@ai-sdk/google 4.0.24 → 4.0.26

package/CHANGELOG.md

@@ -1,5 +1,22 @@
 # @ai-sdk/google
 
+## 4.0.26
+
+### Patch Changes
+
+- c49380c: feat: add experimental streaming speech translation models (`openai.translation('gpt-realtime-translate')` over the OpenAI Realtime translations WebSocket and `google.translation('gemini-3.5-live-translate-preview')` over the Gemini Live API). `connectToWebSocket` in `@ai-sdk/provider-utils` now passes close code and reason to `onClose` (additive, optional parameter).
+- Updated dependencies [0c464d9]
+- Updated dependencies [c49380c]
+  - @ai-sdk/provider-utils@5.0.14
+
+## 4.0.25
+
+### Patch Changes
+
+- Updated dependencies [1e2f324]
+  - @ai-sdk/provider@4.0.4
+  - @ai-sdk/provider-utils@5.0.13
+
 ## 4.0.24
 
 ### Patch Changes

package/dist/index.d.ts

@@ -1,6 +1,7 @@
 import * as _ai_sdk_provider_utils from '@ai-sdk/provider-utils';
-import { InferSchema, FetchFunction } from '@ai-sdk/provider-utils';
-import { ProviderV4, LanguageModelV4, ImageModelV4, EmbeddingModelV4, Experimental_VideoModelV4, SpeechModelV4, FilesV4, Experimental_RealtimeFactoryV4, Experimental_RealtimeModelV4, Experimental_RealtimeModelV4ClientSecretOptions, Experimental_RealtimeModelV4ClientSecretResult, Experimental_RealtimeModelV4ServerEvent, Experimental_RealtimeModelV4ClientEvent, Experimental_RealtimeModelV4SessionConfig } from '@ai-sdk/provider';
+import { InferSchema, FetchFunction, WebSocketConstructor, WORKFLOW_SERIALIZE, WORKFLOW_DESERIALIZE } from '@ai-sdk/provider-utils';
+import * as _ai_sdk_provider from '@ai-sdk/provider';
+import { ProviderV4, LanguageModelV4, ImageModelV4, EmbeddingModelV4, Experimental_VideoModelV4, Experimental_SpeechTranslationModelV4, SpeechModelV4, FilesV4, Experimental_RealtimeFactoryV4, Experimental_RealtimeModelV4, Experimental_RealtimeModelV4ClientSecretOptions, Experimental_RealtimeModelV4ClientSecretResult, Experimental_RealtimeModelV4ServerEvent, Experimental_RealtimeModelV4ClientEvent, Experimental_RealtimeModelV4SessionConfig, Experimental_SpeechTranslationModelV4StreamOptions } from '@ai-sdk/provider';
 
 declare const googleErrorDataSchema: _ai_sdk_provider_utils.LazySchema<{
     error: {
@@ -510,6 +511,12 @@ interface GoogleImageSettings {
     maxImagesPerCall?: number;
 }
 
+type GoogleTranslationModelId = 'gemini-3.5-live-translate-preview' | (string & {});
+declare const googleTranslationModelOptions: _ai_sdk_provider_utils.LazySchema<{
+    echoTargetLanguage?: boolean | undefined;
+}>;
+type GoogleTranslationModelOptions = InferSchema<typeof googleTranslationModelOptions>;
+
 interface GoogleProvider extends ProviderV4 {
     (modelId: GoogleModelId): LanguageModelV4;
     languageModel(modelId: GoogleModelId): LanguageModelV4;
@@ -546,6 +553,14 @@ interface GoogleProvider extends ProviderV4 {
      * Creates a model for video generation.
      */
     videoModel(modelId: GoogleVideoModelId): Experimental_VideoModelV4;
+    /**
+     * Creates an experimental model for streaming speech translation.
+     */
+    translation(modelId: GoogleTranslationModelId): Experimental_SpeechTranslationModelV4;
+    /**
+     * Creates an experimental model for streaming speech translation.
+     */
+    speechTranslationModel(modelId: GoogleTranslationModelId): Experimental_SpeechTranslationModelV4;
     /**
      * Creates a model for speech generation (text-to-speech).
      */
@@ -595,6 +610,11 @@ interface GoogleProviderSettings {
      * Optional function to generate a unique ID for each request.
      */
     generateId?: () => string;
+    /**
+     * Custom WebSocket implementation. This is useful for testing or for
+     * runtimes that need a WebSocket constructor with header support.
+     */
+    webSocket?: WebSocketConstructor;
     /**
      * Custom provider name
      * Defaults to 'google.generative-ai'.
@@ -658,6 +678,33 @@ type GoogleRealtimeModelOptions = {
     };
 };
 
