@absolutejs/voice 0.0.22-beta.475 → 0.0.22-beta.476

package/dist/aiVoiceModel.d.ts

@@ -0,0 +1,10 @@
+import type { AIProviderConfig } from "@absolutejs/ai";
+import type { VoiceAgentModel } from "./agent";
+import type { VoiceSessionRecord } from "./types";
+export type CreateAIVoiceModelOptions = {
+    model: string;
+    provider: AIProviderConfig;
+    signal?: AbortSignal;
+    systemPrompt?: string;
+};
+export declare const createAIVoiceModel: <TContext = unknown, TSession extends VoiceSessionRecord = VoiceSessionRecord, TResult = unknown>(options: CreateAIVoiceModelOptions) => VoiceAgentModel<TContext, TSession, TResult>;

package/dist/index.d.ts

@@ -69,6 +69,8 @@ export { assertVoiceSimulationSuiteEvidence, createVoiceSimulationSuiteRoutes, e
 export { createVoiceWorkflowContract, createVoiceWorkflowContractHandler, createVoiceWorkflowContractPreset, createVoiceWorkflowScenario, recordVoiceWorkflowContractTrace, validateVoiceWorkflowRouteResult, } from "./workflowContract";
 export { createVoiceSessionListRoutes, createVoiceSessionReplayHTMLHandler, createVoiceSessionReplayJSONHandler, createVoiceSessionReplayRoutes, createVoiceSessionsHTMLHandler, createVoiceSessionsJSONHandler, renderVoiceSessionsHTML, summarizeVoiceProviderFallbackRecovery, summarizeVoiceSessions, summarizeVoiceSessionReplay, } from "./sessionReplay";
 export { createVoiceAgent, createVoiceAgentSquad, createVoiceAgentTool, } from "./agent";
+export { createAIVoiceModel } from "./aiVoiceModel";
+export type { CreateAIVoiceModelOptions } from "./aiVoiceModel";
 export { createVoiceRAGTool } from "./ragTool";
 export type { VoiceRAGCollectionLike, VoiceRAGQueryResult, VoiceRAGSearchInput, VoiceRAGToolArgs, VoiceRAGToolOptions, VoiceRAGToolResult, } from "./ragTool";
 export { createVoiceApiRequestTool, createVoiceDTMFTool, createVoiceEndCallTool, createVoiceTransferCallTool, createVoiceVoicemailDetectionTool, } from "./agentTools";

package/dist/index.js

@@ -3369,6 +3369,9 @@ var buildTurnText = (transcripts, partialText, options = {}) => {
   return selectPreferredTranscriptText(finalText, nextPartial);
 };
 
