Configuration for BrowserAudioOutput.
• BufferInputOptions — Options controlling BufferInput behaviour beyond the audio format.
• CartesiaOutputFormat — Cartesia output format configuration passed in each WebSocket message.
• CartesiaTTSConfig — Configuration for the CartesiaTTS provider.
• CompositeVoiceConfig— Main configuration type for the CompositeVoice SDK.
• ConversationHistoryConfig — Configuration for multi-turn conversation history.
• DeepgramAgentConfig — Top-level configuration for the Deepgram Agent API provider.
• DeepgramFluxConfig — Configuration options for the DeepgramFlux provider.
• DeepgramFluxOptions — Deepgram Flux transcription options passed as query parameters on the V2 WebSocket connection URL.
• DeepgramSTTConfig — Configuration options for the DeepgramSTT provider.
• DeepgramTranscriptionOptions — Deepgram-specific transcription options passed as query parameters on the WebSocket connection URL.
• DeepgramTTSConfig — Configuration for the DeepgramTTS provider.
• DeepgramTTSOptions — Deepgram-specific TTS synthesis options.
• DiscordVoiceConfig — Configuration options for the DiscordVoice provider.
• DiscordVoiceConnection — Structural subset of @discordjs/voice’s VoiceConnection.
• EagerLLMConfig — Configuration for the eager LLM pipeline (speculative generation).
• ElevenLabsAgentConfig — Configuration for ElevenLabsAgent.
• ElevenLabsClientToolCall — A client tool call requested by the ElevenLabs agent.
• ElevenLabsSTTConfig — Configuration options for the ElevenLabs STT provider.
• ElevenLabsTTSConfig — Configuration for the ElevenLabsTTS provider.
• EventListenerMap — Typed mapping from event type strings to their corresponding listener signatures.
• EventStreamMessage — A decoded application/vnd.amazon.eventstream message.
• FallbackCapableProvider — A provider that can notify listeners when it swaps its underlying implementation (a fallback chain).
• FallbackSTTOptions — Options for FallbackSTT.
• FishAudioReference — An inline reference audio sample for instant voice cloning.
• FishAudioTTSConfig — Configuration for the FishAudioTTS provider.
• GeminiLiveAgentConfig — Configuration for GeminiLiveAgent.
• GeminiLiveFunctionCall — A function call requested by the Gemini Live agent.
• GeminiLiveFunctionDeclaration — A function declaration the Gemini Live agent can invoke.
• GeminiLLMConfig — Configuration for the Gemini LLM provider.
• GladiaSTTConfig — Configuration options for the GladiaSTT provider.
• GoogleMeetInputConfig — Configuration for GoogleMeetInput.
• GoogleMeetSessionStatus — Session status resource delivered on the session-control data channel.
• GoogleSTTConfig — Configuration for the GoogleSTT provider.
• GoogleSTTWordInfo — Word-level timing information from the Google Cloud STT response.
• GoogleTTSConfig — Configuration for the GoogleTTS provider.
• GroqLLMConfig — Configuration for the Groq LLM provider.
• Guardrail — A pluggable async filter applied to LLM text before it reaches TTS.
• GuardrailApplication — Record of one guardrail that rewrote or blocked text.
• GuardrailAppliedEvent — Emitted when a guardrail rewrites text on its way to the TTS provider.
• GuardrailBlockedEvent — Emitted when a guardrail suppresses text instead of speaking it.
• GuardrailContext — Context passed to Guardrail.check alongside the text to filter.
• GuardrailErrorEvent — Emitted when a guardrail throws or exceeds its timeout.
• GuardrailObserver — Hooks the pipeline calls as guardrails act, so the SDK can turn them into guardrail.* events.
• GuardrailOutcome — Result of running the chain over one piece of text.
• GuardrailResult — What a guardrail decided about the text it was given.
• GuardrailsConfig — Configuration for the guardrail filter chain.
• GuardrailStreamOptions — Per-utterance inputs a GuardrailStream needs to build its context.
• LLMGenerationOptions — Options for controlling a single LLM generation request.
• LLMMessage — A single message in an LLM conversation.
• LLMProvider — Large language model provider interface.
• LLMProviderConfig — Configuration for large language model providers.
• LLMToolCall — A tool invocation emitted by the LLM.
• LLMToolDefinition — Definition of a tool the LLM can invoke.
• LLMToolParameterSchema — Schema for a single tool parameter.
• LLMToolResult — Result sent back to the LLM after executing a tool.
• LMNTTTSConfig — Configuration for the LMNTTTS provider.
• LoggingConfig — Configuration for SDK logging output.
• MicrophoneInputConfig — Configuration for MicrophoneInput.
• MiniMaxTTSConfig — Configuration for the MiniMaxTTS provider.
• MistralLLMConfig — Configuration for the Mistral LLM provider.
• ModerationOptions — Options for createModerationGuardrail.
• ModerationVerdict — Verdict returned by a ModerationFn.
• MurfTTSConfig — Configuration for the MurfTTS provider.
• NativeSTTConfig — Configuration options for the NativeSTT provider.
• NativeTTSConfig — Configuration options for the NativeTTS provider.
• OpenAICompatibleLLMConfig — Configuration for any OpenAI-compatible LLM provider.
