@ariaflowagents/gemini-native-audio 0.9.0

package/dist/types.d.ts

@@ -0,0 +1,150 @@
+import type { AgentDefinition, Foundation } from '@ariaflowagents/core/foundation';
+import type { AutoRetrieveProvider, FlowConfig, HarnessHooks, ToolSet } from '@ariaflowagents/core/types';
+import type { LivePromptAssembler } from '@ariaflowagents/core/capabilities';
+import type { MemoryService } from '@ariaflowagents/core/memory';
+import type { RealtimeAudioClient } from '@ariaflowagents/core/realtime';
+import type { LanguageModel } from 'ai';
+/**
+ * Voice tools use the standard AI SDK ToolSet type.
+ * Define tools using `tool()` from `ai` with `inputSchema` and `execute`.
+ *
+ * All voice tools MUST have an `execute` function. Schema-only tools
+ * (tools without execute) are not supported in voice agents.
+ */
+export type VoiceToolSet = ToolSet;
+/** @deprecated Use `ToolSet[string]` or define tools with `tool()` from `ai`. */
+export type VoiceToolDef = ToolSet[string];
+/**
+ * Voice agent configuration — supports both flat tool agents (v1) and
+ * flow-aware agents (v2) when `flow` is provided.
+ */
+export interface VoiceAgentConfig extends AgentDefinition {
+    tools?: ToolSet;
+    /** Gemini Live voice preset. */
+    voice?: string;
+    /** Optional flow config. When set, VoiceEngine uses CapabilityCallWorker. */
+    flow?: FlowConfig;
+    /** The initial node ID when `flow` is set. Defaults to the first node in the flow. */
+    initialNode?: string;
+}
+/**
+ * Shared interface implemented by both CallWorker and CapabilityCallWorker.
+ * VoiceEngine holds workers through this interface.
+ */
+export interface WorkerLike {
+    readonly callId: string;
+    start(): Promise<void>;
+    stop(): Promise<void>;
+}
+/** Gemini model/API configuration. */
+export interface GeminiConfig {
+    apiKey: string;
+    model?: string;
+}
+/** Top-level VoiceEngine configuration. */
+export interface VoiceEngineConfig {
+    foundation: Foundation;
+    agents: VoiceAgentConfig[];
+    defaultAgentId: string;
+    /**
+     * Gemini API configuration. Required when using the default Gemini provider.
+     * Optional when a custom `createModelClient` factory is provided.
+     */
+    gemini?: GeminiConfig;
+    /**
+     * Custom factory for creating a RealtimeAudioClient per call.
+     * When provided, this factory is used instead of creating a GeminiLiveSession.
+     * This is the provider abstraction point — use it to plug in OpenAI Realtime,
+     * a custom WebSocket client, or any other provider.
+     *
+     * Example (OpenAI):
+     * ```typescript
+     * createModelClient: (agent) => new OpenAIRealtimeClient({ apiKey: '...' })
+     * ```
+     */
+    createModelClient?: (agent: VoiceAgentConfig) => RealtimeAudioClient;
+    /**
+     * Optional LivePromptAssembler (port) for building audio-optimized prompts.
+     * Used by the core RealtimeRuntime via OrchestrationAuthority.
+     */
+    promptAssembler?: LivePromptAssembler;
+    /**
+     * Optional MemoryService for cross-session long-term memory.
+     * When provided, the authority handles memory preloading and ingestion.
+     */
+    memoryService?: MemoryService;
+    /**
+     * Hook callbacks for lifecycle events (onStart, onEnd, onToolResult, etc.).
+     * Passed to the OrchestrationAuthority for full hook parity with text Runtime.
+     */
+    hooks?: HarnessHooks;
+    /**
+     * Controls automatic memory ingestion behavior.
+     * - 'onEnd': Ingest after session close (default when memoryService is set)
+     * - 'manual': Developer must call memoryService.addSessionToMemory() explicitly
+     * - 'hook': Fires onMemoryIngest hook
+     */
+    memoryIngestion?: 'onEnd' | 'manual' | 'hook';
+    /**
+     * Optional dedicated text model used to verify extraction-node data in
+     * realtime sessions. When omitted, the core authority stays in conservative
+     * fallback mode rather than claiming full extraction parity.
+     */
+    extractionModel?: LanguageModel;
+    /**
+     * Optional auto-retrieve for realtime sessions. Passed to DefaultOrchestrationAuthority
+     * so each turn can refresh `knowledge:realtime` in session working memory.
+     */
+    autoRetrieveProvider?: AutoRetrieveProvider;
+}
+/** Transport session abstraction for audio I/O. */
+export interface TransportSession {
+    /** Send audio frames to the client. */
+    sendAudio(data: Uint8Array): void;
+    /** Receive audio frames from the client. */
+    onAudio(handler: (data: Uint8Array) => void): void;
+    /** Called when the session ends. */
+    onClose(handler: () => void): void;
+    /** Close the transport. */
+    close(): void;
+}
+/** Parameters for accepting a new call. */
+export interface AcceptCallParams {
+    callId: string;
+    sessionId?: string;
+    userId?: string;
+    transport: TransportSession;
+    agentId?: string;
+}
+/**
+ * Realtime events emitted by the Gemini Live session.
+ */
+export type RealtimeEvent = {
+    type: 'audio';
+    data: Uint8Array;
+} | {
+    type: 'transcript';
+    text: string;
+    role: 'user' | 'assistant';
+} | {
+    type: 'tool-call';
+    id: string;
+    name: string;
+    args: unknown;
+} | {
+    type: 'tool-result';
+    id: string;
+    name: string;
+    result: unknown;
+} | {
+    type: 'interrupted';
+} | {
+    type: 'turn-complete';
+} | {
+    type: 'session-resumed';
+    newHandle: string;
+} | {
+    type: 'error';
+    error: string;
+};
+//# sourceMappingURL=types.d.ts.map

