@absolutejs/voice 0.0.22-beta.5 → 0.0.22-beta.500

package/dist/types.d.ts

@@ -1,24 +1,29 @@
-import type { SessionStore } from '@absolutejs/absolute';
-import type { VoiceOpsDispositionTaskPolicies, VoiceOpsTaskAssignmentRule, VoiceOpsTaskAssignmentRules, VoiceIntegrationWebhookConfig, StoredVoiceIntegrationEvent, StoredVoiceOpsTask, VoiceIntegrationEventStore, VoiceOpsTaskPolicy, VoiceOpsTask, VoiceOpsTaskStore } from './ops';
-import type { VoiceIntegrationSink } from './opsSinks';
-import type { StoredVoiceCallReviewArtifact, VoiceCallReviewArtifact, VoiceCallReviewStore } from './testing/review';
-import type { VoiceTraceEventStore } from './trace';
+import type { SessionStore } from "@absolutejs/absolute";
+import type { MediaWebRTCStatsCollector, MediaWebRTCStatsReport, MediaWebRTCStatsReportInput, MediaWebRTCStreamContinuityInput, MediaWebRTCStreamContinuityReport } from "@absolutejs/media";
+import type { VoiceOpsDispositionTaskPolicies, VoiceOpsTaskAssignmentRule, VoiceOpsTaskAssignmentRules, VoiceIntegrationWebhookConfig, StoredVoiceIntegrationEvent, StoredVoiceOpsTask, VoiceIntegrationEventStore, VoiceOpsTaskPolicy, VoiceOpsTask, VoiceOpsTaskStore } from "./ops";
+import type { VoiceIntegrationSink } from "./opsSinks";
+import type { StoredVoiceCallReviewArtifact, VoiceCallReviewArtifact, VoiceCallReviewStore } from "./testing/review";
+import type { VoiceTraceEventStore } from "./trace";
+import type { VoiceLiveOpsControlState } from "./liveOps";
+import type { VoiceAuditActor, VoiceAuditEventStore } from "./audit";
+import type { VoiceProfileSwitchGuardDecision, VoiceProfileSwitchGuardMode, VoiceProfileSwitchObservedSignals } from "./profileSwitchRecommendation";
+import type { VoiceRealCallProfileDefaultsReport, VoiceRealCallProfileHistoryReport } from "./proofTrends";
 export type AudioFormat = {
-    container: 'raw';
-    encoding: 'alaw' | 'mulaw' | 'pcm_s16le';
+    container: "raw";
+    encoding: "alaw" | "mulaw" | "pcm_s16le";
     sampleRateHz: number;
     channels: 1 | 2;
 };
 export type AudioChunk = ArrayBuffer | ArrayBufferView;
 export type VoiceLanguageStrategy = {
-    mode: 'auto-detect';
+    mode: "auto-detect";
     allowedLanguages?: string[];
 } | {
-    mode: 'fixed';
+    mode: "fixed";
     primaryLanguage: string;
     secondaryLanguages?: string[];
 } | {
-    mode: 'allow-switching';
+    mode: "allow-switching";
     primaryLanguage?: string;
     secondaryLanguages: string[];
 };
@@ -28,7 +33,7 @@ export type VoicePhraseHint = {
     boost?: number;
     metadata?: Record<string, unknown>;
 };
-export type VoiceCorrectionRiskTier = 'safe' | 'balanced' | 'risky';
+export type VoiceCorrectionRiskTier = "safe" | "balanced" | "risky";
 export type VoiceDomainTerm = {
     text: string;
     aliases?: string[];
@@ -44,12 +49,18 @@ export type VoiceLexiconEntry = {
     metadata?: Record<string, unknown>;
     pronunciation?: string;
 };
+export type VoiceTranscriptSentiment = {
+    label: "negative" | "neutral" | "positive" | (string & {});
+    metadata?: Record<string, unknown>;
+    score?: number;
+};
 export type Transcript = {
     id: string;
     text: string;
     isFinal: boolean;
     confidence?: number;
     language?: string;
+    sentiment?: VoiceTranscriptSentiment;
     speaker?: string | number;
     startedAtMs?: number;
     endedAtMs?: number;
@@ -65,7 +76,7 @@ export type VoiceTranscriptQuality = {
     fallback?: VoiceFallbackDiagnostics;
     partialTranscriptCount: number;
     selectedTranscriptCount: number;
-    source: 'fallback' | 'primary';
+    source: "fallback" | "primary";
 };
 export type VoiceTurnCorrectionDiagnostics = {
     attempted: boolean;
@@ -83,7 +94,7 @@ export type VoiceTurnCostEstimate = {
     primaryAudioMs: number;
     totalBillableAudioMs: number;
 };
-export type VoiceFallbackSelectionReason = 'fallback-empty' | 'primary-empty' | 'word-count-margin' | 'confidence-margin' | 'word-count-tiebreak' | 'kept-primary';
+export type VoiceFallbackSelectionReason = "fallback-empty" | "primary-empty" | "word-count-margin" | "confidence-margin" | "word-count-tiebreak" | "kept-primary";
 export type VoiceFallbackDiagnostics = {
     attempted: boolean;
     fallbackConfidence?: number;
@@ -94,31 +105,31 @@ export type VoiceFallbackDiagnostics = {
     primaryWordCount: number;
     selected: boolean;
     selectionReason: VoiceFallbackSelectionReason;