+// src/types.ts
+var ttsAdapterSessionCanCancel = (session) => typeof session.cancel === "function";
+
 // src/session.ts
 var DEFAULT_RECONNECT_TIMEOUT = 30000;
 var DEFAULT_MAX_RECONNECT_ATTEMPTS = 10;
@@ -3789,6 +3792,27 @@ var createVoiceSession = (options) => {
       });
     }
   };
+  const cancelActiveTTS = async (reason) => {
+    const activeSession = ttsSession;
+    const cancelledTurnId = activeTTSTurnId;
+    if (!activeSession || cancelledTurnId === undefined) {
+      return;
+    }
+    activeTTSTurnId = undefined;
+    if (!ttsAdapterSessionCanCancel(activeSession)) {
+      return;
+    }
+    try {
+      await activeSession.cancel(reason);
+    } catch (error) {
+      logger.warn("voice tts adapter cancel failed", {
+        error: toError(error).message,
+        reason,
+        sessionId: options.id,
+        turnId: cancelledTurnId
+      });
+    }
+  };
   const sendAssistantAudio = async (chunk, input) => {
     const normalizedChunk = chunk instanceof Uint8Array ? new Uint8Array(chunk) : chunk instanceof ArrayBuffer ? new Uint8Array(chunk.slice(0)) : new Uint8Array(chunk.buffer.slice(chunk.byteOffset, chunk.byteOffset + chunk.byteLength));
     await send({
@@ -5097,6 +5121,9 @@ var createVoiceSession = (options) => {
       pushTurnAudio(conditionedAudio);
     }
     if (audioLevel >= turnDetection.speechThreshold) {
+      if (!speechDetected && activeTTSTurnId !== undefined) {
+        cancelActiveTTS("barge-in");
+      }
       speechDetected = true;
       clearSilenceTimer();
     } else if (speechDetected) {
@@ -34559,6 +34586,76 @@ var createVoiceWorkflowContractHandler = (input) => {
     return result;
   };
 };
+// src/aiVoiceModel.ts
+var toProviderMessages = (messages) => {
+  const out = [];
+  for (const message of messages) {
+    if (message.role === "tool") {
+      out.push({
+        content: [
+          {
+            content: message.content,
+            tool_use_id: message.toolCallId ?? message.name ?? "",
+            type: "tool_result"
+          }
+        ],
+        role: "user"
+      });
+      continue;
+    }
+    if (message.role === "system") {
+      out.push({ content: message.content, role: "user" });
+      continue;
+    }
+    out.push({ content: message.content, role: message.role });
+  }
+  return out;
+};
+var toProviderTools = (tools) => {
+  if (tools.length === 0) {
+    return;
+  }
+  return tools.map((tool) => ({
+    description: tool.description ?? "",
+    input_schema: tool.parameters ?? {
+      properties: {},
+      type: "object"
+    },
+    name: tool.name
+  }));
+};
+var createAIVoiceModel = (options) => ({
+  generate: async (input) => {
+    const systemPrompt = input.system ?? options.systemPrompt;
+    const stream = options.provider.stream({
+      messages: toProviderMessages(input.messages),
+      model: options.model,
+      signal: options.signal,
+      systemPrompt,
+      tools: toProviderTools(input.tools)
+    });
+    let assistantText = "";
+    const toolCalls = [];
+    for await (const chunk of stream) {
+      if (chunk.type === "text") {
+        assistantText += chunk.content;
+      } else if (chunk.type === "tool_use") {
+        toolCalls.push({
+          args: chunk.input ?? {},
+          id: chunk.id,
+          name: chunk.name
+        });
+      }
+    }
+    const output = {
+      assistantText
+    };
+    if (toolCalls.length > 0) {
+      output.toolCalls = toolCalls;
+    }
+    return output;
+  }
+});
 // src/ragTool.ts
 var DEFAULT_TOOL_NAME = "searchKnowledgeBase";
 var DEFAULT_DESCRIPTION = "Search the knowledge base and return short grounded citations. Use this whenever the caller asks a question that may be answered by indexed reference material.";
@@ -44991,6 +45088,7 @@ export {
   verifyVoiceOpsWebhookSignature,
   validateVoiceWorkflowRouteResult,
   validateVoiceObservabilityExportRecord,
+  ttsAdapterSessionCanCancel,
   transcodeTwilioInboundPayloadToPCM16,
   transcodePCMToTwilioOutboundPayload,
   summarizeVoiceTurnQuality,
@@ -45614,6 +45712,7 @@ export {
   createDomainPhraseHints,
   createDomainLexicon,
   createAnthropicVoiceAssistantModel,
+  createAIVoiceModel,
   conditionAudioChunk,
   completeVoiceOpsTask,
   compareVoiceEvalBaseline,

package/dist/testing/index.js

@@ -5337,6 +5337,9 @@ var resolveLogger = (logger) => ({
   ...logger
 });
 