package/dist/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,eAAe,EAAE,UAAU,EAAE,MAAM,iCAAiC,CAAC;AACnF,OAAO,KAAK,EAAE,oBAAoB,EAAE,UAAU,EAAE,YAAY,EAAE,OAAO,EAAE,MAAM,4BAA4B,CAAC;AAC1G,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,mCAAmC,CAAC;AAC7E,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAC;AACjE,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,+BAA+B,CAAC;AACzE,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,IAAI,CAAC;AAExC;;;;;;GAMG;AACH,MAAM,MAAM,YAAY,GAAG,OAAO,CAAC;AAEnC,iFAAiF;AACjF,MAAM,MAAM,YAAY,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC;AAE3C;;;GAGG;AACH,MAAM,WAAW,gBAAiB,SAAQ,eAAe;IACvD,KAAK,CAAC,EAAE,OAAO,CAAC;IAChB,gCAAgC;IAChC,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,6EAA6E;IAC7E,IAAI,CAAC,EAAE,UAAU,CAAC;IAClB,sFAAsF;IACtF,WAAW,CAAC,EAAE,MAAM,CAAC;CACtB;AAED;;;GAGG;AACH,MAAM,WAAW,UAAU;IACzB,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IACxB,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC;IACvB,IAAI,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC;CACvB;AAED,sCAAsC;AACtC,MAAM,WAAW,YAAY;IAC3B,MAAM,EAAE,MAAM,CAAC;IACf,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,2CAA2C;AAC3C,MAAM,WAAW,iBAAiB;IAChC,UAAU,EAAE,UAAU,CAAC;IACvB,MAAM,EAAE,gBAAgB,EAAE,CAAC;IAC3B,cAAc,EAAE,MAAM,CAAC;IACvB;;;OAGG;IACH,MAAM,CAAC,EAAE,YAAY,CAAC;IACtB;;;;;;;;;;OAUG;IACH,iBAAiB,CAAC,EAAE,CAAC,KAAK,EAAE,gBAAgB,KAAK,mBAAmB,CAAC;IACrE;;;OAGG;IACH,eAAe,CAAC,EAAE,mBAAmB,CAAC;IACtC;;;OAGG;IACH,aAAa,CAAC,EAAE,aAAa,CAAC;IAC9B;;;OAGG;IACH,KAAK,CAAC,EAAE,YAAY,CAAC;IACrB;;;;;OAKG;IACH,eAAe,CAAC,EAAE,OAAO,GAAG,QAAQ,GAAG,MAAM,CAAC;IAC9C;;;;OAIG;IACH,eAAe,CAAC,EAAE,aAAa,CAAC;IAChC;;;OAGG;IACH,oBAAoB,CAAC,EAAE,oBAAoB,CAAC;CAC7C;AAED,mDAAmD;AACnD,MAAM,WAAW,gBAAgB;IAC/B,uCAAuC;IACvC,SAAS,CAAC,IAAI,EAAE,UAAU,GAAG,IAAI,CAAC;IAClC,4CAA4C;IAC5C,OAAO,CAAC,OAAO,EAAE,CAAC,IAAI,EAAE,UAAU,KAAK,IAAI,GAAG,IAAI,CAAC;IACnD,oCAAoC;IACpC,OAAO,CAAC,OAAO,EAAE,MAAM,IAAI,GAAG,IAAI,CAAC;IACnC,2BAA2B;IAC3B,KAAK,IAAI,IAAI,CAAC;CACf;AAED,2CAA2C;AAC3C,MAAM,WAAW,gBAAgB;IAC/B,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,SAAS,EAAE,gBAAgB,CAAC;IAC5B,OAAO,CAAC,EAAE,MAAM,CAAC;CAClB;AAED;;GAEG;AACH,MAAM,MAAM,aAAa,GACrB;IAAE,IAAI,EAAE,OAAO,CAAC;IAAC,IAAI,EAAE,UAAU,CAAA;CAAE,GACnC;IAAE,IAAI,EAAE,YAAY,CAAC;IAAC,IAAI,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,GAAG,WAAW,CAAA;CAAE,GAChE;IAAE,IAAI,EAAE,WAAW,CAAC;IAAC,EAAE,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,OAAO,CAAA;CAAE,GAC9D;IAAE,IAAI,EAAE,aAAa,CAAC;IAAC,EAAE,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAC;IAAC,MAAM,EAAE,OAAO,CAAA;CAAE,GAClE;IAAE,IAAI,EAAE,aAAa,CAAA;CAAE,GACvB;IAAE,IAAI,EAAE,eAAe,CAAA;CAAE,GACzB;IAAE,IAAI,EAAE,iBAAiB,CAAC;IAAC,SAAS,EAAE,MAAM,CAAA;CAAE,GAC9C;IAAE,IAAI,EAAE,OAAO,CAAC;IAAC,KAAK,EAAE,MAAM,CAAA;CAAE,CAAC"}