-    trigger: 'empty-turn' | 'low-confidence' | 'empty-or-low-confidence' | 'always';
+    trigger: "empty-turn" | "low-confidence" | "empty-or-low-confidence" | "always";
 };
 export type VoicePartialEvent = {
-    type: 'partial';
+    type: "partial";
     transcript: Transcript;
     receivedAt: number;
 };
 export type VoiceFinalEvent = {
-    type: 'final';
+    type: "final";
     transcript: Transcript;
     receivedAt: number;
 };
 export type VoiceEndOfTurnEvent = {
-    type: 'endOfTurn';
-    reason: 'vendor' | 'silence' | 'manual';
+    type: "endOfTurn";
+    reason: "vendor" | "silence" | "manual";
     receivedAt: number;
 };
 export type VoiceErrorEvent = {
-    type: 'error';
+    type: "error";
     error: Error;
     recoverable: boolean;
     code?: string;
 };
 export type VoiceCloseEvent = {
-    type: 'close';
+    type: "close";
     code?: number;
     reason?: string;
     recoverable?: boolean;
@@ -144,11 +155,11 @@ export type STTAdapterOpenOptions = {
     signal?: AbortSignal;
 };
 export type STTAdapter<TOptions extends STTAdapterOpenOptions = STTAdapterOpenOptions> = {
-    kind: 'stt';
+    kind: "stt";
     open: (options: TOptions) => Promise<STTAdapterSession> | STTAdapterSession;
 };
 export type TTSAudioEvent = {
-    type: 'audio';
+    type: "audio";
     chunk: AudioChunk;
     format: AudioFormat;
     receivedAt: number;
@@ -161,15 +172,26 @@ export type TTSSessionEventMap = {
 export type TTSAdapterSession = {
     on: <K extends keyof TTSSessionEventMap>(event: K, handler: (payload: TTSSessionEventMap[K]) => void | Promise<void>) => () => void;
     send: (text: string) => Promise<void>;
+    cancel?: (reason?: string) => Promise<void>;
     close: (reason?: string) => Promise<void>;
 };
+export declare const ttsAdapterSessionCanCancel: (session: TTSAdapterSession) => session is TTSAdapterSession & {
+    cancel: (reason?: string) => Promise<void>;
+};
+export type VoiceTTSProsody = {
+    emphasis?: number;
+    pitch?: number;
+    speed?: number;
+    style?: string;
+};
 export type TTSAdapterOpenOptions = {
     sessionId: string;
     lexicon?: VoiceLexiconEntry[];
+    prosody?: VoiceTTSProsody;
     signal?: AbortSignal;
 };
 export type TTSAdapter<TOptions extends TTSAdapterOpenOptions = TTSAdapterOpenOptions> = {
-    kind: 'tts';
+    kind: "tts";
     open: (options: TOptions) => Promise<TTSAdapterSession> | TTSAdapterSession;
 };
 export type RealtimeSessionEventMap = STTSessionEventMap & {
@@ -185,20 +207,33 @@ export type RealtimeAdapterOpenOptions = {
     format: AudioFormat;
     languageStrategy?: VoiceLanguageStrategy;
     lexicon?: VoiceLexiconEntry[];
+    modalities?: ReadonlyArray<"audio" | "text">;
     phraseHints?: VoicePhraseHint[];
+    promptCacheKey?: string;
+    semanticVAD?: import("./assistantMode").VoiceSemanticVADConfig;
     signal?: AbortSignal;
 };
 export type RealtimeAdapter<TOptions extends RealtimeAdapterOpenOptions = RealtimeAdapterOpenOptions> = {
-    kind: 'realtime';
+    kind: "realtime";
     open: (options: TOptions) => Promise<RealtimeAdapterSession> | RealtimeAdapterSession;
 };
-export type VoiceSessionStatus = 'active' | 'reconnecting' | 'completed' | 'failed';
+export type VoiceSessionStatus = "active" | "reconnecting" | "completed" | "failed";
+export type VoiceReconnectClientStatus = "idle" | "reconnecting" | "resumed" | "exhausted";
+export type VoiceReconnectClientState = {
+    attempts: number;
+    lastDisconnectAt?: number;
+    lastResumedAt?: number;
+    maxAttempts: number;
+    nextAttemptAt?: number;
+    status: VoiceReconnectClientStatus;
+};
 export type VoiceTurnRecord<TResult = unknown> = {
     id: string;
     text: string;
     quality?: VoiceTranscriptQuality;
     transcripts: Transcript[];
     assistantText?: string;
+    attachments?: import("./agent").VoiceAgentMessageAttachment[];
     committedAt: number;
     result?: TResult;
 };
@@ -253,10 +288,10 @@ export type VoiceSessionSummary = {
     status: VoiceSessionStatus;
     turnCount: number;
 };
-export type VoiceCallDisposition = 'completed' | 'transferred' | 'escalated' | 'voicemail' | 'no-answer' | 'failed' | 'closed';
+export type VoiceCallDisposition = "completed" | "transferred" | "escalated" | "voicemail" | "no-answer" | "failed" | "silence-timeout" | "closed";
 export type VoiceCallLifecycleEvent = {
     at: number;