+type GoogleTranslationModelConfig = {
+    provider: string;
+    baseURL: string;
+    headers: () => Record<string, string | undefined>;
+    webSocket?: WebSocketConstructor;
+    _internal?: {
+        currentDate?: () => Date;
+        finishGraceMs?: number;
+    };
+};
+declare class GoogleTranslationModel implements Experimental_SpeechTranslationModelV4 {
+    readonly specificationVersion = "v4";
+    readonly modelId: GoogleTranslationModelId;
+    private readonly config;
+    static [WORKFLOW_SERIALIZE](model: GoogleTranslationModel): {
+        modelId: string;
+        config: _ai_sdk_provider.JSONObject;
+    };
+    static [WORKFLOW_DESERIALIZE](options: {
+        modelId: GoogleTranslationModelId;
+        config: GoogleTranslationModelConfig;
+    }): GoogleTranslationModel;
+    get provider(): string;
+    constructor(modelId: GoogleTranslationModelId, config: GoogleTranslationModelConfig);
+    doStream(options: Experimental_SpeechTranslationModelV4StreamOptions): Promise<Awaited<ReturnType<Experimental_SpeechTranslationModelV4['doStream']>>>;
+}
+
 declare const VERSION: string;
 
-export { GoogleRealtimeModel as Experimental_GoogleRealtimeModel, type GoogleRealtimeModelConfig as Experimental_GoogleRealtimeModelConfig, type GoogleRealtimeModelId as Experimental_GoogleRealtimeModelId, type GoogleRealtimeModelOptions as Experimental_GoogleRealtimeModelOptions, type GoogleEmbeddingModelOptions, type GoogleErrorData, type GoogleFilesUploadOptions, type GoogleEmbeddingModelOptions as GoogleGenerativeAIEmbeddingProviderOptions, type GoogleImageModelOptions as GoogleGenerativeAIImageProviderOptions, type GoogleProvider as GoogleGenerativeAIProvider, type GoogleProviderMetadata as GoogleGenerativeAIProviderMetadata, type GoogleLanguageModelOptions as GoogleGenerativeAIProviderOptions, type GoogleProviderSettings as GoogleGenerativeAIProviderSettings, type GoogleVideoModelId as GoogleGenerativeAIVideoModelId, type GoogleVideoModelOptions as GoogleGenerativeAIVideoProviderOptions, type GoogleImageModelOptions, type GoogleInteractionsAgentName, type GoogleInteractionsModelId, type GoogleInteractionsProviderMetadata, type GoogleLanguageModelInteractionsOptions, type GoogleLanguageModelOptions, type GoogleProvider, type GoogleProviderMetadata, type GoogleProviderSettings, type GoogleSpeechModelId, type GoogleSpeechModelOptions, type GoogleVideoModelId, type GoogleVideoModelOptions, VERSION, createGoogle, createGoogle as createGoogleGenerativeAI, google };
+export { GoogleRealtimeModel as Experimental_GoogleRealtimeModel, type GoogleRealtimeModelConfig as Experimental_GoogleRealtimeModelConfig, type GoogleRealtimeModelId as Experimental_GoogleRealtimeModelId, type GoogleRealtimeModelOptions as Experimental_GoogleRealtimeModelOptions, GoogleTranslationModel as Experimental_GoogleTranslationModel, type GoogleTranslationModelConfig as Experimental_GoogleTranslationModelConfig, type GoogleTranslationModelId as Experimental_GoogleTranslationModelId, type GoogleTranslationModelOptions as Experimental_GoogleTranslationModelOptions, type GoogleEmbeddingModelOptions, type GoogleErrorData, type GoogleFilesUploadOptions, type GoogleEmbeddingModelOptions as GoogleGenerativeAIEmbeddingProviderOptions, type GoogleImageModelOptions as GoogleGenerativeAIImageProviderOptions, type GoogleProvider as GoogleGenerativeAIProvider, type GoogleProviderMetadata as GoogleGenerativeAIProviderMetadata, type GoogleLanguageModelOptions as GoogleGenerativeAIProviderOptions, type GoogleProviderSettings as GoogleGenerativeAIProviderSettings, type GoogleVideoModelId as GoogleGenerativeAIVideoModelId, type GoogleVideoModelOptions as GoogleGenerativeAIVideoProviderOptions, type GoogleImageModelOptions, type GoogleInteractionsAgentName, type GoogleInteractionsModelId, type GoogleInteractionsProviderMetadata, type GoogleLanguageModelInteractionsOptions, type GoogleLanguageModelOptions, type GoogleProvider, type GoogleProviderMetadata, type GoogleProviderSettings, type GoogleSpeechModelId, type GoogleSpeechModelOptions, type GoogleVideoModelId, type GoogleVideoModelOptions, VERSION, createGoogle, createGoogle as createGoogleGenerativeAI, google };