• OpenAILLMConfig — Configuration for the OpenAI LLM provider.
• OpenAIRealtimeAgentConfig — Configuration for OpenAIRealtimeAgent.
• OpenAIRealtimeSemanticVad — Semantic VAD turn-detection configuration for OpenAI Realtime transcription sessions.
• OpenAIRealtimeServerVad — Server VAD turn-detection configuration for OpenAI Realtime transcription sessions.
• OpenAIRealtimeSTTConfig — Configuration options for the OpenAIRealtimeSTT provider.
• OpenAITTSConfig — Configuration for the OpenAITTS provider.
• ParsedAudioMetadata — Audio format parameters recovered from a container header.
• PatternRedactionOptions — Options for createPatternRedactionGuardrail.
• PIIRedactionOptions — Options for createPIIRedactionGuardrail.
• PollyTTSConfig — Configuration for the PollyTTS provider.
• PresignAwsUrlOptions — Options for presignAwsUrl.
• PronunciationOptions — Options for createPronunciationGuardrail.
• ProviderFallbackEvent — Emitted when a provider fallback chain swaps the active provider.
• ProviderFallbackInfo — Notification payload emitted when a fallback chain swaps the active provider.
• QueueOverflowEvent — Emitted when an AudioBufferQueue drops chunks due to overflow.
• QueueStats — Statistics snapshot from an AudioBufferQueue instance.
• QueueStatsEvent — Emitted when queue statistics are requested via getQueueStats().
• RealtimeAgentFunctionCall — A function call requested by the Realtime agent.
• RealtimeAgentTool — A function tool the Realtime agent can invoke.
• RealtimeAgentTurnDetection — Turn-detection configuration for the Realtime session.
• ReconnectionConfig — Configuration for automatic WebSocket reconnection with exponential backoff.
• RecoveryEvent — Event payload emitted by the RecoveryOrchestrator during recovery attempts.
• RecoveryStrategy — Configuration for the recovery strategy applied by RecoveryOrchestrator.
• RedactionPattern — A named pattern and the text spoken in its place.
• ResolvedPipeline — A fully resolved 5-role pipeline with a provider assigned to each slot.
• RevAISTTConfig — Configuration options for the RevAISTT provider.
• RimeTTSConfig — Configuration for the RimeTTS provider.
• SignAwsRequestOptions — Options for signAwsRequestHeaders.
• SileroVADOptions — Options for SileroVAD.
• SmallestTTSConfig — Configuration for the SmallestTTS provider.
• SonioxSTTConfig — Configuration options for the SonioxSTT provider.
• SpeechifyTTSConfig — Configuration for the SpeechifyTTS provider.
• SpeechmaticsSTTConfig — Configuration options for the SpeechmaticsSTT provider.
• STTProviderConfig — Configuration for speech-to-text providers.
• TeamsCallConfig — Configuration for TeamsCall.
• TeamsTokenCredential — Minimal structural type for an ACS CommunicationTokenCredential.
• ToolAwareLLMProvider — LLM provider with optional tool use support.
• TranscribeSTTConfig — Configuration options for the TranscribeSTT provider.
• TranscriptionPreflightEvent — Emitted when a provider sends a preflight/eager-end-of-turn signal.
• TranscriptionResult — Result of a speech-to-text transcription.
• TranscriptionSpeechFinalEvent — Emitted when an utterance is fully complete.
• TTSProviderConfig — Configuration for text-to-speech providers.
• TurnDurations — Derived phase durations in milliseconds.
• TurnMetricsEvent — Emitted once per conversation turn with the turn’s latency breakdown.
• TurnMetricsSummary — One completed (or interrupted) conversation turn’s timing summary.
• TurnTakingConfig — Configuration for turn-taking behavior between the user and the agent.
• TurnTimestamps — Absolute timestamps (epoch ms) captured during one conversation turn.
• TwilioMediaStreamConfig — Configuration options for TwilioMediaStream.
• TwilioStreamSocket — Browser-style event subscription (socket.addEventListener(‘message’, cb)).
• VADBargeInEvent — Emitted when local VAD speech detection interrupted the agent.
• VADConfig — Configuration for local voice activity detection.
• VADEngine — Contract for a voice-activity-detection model.
• VADProcessorOptions — Tuning options for VADProcessor.
• VADSpeechEndEvent — Emitted when the local VAD detects the user has stopped speaking.
• VADSpeechEndInfo — Payload delivered when the processor detects the end of user speech.
• VADSpeechStartEvent — Emitted when the local VAD confirms the user has started speaking.
• VADSpeechStartInfo — Payload delivered when the processor detects the start of user speech.
• VonageAudioSocketConfig — Configuration options for VonageAudioSocket.
• VonageSocket — Browser-style event subscription (socket.addEventListener(‘message’, cb)).
• WebLLMLLMConfig — Configuration for the WebLLM in-browser LLM provider.
• WebLLMLoadProgress — Progress information emitted during model download and WebGPU shader compilation.
• WebRTCInputConfig — Configuration for WebRTCInput.
• WebRTCOutputConfig — Configuration for WebRTCOutput.
• ZoomRtmsInputConfig — Configuration options for the ZoomRtmsInput provider.
• ZoomRtmsSession — Session parameters delivered by the meeting.rtms_started webhook.