-    type: 'start' | 'end' | 'transfer' | 'escalation' | 'voicemail' | 'no-answer';
+    type: "start" | "end" | "transfer" | "escalation" | "voicemail" | "no-answer";
     disposition?: VoiceCallDisposition;
     metadata?: Record<string, unknown>;
     reason?: string;
@@ -269,6 +304,65 @@ export type VoiceCallLifecycleState = {
     lastEventAt: number;
     startedAt: number;
 };
+export type VoiceHandoffAction = "escalate" | "no-answer" | "transfer" | "voicemail";
+export type VoiceHandoffStatus = "delivered" | "failed" | "skipped";
+export type VoiceHandoffResult = {
+    deliveredAt?: number;
+    deliveredTo?: string;
+    error?: string;
+    metadata?: Record<string, unknown>;
+    status: VoiceHandoffStatus;
+};
+export type VoiceHandoffDeliveryQueueStatus = VoiceHandoffStatus | "pending";
+export type StoredVoiceHandoffDelivery<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = {
+    action: VoiceHandoffAction;
+    context: TContext;
+    createdAt: number;
+    deliveredAt?: number;
+    deliveries?: Record<string, VoiceHandoffResult & {
+        adapterId: string;
+        adapterKind?: string;
+    }>;
+    deliveryAttempts?: number;
+    deliveryError?: string;
+    deliveryStatus: VoiceHandoffDeliveryQueueStatus;
+    id: string;
+    metadata?: Record<string, unknown>;
+    reason?: string;
+    result?: TResult;
+    session: TSession;
+    sessionId: string;
+    target?: string;
+    updatedAt: number;
+};
+export type VoiceHandoffDeliveryStore<TDelivery extends StoredVoiceHandoffDelivery = StoredVoiceHandoffDelivery> = {
+    get: (id: string) => Promise<TDelivery | undefined> | TDelivery | undefined;
+    list: () => Promise<TDelivery[]> | TDelivery[];
+    remove: (id: string) => Promise<void> | void;
+    set: (id: string, delivery: TDelivery) => Promise<void> | void;
+};
+export type VoiceHandoffInput<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = {
+    action: VoiceHandoffAction;
+    api: VoiceSessionHandle<TContext, TSession, TResult>;
+    context: TContext;
+    metadata?: Record<string, unknown>;
+    reason?: string;
+    result?: TResult;
+    session: TSession;
+    target?: string;
+};
+export type VoiceHandoffAdapter<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = {
+    actions?: VoiceHandoffAction[];
+    handoff: (input: VoiceHandoffInput<TContext, TSession, TResult>) => Promise<VoiceHandoffResult> | VoiceHandoffResult;
+    id: string;
+    kind?: string;
+};
+export type VoiceHandoffConfig<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = {
+    adapters: VoiceHandoffAdapter<TContext, TSession, TResult>[];
+    deliveryQueue?: VoiceHandoffDeliveryStore<StoredVoiceHandoffDelivery<TContext, TSession, TResult>>;
+    enqueueOnly?: boolean;
+    failMode?: "record" | "throw";
+};
 export type VoiceSessionStore<TSession extends VoiceSessionRecord = VoiceSessionRecord> = SessionStore<TSession, VoiceSessionSummary>;
 export type VoiceLogger = {
     debug?: (message: string, meta?: Record<string, unknown>) => void;
@@ -277,15 +371,15 @@ export type VoiceLogger = {
     error?: (message: string, meta?: Record<string, unknown>) => void;
 };
 export type VoiceReconnectConfig = {
-    strategy?: 'resume-last-turn' | 'restart' | 'fail';
+    strategy?: "resume-last-turn" | "restart" | "fail";
     timeout?: number;
     maxAttempts?: number;
 };
-export type VoiceRuntimePreset = 'default' | 'chat' | 'guided-intake' | 'dictation' | 'noisy-room' | 'pstn-balanced' | 'pstn-fast' | 'reliability';
-export type VoiceSTTLifecycle = 'continuous' | 'turn-scoped';
-export type VoiceTurnProfile = 'fast' | 'balanced' | 'long-form';
-export type VoiceTurnQualityProfile = 'general' | 'accent-heavy' | 'noisy-room' | 'short-command';
-export type VoiceTurnFallbackTrigger = 'empty-turn' | 'low-confidence' | 'empty-or-low-confidence' | 'always';
+export type VoiceRuntimePreset = "default" | "chat" | "guided-intake" | "dictation" | "noisy-room" | "pstn-balanced" | "pstn-fast" | "reliability";
+export type VoiceSTTLifecycle = "continuous" | "turn-scoped";
+export type VoiceTurnProfile = "fast" | "balanced" | "long-form";
+export type VoiceTurnQualityProfile = "general" | "accent-heavy" | "noisy-room" | "short-command";
+export type VoiceTurnFallbackTrigger = "empty-turn" | "low-confidence" | "empty-or-low-confidence" | "always";
 export type VoiceSTTFallbackConfig = {
     adapter: STTAdapter;