package/dist/types.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=types.js.map

package/dist/types.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":""}

package/package.json

@@ -0,0 +1,44 @@
+{
+  "name": "@ariaflowagents/gemini-native-audio",
+  "version": "0.9.0",
+  "description": "Gemini Live Native Audio integration for AriaFlow — speech-to-speech voice agents with built-in STT/TTS",
+  "publishConfig": {
+    "access": "public"
+  },
+  "files": [
+    "dist",
+    "README.md"
+  ],
+  "type": "module",
+  "main": "dist/index.js",
+  "types": "dist/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "default": "./dist/index.js"
+    }
+  },
+  "scripts": {
+    "build": "tsc -p tsconfig.json",
+    "clean": "rm -rf dist",
+    "test": "npm run build && node --test test/*.test.js"
+  },
+  "dependencies": {
+    "@google/genai": "^1.0.0",
+    "ws": "^8.19.0",
+    "zod-to-json-schema": "^3.24.0"
+  },
+  "peerDependencies": {
+    "@ariaflowagents/core": "workspace:*",
+    "ai": "^6.0.0",
+    "zod": "^3.0.0"
+  },
+  "devDependencies": {
+    "@ariaflowagents/core": "workspace:*",
+    "@types/node": "^20.11.0",
+    "@types/ws": "^8.5.13",
+    "ai": "^6.0.0",
+    "typescript": "^5.3.0",
+    "zod": "^3.23.0"
+  }
+}