package/dist/index.js

@@ -7,7 +7,7 @@ import {
 } from "@ai-sdk/provider-utils";
 
 // src/version.ts
-var VERSION = true ? "4.0.24" : "0.0.0-test";
+var VERSION = true ? "4.0.26" : "0.0.0-test";
 
 // src/google-embedding-model.ts
 import {
@@ -6930,6 +6930,24 @@ function pruneUndefined(obj) {
   return result;
 }
 
+// src/get-realtime-base-url.ts
+function getRealtimeBaseURL(baseURL) {
+  const url = new URL(baseURL);
+  const pathSegments = url.pathname.split("/");
+  const version = pathSegments.at(-1);
+  if (version === "v1beta" || version === "v1alpha") {
+    pathSegments.pop();
+    url.pathname = pathSegments.join("/") || "/";
+  }
+  return url;
+}
+function getRealtimeWebSocketURL(baseURL, webSocketPath) {
+  const url = getRealtimeBaseURL(baseURL);
+  url.protocol = url.protocol === "https:" ? "wss:" : "ws:";
+  url.pathname = `${url.pathname.replace(/\/$/, "")}/ws/${webSocketPath}`;
+  return url;
+}
+
 // src/realtime/google-realtime-event-mapper.ts
 import { safeParseJSON } from "@ai-sdk/provider-utils";
 function isRecord(value) {
@@ -7257,26 +7275,13 @@ function buildGoogleSessionConfig(config, modelId) {
 
 // src/realtime/google-realtime-model.ts
 var realtimeWebSocketPath = "google.ai.generativelanguage.v1alpha.GenerativeService.BidiGenerateContentConstrained";
-function getRealtimeBaseURL(baseURL) {
-  const url = new URL(baseURL);
-  const pathSegments = url.pathname.split("/");
-  const version = pathSegments.at(-1);
-  if (version === "v1beta" || version === "v1alpha") {
-    pathSegments.pop();
-    url.pathname = pathSegments.join("/") || "/";
-  }
-  return url;
-}
 function getAuthTokensURL(baseURL) {
   const url = getRealtimeBaseURL(baseURL);
   url.pathname = `${url.pathname.replace(/\/$/, "")}/v1alpha/auth_tokens`;
   return url.toString();
 }
 function getWebSocketURL(baseURL) {
-  const url = getRealtimeBaseURL(baseURL);
-  url.protocol = url.protocol === "https:" ? "wss:" : "ws:";
-  url.pathname = `${url.pathname.replace(/\/$/, "")}/ws/${realtimeWebSocketPath}`;
-  return url.toString();
+  return getRealtimeWebSocketURL(baseURL, realtimeWebSocketPath).toString();
 }
 var GoogleRealtimeModel = class {
   constructor(modelId, config) {
@@ -7351,6 +7356,459 @@ var GoogleRealtimeModel = class {
   }
 };
 