     trigger?: VoiceTurnFallbackTrigger;
@@ -338,11 +432,26 @@ export type VoiceSocket = {
     send: (data: string | Uint8Array | ArrayBuffer) => void | Promise<void>;
     close: (code?: number, reason?: string) => void | Promise<void>;
 };
+export type VoiceMonitorRuntimeSessionBinding = {
+    deregister: (reason?: string) => void;
+    emitAudio: (chunk: Uint8Array | ArrayBuffer, options?: {
+        source?: "assistant" | "caller" | (string & {});
+    }) => void;
+};
+export type VoiceMonitorRuntimeRegisterInput = {
+    handle: VoiceSessionHandle<unknown, VoiceSessionRecord, unknown>;
+    metadata?: Record<string, unknown>;
+    sessionId: string;
+};
+export type VoiceMonitorRuntimeBinding = {
+    registerSession: (input: VoiceMonitorRuntimeRegisterInput) => VoiceMonitorRuntimeSessionBinding;
+};
 export type VoiceSessionHandle<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = {
     id: string;
     connect: (socket: VoiceSocket) => Promise<void>;
     receiveAudio: (audio: AudioChunk) => Promise<void>;
-    commitTurn: (reason?: VoiceEndOfTurnEvent['reason']) => Promise<void>;
+    attachUserMedia: (attachment: import("./agent").VoiceAgentMessageAttachment) => Promise<void>;
+    commitTurn: (reason?: VoiceEndOfTurnEvent["reason"]) => Promise<void>;
     disconnect: (event?: VoiceCloseEvent) => Promise<void>;
     complete: (result?: TResult) => Promise<void>;
     escalate: (input: {
@@ -364,6 +473,7 @@ export type VoiceSessionHandle<TContext = unknown, TSession extends VoiceSession
         reason?: string;
         result?: TResult;
         target: string;
+        transferMode?: "cold" | "warm";
     }) => Promise<void>;
     close: (reason?: string) => Promise<void>;
     snapshot: () => Promise<TSession>;
@@ -376,6 +486,7 @@ export type VoiceRouteResult<TResult = unknown> = {
         metadata?: Record<string, unknown>;
         reason?: string;
         target: string;
+        transferMode?: "cold" | "warm";
     };
     escalate?: {
         metadata?: Record<string, unknown>;
@@ -416,6 +527,10 @@ export type VoiceLexiconResolver<TContext = unknown> = (input: {
 }) => Promise<VoiceLexiconEntry[] | void> | VoiceLexiconEntry[] | void;
 export type VoiceOnTurnObjectHandler<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = (input: {
     context: TContext;
+    liveOps?: {
+        control: VoiceLiveOpsControlState;
+        injectedInstruction?: string;
+    };
     session: TSession;
     turn: VoiceTurnRecord;
     api: VoiceSessionHandle<TContext, TSession, TResult>;
@@ -500,7 +615,7 @@ export type VoiceRuntimeOpsConfig<TContext = unknown, TSession extends VoiceSess
         disposition: VoiceCallDisposition;
         review: StoredVoiceCallReviewArtifact;
         session: TSession;
-    }) => Promise<Omit<VoiceOpsTask, 'id'> | VoiceOpsTask | StoredVoiceOpsTask | null | void> | Omit<VoiceOpsTask, 'id'> | VoiceOpsTask | StoredVoiceOpsTask | null | void;
+    }) => Promise<Omit<VoiceOpsTask, "id"> | VoiceOpsTask | StoredVoiceOpsTask | null | void> | Omit<VoiceOpsTask, "id"> | VoiceOpsTask | StoredVoiceOpsTask | null | void;
     resolveTaskPolicy?: (input: {
         api: VoiceSessionHandle<TContext, TSession, TResult>;
         context: TContext;
@@ -537,7 +652,37 @@ export type VoiceRuntimeOpsConfig<TContext = unknown, TSession extends VoiceSess
     tasks?: VoiceOpsTaskStore;
     webhook?: VoiceIntegrationWebhookConfig;
 };
-export type VoiceNormalizedRouteConfig<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = Omit<VoiceRouteConfig<TContext, TSession, TResult>, 'onTurn'> & {
+export type VoiceLiveOpsRuntimeConfig = {
+    getControl: (sessionId: string) => Promise<VoiceLiveOpsControlState | null | undefined> | VoiceLiveOpsControlState | null | undefined;
+};
+export type VoiceProfileSwitchGuardResolverInput<TContext = unknown> = {
+    context: TContext;
+    scenarioId?: string;
+    sessionId: string;
+};
+export type VoicePluginProfileSwitchGuardConfig<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = {
+    actor?: VoiceAuditActor;
+    allowedProfileIds?: string[] | ((input: VoiceProfileSwitchGuardResolverInput<TContext>) => Promise<string[] | undefined> | string[] | undefined);
+    audit?: VoiceAuditEventStore;
+    blockedProfileIds?: string[] | ((input: VoiceProfileSwitchGuardResolverInput<TContext>) => Promise<string[] | undefined> | string[] | undefined);