+// src/types.ts
+var ttsAdapterSessionCanCancel = (session) => typeof session.cancel === "function";
+
 // src/session.ts
 var DEFAULT_RECONNECT_TIMEOUT = 30000;
 var DEFAULT_MAX_RECONNECT_ATTEMPTS2 = 10;
@@ -5757,6 +5760,27 @@ var createVoiceSession = (options) => {
       });
     }
   };
+  const cancelActiveTTS = async (reason) => {
+    const activeSession = ttsSession;
+    const cancelledTurnId = activeTTSTurnId;
+    if (!activeSession || cancelledTurnId === undefined) {
+      return;
+    }
+    activeTTSTurnId = undefined;
+    if (!ttsAdapterSessionCanCancel(activeSession)) {
+      return;
+    }
+    try {
+      await activeSession.cancel(reason);
+    } catch (error) {
+      logger.warn("voice tts adapter cancel failed", {
+        error: toError(error).message,
+        reason,
+        sessionId: options.id,
+        turnId: cancelledTurnId
+      });
+    }
+  };
   const sendAssistantAudio = async (chunk, input) => {
     const normalizedChunk = chunk instanceof Uint8Array ? new Uint8Array(chunk) : chunk instanceof ArrayBuffer ? new Uint8Array(chunk.slice(0)) : new Uint8Array(chunk.buffer.slice(chunk.byteOffset, chunk.byteOffset + chunk.byteLength));
     await send({
@@ -7065,6 +7089,9 @@ var createVoiceSession = (options) => {
       pushTurnAudio(conditionedAudio);
     }
     if (audioLevel >= turnDetection.speechThreshold) {
+      if (!speechDetected && activeTTSTurnId !== undefined) {
+        cancelActiveTTS("barge-in");
+      }
       speechDetected = true;
       clearSilenceTimer();
     } else if (speechDetected) {
@@ -13154,8 +13181,9 @@ var runTTSAdapterFixture = async (adapter, fixture, options = {}) => {
     sessionId: `tts-benchmark:${fixture.id}`,
     ...openOptions ?? {}
   });
+  const sessionOn = session.on;
   const unsubscribers = [
-    session.on("audio", ({ chunk, format, receivedAt }) => {
+    sessionOn("audio", ({ chunk, format, receivedAt }) => {
       const normalizedChunk = chunk instanceof Uint8Array ? chunk : chunk instanceof ArrayBuffer ? new Uint8Array(chunk) : new Uint8Array(chunk.buffer, chunk.byteOffset, chunk.byteLength);
       audioChunkCount += 1;
       totalAudioBytes += normalizedChunk.byteLength;
@@ -13175,10 +13203,10 @@ var runTTSAdapterFixture = async (adapter, fixture, options = {}) => {
         }, options.interruptAfterFirstAudioMs);
       }
     }),
-    session.on("error", () => {
+    sessionOn("error", () => {
       errorCount += 1;
     }),
-    session.on("close", () => {
+    sessionOn("close", () => {
       closeCount += 1;
       closed = true;
       closedAt = Date.now();

package/dist/types.d.ts

@@ -166,8 +166,12 @@ 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 TTSAdapterOpenOptions = {
     sessionId: string;
     lexicon?: VoiceLexiconEntry[];

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@absolutejs/voice",
-  "version": "0.0.22-beta.475",
+  "version": "0.0.22-beta.476",
   "description": "Voice primitives and Elysia plugin for AbsoluteJS",
   "repository": {
     "type": "git",
@@ -160,7 +160,7 @@
     "bench:stt": "bun run ./scripts/benchmark-stt.ts all",
     "bench:assemblyai:sessions": "bun run ./scripts/benchmark-session.ts assemblyai",
     "bench:openai:sessions": "bun run ./scripts/benchmark-session.ts openai",
-    "build": "bun run ./scripts/build-htmx-bootstrap-asset.ts && rm -rf dist && bun build ./src/index.ts ./src/client/index.ts ./src/react/index.ts ./src/vue/index.ts ./src/svelte/index.ts ./src/angular/index.ts ./src/testing/index.ts --outdir dist --target bun --external elysia --external react --external vue --external @angular/core --external @absolutejs/absolute --external @absolutejs/media && bun build ./src/client/htmxBootstrap.ts --outdir dist/client --target browser --format esm && tsc --emitDeclarationOnly --project tsconfig.json",
+    "build": "bun run ./scripts/build-htmx-bootstrap-asset.ts && rm -rf dist && bun build ./src/index.ts ./src/client/index.ts ./src/react/index.ts ./src/vue/index.ts ./src/svelte/index.ts ./src/angular/index.ts ./src/testing/index.ts --outdir dist --target bun --external elysia --external react --external vue --external @angular/core --external @absolutejs/absolute --external @absolutejs/ai --external @absolutejs/media && bun build ./src/client/htmxBootstrap.ts --outdir dist/client --target browser --format esm && tsc --emitDeclarationOnly --project tsconfig.json",
     "format": "prettier --write \"./**/*.{js,jsx,ts,tsx,json,md}\"",
     "lint": "eslint ./src",
     "release": "bun run format && bun run build && bun publish",
@@ -229,12 +229,16 @@
   },
   "peerDependencies": {
     "@absolutejs/absolute": ">=0.19.0-beta.646",
+    "@absolutejs/ai": ">=0.0.5",
     "@angular/core": ">=21.0.0",
     "elysia": ">=1.4.18",
     "react": ">=19.0.0",
     "vue": ">=3.5.0"
   },
   "peerDependenciesMeta": {
+    "@absolutejs/ai": {
+      "optional": true
+    },
     "@angular/core": {
       "optional": true
     },
@@ -250,6 +254,7 @@
   },
   "devDependencies": {
     "@absolutejs/absolute": "0.19.0-beta.646",
+    "@absolutejs/ai": "0.0.5",
     "@angular/core": "^21.0.0",
     "@types/bun": "1.3.9",
     "@types/react": "19.2.0",