+    currentProfileId?: string | ((input: VoiceProfileSwitchGuardResolverInput<TContext>) => Promise<string | undefined> | string | undefined);
+    defaultProfileId?: string;
+    defaults: VoiceRealCallProfileDefaultsReport | VoiceRealCallProfileHistoryReport | ((input: VoiceProfileSwitchGuardResolverInput<TContext>) => Promise<VoiceRealCallProfileDefaultsReport | VoiceRealCallProfileHistoryReport> | VoiceRealCallProfileDefaultsReport | VoiceRealCallProfileHistoryReport);
+    metadata?: Record<string, unknown> | ((input: VoiceProfileSwitchGuardResolverInput<TContext>) => Promise<Record<string, unknown> | undefined> | Record<string, unknown> | undefined);
+    minConfidence?: number | ((input: VoiceProfileSwitchGuardResolverInput<TContext>) => Promise<number | undefined> | number | undefined);
+    maxAutoSwitchesPerSession?: number | ((input: VoiceProfileSwitchGuardResolverInput<TContext>) => Promise<number | undefined> | number | undefined);
+    mode?: VoiceProfileSwitchGuardMode | ((input: VoiceProfileSwitchGuardResolverInput<TContext>) => Promise<VoiceProfileSwitchGuardMode | undefined> | VoiceProfileSwitchGuardMode | undefined);
+    observed?: VoiceProfileSwitchObservedSignals | ((input: VoiceProfileSwitchGuardResolverInput<TContext>) => Promise<VoiceProfileSwitchObservedSignals | undefined> | VoiceProfileSwitchObservedSignals | undefined);
+    onDecision?: (input: {
+        context: TContext;
+        decision: VoiceProfileSwitchGuardDecision;
+        scenarioId?: string;
+        sessionId: string;
+    }) => Promise<void> | void;
+    sessionMetadataKey?: string | false;
+    trace?: false | VoiceTraceEventStore;
+};
+export type VoiceNormalizedRouteConfig<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = Omit<VoiceRouteConfig<TContext, TSession, TResult>, "onTurn"> & {
     onTurn: VoiceOnTurnObjectHandler<TContext, TSession, TResult>;
 };
 export type VoiceScenario = {
@@ -557,9 +702,11 @@ export type VoicePluginConfig<TContext = unknown, TSession extends VoiceSessionR
     lexicon?: VoiceLexiconEntry[] | VoiceLexiconResolver<TContext>;
     phraseHints?: VoicePhraseHint[] | VoicePhraseHintResolver<TContext>;
     preset?: VoiceRuntimePreset;
-    stt: STTAdapter;
+    stt?: STTAdapter;
     sttFallback?: VoiceSTTFallbackConfig;
     sttLifecycle?: VoiceSTTLifecycle;
+    realtime?: RealtimeAdapter;
+    realtimeInputFormat?: AudioFormat;
     tts?: TTSAdapter;
     session: VoiceSessionStore<NoInfer<TSession>>;
     reconnect?: VoiceReconnectConfig;
@@ -567,100 +714,177 @@ export type VoicePluginConfig<TContext = unknown, TSession extends VoiceSessionR
     audioConditioning?: VoiceAudioConditioningConfig;
     logger?: VoiceLogger;
     htmx?: boolean | VoiceHTMXConfig<TSession, NoInfer<TResult>>;
+    handoff?: VoiceHandoffConfig<TContext, TSession, TResult>;
     ops?: VoiceRuntimeOpsConfig<TContext, TSession, TResult>;
+    liveOps?: VoiceLiveOpsRuntimeConfig;
+    monitor?: VoiceMonitorRuntimeBinding;
+    profileSwitchGuard?: VoicePluginProfileSwitchGuardConfig<TContext, TSession, TResult>;
     trace?: VoiceTraceEventStore;
 } & VoiceRouteConfig<TContext, TSession, TResult>;
+export type VoiceSessionRecordingConfig = {
+    channels?: ReadonlyArray<"assistant" | "user">;
+    maxBytesPerChannel?: number;
+    store: import("./recordingStore").VoiceRecordingStore;
+    userInputFormat?: AudioFormat;
+};
 export type CreateVoiceSessionOptions<TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = {
     costTelemetry?: VoiceCostTelemetryConfig<TContext, TSession, TResult>;
     id: string;
     context: TContext;
     socket: VoiceSocket;
-    stt: STTAdapter;
+    stt?: STTAdapter;
+    realtime?: RealtimeAdapter;
+    realtimeInputFormat?: AudioFormat;
     tts?: TTSAdapter;
     languageStrategy?: VoiceLanguageStrategy;
     lexicon?: VoiceLexiconEntry[];
     sttFallback?: VoiceResolvedSTTFallbackConfig;
     store: VoiceSessionStore<TSession>;
     trace?: VoiceTraceEventStore;
+    recording?: VoiceSessionRecordingConfig;
+    callSilenceTimeoutMs?: number;
+    amd?: import("./amdDetector").VoiceAMDDetector<TContext, TSession, TResult>;
+    costAccountant?: import("./costAccounting").VoiceCostAccountant;
+    costTelephony?: {
+        provider?: string;
+    };
+    redact?: import("./redaction").VoiceTranscriptRedactor;
+    semanticTurnDetector?: import("./semanticTurn").VoiceSemanticTurnDetector;
+    assistantMode?: import("./assistantMode").VoiceAssistantMode;
+    modalities?: ReadonlyArray<"audio" | "text">;
+    prosody?: VoiceTTSProsody;
     reconnect: Required<VoiceReconnectConfig>;
     phraseHints?: VoicePhraseHint[];
+    sessionMetadata?: Record<string, unknown>;
     scenarioId?: string;
     sttLifecycle: VoiceSTTLifecycle;
     turnDetection: VoiceResolvedTurnDetectionConfig;
     audioConditioning?: VoiceResolvedAudioConditioningConfig;
+    handoff?: VoiceHandoffConfig<TContext, TSession, TResult>;
+    liveOps?: VoiceLiveOpsRuntimeConfig;
     route: VoiceNormalizedRouteConfig<TContext, TSession, TResult>;
     logger?: VoiceLogger;
 };
 export type CreateVoiceSession = <TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown>(options: CreateVoiceSessionOptions<TContext, TSession, TResult>) => VoiceSessionHandle<TContext, TSession, TResult>;
 export type VoiceClientStartMessage = {
-    type: 'start';
+    type: "start";
     sessionId?: string;
     scenarioId?: string;
 };
 export type VoiceClientEndTurnMessage = {
-    type: 'end_turn';
+    type: "end_turn";
 };
 export type VoiceClientCloseMessage = {
-    type: 'close';
+    type: "close";
+    reason?: string;
+};
+export type VoiceClientCallControlMessage = {
+    type: "call_control";
+    action: "complete" | "escalate" | "no-answer" | "transfer" | "voicemail";
+    metadata?: Record<string, unknown>;
     reason?: string;
+    target?: string;
 };
 export type VoiceClientPingMessage = {
-    type: 'ping';
+    type: "ping";
 };
-export type VoiceClientMessage = VoiceClientStartMessage | VoiceClientEndTurnMessage | VoiceClientCloseMessage | VoiceClientPingMessage;
+export type VoiceClientMessage = VoiceClientStartMessage | VoiceClientEndTurnMessage | VoiceClientCloseMessage | VoiceClientCallControlMessage | VoiceClientPingMessage;
 export type VoiceServerSessionMessage = {
-    type: 'session';
+    type: "session";
     sessionId: string;
     status: VoiceSessionStatus;
     scenarioId?: string;
+    sessionMetadata?: Record<string, unknown>;
+};
+export type VoiceServerReplayMessage<TResult = unknown> = {
+    type: "replay";
+    assistantTexts: string[];
+    call?: VoiceCallLifecycleState;
+    partial: string;
+    scenarioId?: string;
+    sessionId: string;
+    sessionMetadata?: Record<string, unknown>;
+    status: VoiceSessionStatus;
+    turns: VoiceTurnRecord<TResult>[];
 };
 export type VoiceServerPartialMessage = {
-    type: 'partial';
+    type: "partial";
     transcript: Transcript;
 };
 export type VoiceServerFinalMessage = {
-    type: 'final';
+    type: "final";
     transcript: Transcript;
 };
 export type VoiceServerTurnMessage<TResult = unknown> = {
-    type: 'turn';
+    type: "turn";
     turn: VoiceTurnRecord<TResult>;
 };
 export type VoiceServerAssistantMessage = {
-    type: 'assistant';
+    type: "assistant";
     text: string;
     turnId?: string;
 };
 export type VoiceServerAudioMessage = {
-    type: 'audio';
+    type: "audio";
     chunkBase64: string;
     format: AudioFormat;
     receivedAt: number;
     turnId?: string;
 };
 export type VoiceServerCompleteMessage = {
-    type: 'complete';
+    type: "complete";
+    sessionId: string;
+};
+export type VoiceServerCallLifecycleMessage = {
+    type: "call_lifecycle";
+    event: VoiceCallLifecycleEvent;
     sessionId: string;
 };
 export type VoiceServerErrorMessage = {
-    type: 'error';
+    type: "error";
     message: string;
     recoverable?: boolean;
 };
 export type VoiceServerPongMessage = {
-    type: 'pong';
+    type: "pong";
+};
+export type VoiceServerConnectionMessage = {
+    type: "connection";
+    reconnect: VoiceReconnectClientState;
 };
-export type VoiceServerMessage<TResult = unknown> = VoiceServerSessionMessage | VoiceServerPartialMessage | VoiceServerFinalMessage | VoiceServerTurnMessage<TResult> | VoiceServerAssistantMessage | VoiceServerAudioMessage | VoiceServerCompleteMessage | VoiceServerErrorMessage | VoiceServerPongMessage;
+export type VoiceServerMessage<TResult = unknown> = VoiceServerSessionMessage | VoiceServerReplayMessage<TResult> | VoiceServerPartialMessage | VoiceServerFinalMessage | VoiceServerTurnMessage<TResult> | VoiceServerAssistantMessage | VoiceServerAudioMessage | VoiceServerCallLifecycleMessage | VoiceServerCompleteMessage | VoiceServerErrorMessage | VoiceServerPongMessage | VoiceServerConnectionMessage;
 export type VoiceConnectionOptions = {
+    browserMedia?: false | VoiceBrowserMediaReporterOptions;
     protocols?: string[];
     scenarioId?: string;
     reconnect?: boolean;
+    reconnectReportPath?: string;
     maxReconnectAttempts?: number;
     pingInterval?: number;
     sessionId?: string;
 };
+export type VoiceBrowserMediaReportPayload = {
+    at: number;
+    continuity?: MediaWebRTCStreamContinuityReport;
+    report: MediaWebRTCStatsReport;
+    scenarioId?: string | null;
+    sessionId?: string | null;
+};
+export type VoiceBrowserMediaReporterOptions = Omit<MediaWebRTCStatsReportInput, "peerConnection"> & {
+    fetch?: typeof fetch;
+    getPeerConnection?: (() => MediaWebRTCStatsCollector | null | undefined) | (() => Promise<MediaWebRTCStatsCollector | null | undefined>);
+    getScenarioId?: () => string | null | undefined;
+    getSessionId?: () => string | null | undefined;
+    intervalMs?: number;
+    continuity?: false | Omit<MediaWebRTCStreamContinuityInput, "previousStats" | "stats">;
+    onError?: (error: unknown) => void;
+    onReport?: (payload: VoiceBrowserMediaReportPayload) => void;
+    path?: string;
+    peerConnection?: MediaWebRTCStatsCollector;
+};
 export type VoiceCaptureOptions = {
     channelCount?: 1 | 2;
+    onAudio?: (audio: Uint8Array | ArrayBuffer, sendAudio: (audio: Uint8Array | ArrayBuffer) => void) => void;
     onLevel?: (level: number) => void;
     sampleRateHz?: number;
 };
@@ -674,6 +898,46 @@ export type VoiceBargeInOptions = {
     enabled?: boolean;
     interruptOnPartial?: boolean;
     interruptThreshold?: number;
+    monitor?: VoiceBargeInMonitor;
+};
+export type VoiceBargeInTriggerReason = "input-level" | "manual-audio" | "manual-interrupt" | "partial-transcript";
+export type VoiceBargeInMonitorEvent = {
+    at: number;
+    id: string;
+    latencyMs?: number;
+    playbackStopLatencyMs?: number;
+    reason: VoiceBargeInTriggerReason;
+    sessionId?: string | null;
+    status: "requested" | "stopped" | "skipped";
+    thresholdMs?: number;
+};
+export type VoiceBargeInMonitorSnapshot = {
+    averageLatencyMs?: number;
+    events: VoiceBargeInMonitorEvent[];
+    failed: number;
+    lastEvent?: VoiceBargeInMonitorEvent;
+    passed: number;
+    status: "empty" | "fail" | "pass" | "warn";
+    thresholdMs: number;
+    total: number;
+};
+export type VoiceBargeInMonitor = {
+    getSnapshot: () => VoiceBargeInMonitorSnapshot;
+    recordRequested: (input: {
+        reason: VoiceBargeInTriggerReason;
+        sessionId?: string | null;
+    }) => VoiceBargeInMonitorEvent;
+    recordSkipped: (input: {
+        reason: VoiceBargeInTriggerReason;
+        sessionId?: string | null;
+    }) => VoiceBargeInMonitorEvent;
+    recordStopped: (input: {
+        latencyMs?: number;
+        playbackStopLatencyMs?: number;
+        reason: VoiceBargeInTriggerReason;
+        sessionId?: string | null;
+    }) => VoiceBargeInMonitorEvent;
+    subscribe: (subscriber: () => void) => () => void;
 };
 export type VoiceAudioPlayerOptions = {
     autoStart?: boolean;
@@ -684,10 +948,10 @@ export type VoiceDuplexControllerOptions = VoiceControllerOptions & {
     audioPlayer?: VoiceAudioPlayerOptions;
     bargeIn?: VoiceBargeInOptions;
 };
-export type VoiceSTTRoutingGoal = 'best' | 'low-cost';
-export type VoiceSTTRoutingCorrectionMode = 'generic' | 'none' | 'risky-turn';
+export type VoiceSTTRoutingGoal = "best" | "low-cost";
+export type VoiceSTTRoutingCorrectionMode = "generic" | "none" | "risky-turn";
 export type VoiceSTTRoutingStrategy = {
-    benchmarkSessionTarget: 'deepgram-corrected' | 'deepgram-flux';
+    benchmarkSessionTarget: "deepgram-corrected" | "deepgram-flux";
     correctionMode: VoiceSTTRoutingCorrectionMode;
     goal: VoiceSTTRoutingGoal;
     notes: string[];
@@ -701,7 +965,7 @@ export type VoiceHTMXRenderInput<TResult = unknown, TSession extends VoiceSessio
     result?: TResult;
     session?: TSession;
     sessionId?: string;
-    status: VoiceSessionStatus | 'idle';
+    status: VoiceSessionStatus | "idle";
     turnCount: number;
     turns: VoiceTurnRecord<TResult>[];
 };
@@ -728,9 +992,12 @@ export type VoiceHTMXOptions<TSession extends VoiceSessionRecord = VoiceSessionR
 };
 export type VoiceHTMXConfig<TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown> = VoiceHTMXRenderer<TSession, TResult> | VoiceHTMXOptions<TSession, TResult>;
 export type VoiceStreamState<TResult = unknown> = {
+    call: VoiceCallLifecycleState | null;
+    sessionMetadata: Record<string, unknown> | null;
     sessionId: string | null;
     scenarioId: string | null;
-    status: VoiceSessionStatus | 'idle';
+    status: VoiceSessionStatus | "idle";
+    reconnect: VoiceReconnectClientState;
     partial: string;
     turns: VoiceTurnRecord<TResult>[];
     assistantTexts: string[];
@@ -744,6 +1011,8 @@ export type VoiceStreamState<TResult = unknown> = {
     isConnected: boolean;
 };
 export type VoiceStream<TResult = unknown> = {
+    call: VoiceCallLifecycleState | null;
+    callControl: (message: Omit<VoiceClientCallControlMessage, "type">) => void;
     close: () => void;
     start: (input?: {
         scenarioId?: string;
@@ -755,10 +1024,13 @@ export type VoiceStream<TResult = unknown> = {
     getSnapshot: () => VoiceStreamState<TResult>;
     isConnected: boolean;
     partial: string;
+    reconnect: VoiceReconnectClientState;
     sendAudio: (audio: Uint8Array | ArrayBuffer) => void;
+    simulateDisconnect: () => void;
     sessionId: string | null;
+    sessionMetadata: Record<string, unknown> | null;
     scenarioId: string | null;
-    status: VoiceSessionStatus | 'idle';
+    status: VoiceSessionStatus | "idle";
     subscribe: (subscriber: () => void) => () => void;
     turns: VoiceTurnRecord<TResult>[];
     assistantTexts: string[];
@@ -784,7 +1056,7 @@ export type VoiceAudioPlayerState = {
     queuedChunkCount: number;
 };
 export type VoiceAudioPlayerSource = {
-    assistantAudio: VoiceStreamState['assistantAudio'];
+    assistantAudio: VoiceStreamState["assistantAudio"];
     subscribe: (subscriber: () => void) => () => void;
 };
 export type VoiceAudioPlayer = {
@@ -810,6 +1082,8 @@ export type VoiceBargeInBinding = {
 };
 export type VoiceController<TResult = unknown> = {
     bindHTMX: (options: VoiceHTMXBindingOptions) => () => void;
+    call: VoiceCallLifecycleState | null;
+    callControl: (message: Omit<VoiceClientCallControlMessage, "type">) => void;
     close: () => void;
     endTurn: () => void;
     start: (input?: {
@@ -822,12 +1096,15 @@ export type VoiceController<TResult = unknown> = {
     isConnected: boolean;
     isRecording: boolean;
     partial: string;
+    reconnect: VoiceReconnectClientState;
     recordingError: string | null;
     sendAudio: (audio: Uint8Array | ArrayBuffer) => void;
+    simulateDisconnect: () => void;
     sessionId: string | null;
+    sessionMetadata: Record<string, unknown> | null;
     scenarioId: string | null;
     startRecording: () => Promise<void>;
-    status: VoiceSessionStatus | 'idle';
+    status: VoiceSessionStatus | "idle";
     stopRecording: () => void;
     subscribe: (subscriber: () => void) => () => void;
     toggleRecording: () => Promise<void>;
@@ -851,36 +1128,54 @@ export type VoiceHTMXBindingOptions = {
     sessionQueryParam?: string;
 };
 export type VoiceStoreAction<TResult = unknown> = {
-    type: 'session';
+    type: "session";
     sessionId: string;
+    sessionMetadata?: Record<string, unknown>;
+    scenarioId?: string;
+    status: VoiceSessionStatus;
+} | {
+    type: "replay";
+    assistantTexts: string[];
+    call?: VoiceCallLifecycleState;
+    partial: string;
     scenarioId?: string;
+    sessionId: string;
+    sessionMetadata?: Record<string, unknown>;
     status: VoiceSessionStatus;
+    turns: VoiceTurnRecord<TResult>[];
 } | {
-    type: 'partial';
+    type: "call_lifecycle";
+    event: VoiceCallLifecycleEvent;
+    sessionId: string;
+} | {
+    type: "partial";
     transcript: Transcript;
 } | {
-    type: 'final';
+    type: "final";
     transcript: Transcript;
 } | {
-    type: 'turn';
+    type: "turn";
     turn: VoiceTurnRecord<TResult>;
 } | {
-    type: 'assistant';
+    type: "assistant";
     text: string;
 } | {
-    type: 'audio';
+    type: "audio";
     chunk: Uint8Array;
     format: AudioFormat;
     receivedAt: number;
     turnId?: string;
 } | {
-    type: 'complete';
+    type: "complete";
     sessionId: string;
 } | {
-    type: 'error';
+    type: "error";
     message: string;
 } | {
-    type: 'connected';
+    type: "connected";
+} | {
+    type: "connection";
+    reconnect: VoiceReconnectClientState;
 } | {
-    type: 'disconnected';
+    type: "disconnected";
 };