+// src/translation/google-translation-model.ts
+import {
+  InvalidArgumentError
+} from "@ai-sdk/provider";
+import {
+  connectToWebSocket,
+  combineHeaders as combineHeaders9,
+  convertBase64ToUint8Array as convertBase64ToUint8Array2,
+  convertToBase64 as convertToBase644,
+  parseProviderOptions as parseProviderOptions8,
+  safeParseJSON as safeParseJSON2,
+  serializeModelOptions as serializeModelOptions6,
+  WORKFLOW_DESERIALIZE as WORKFLOW_DESERIALIZE6,
+  WORKFLOW_SERIALIZE as WORKFLOW_SERIALIZE6,
+  waitForWebSocketBufferDrain
+} from "@ai-sdk/provider-utils";
+
+// src/translation/google-translation-model-options.ts
+import {
+  lazySchema as lazySchema20,
+  zodSchema as zodSchema20
+} from "@ai-sdk/provider-utils";
+import { z as z22 } from "zod/v4";
+var googleTranslationModelOptions = lazySchema20(
+  () => zodSchema20(
+    z22.object({
+      /**
+       * Whether input audio already in the target language should be echoed
+       * instead of producing silence.
+       */
+      echoTargetLanguage: z22.boolean().optional()
+    })
+  )
+);
+
+// src/translation/google-translation-model.ts
+var liveWebSocketPath = "google.ai.generativelanguage.v1beta.GenerativeService.BidiGenerateContent";
+var defaultFinishGraceMs = 1e3;
+var googleLiveOutputAudioRate = 24e3;
+var pcm16SilenceAmplitudeThreshold = 128;
+function getLiveWebSocketURL(baseURL, apiKey) {
+  const url = getRealtimeWebSocketURL(baseURL, liveWebSocketPath);
+  url.searchParams.set("key", apiKey);
+  return url;
+}
+var GoogleTranslationModel = class _GoogleTranslationModel {
+  constructor(modelId, config) {
+    this.specificationVersion = "v4";
+    this.modelId = modelId;
+    this.config = config;
+  }
+  static [WORKFLOW_SERIALIZE6](model) {
+    return serializeModelOptions6({
+      modelId: model.modelId,
+      config: model.config
+    });
+  }
+  static [WORKFLOW_DESERIALIZE6](options) {
+    return new _GoogleTranslationModel(options.modelId, options.config);
+  }
+  get provider() {
+    return this.config.provider;
+  }
+  async doStream(options) {
+    var _a, _b, _c, _d, _e, _f;
+    if (options.targetLanguage == null) {
+      throw new InvalidArgumentError({
+        argument: "targetLanguage",
+        message: `targetLanguage is required for translation model '${this.modelId}'.`
+      });
+    }
+    const currentDate = (_c = (_b = (_a = this.config._internal) == null ? void 0 : _a.currentDate) == null ? void 0 : _b.call(_a)) != null ? _c : /* @__PURE__ */ new Date();
+    const googleOptions = await parseProviderOptions8({
+      provider: "google",
+      providerOptions: options.providerOptions,
+      schema: googleTranslationModelOptions
+    });
+    const warnings = [];
+    validateGoogleTranslationInputAudioFormat(options.inputAudioFormat);
+    if (options.sourceLanguage != null) {
+      warnings.push({
+        type: "unsupported",
+        feature: "sourceLanguage",
+        details: "The Gemini Live translation API auto-detects the source language and does not accept a source language."
+      });
+    }
+    if (options.outputAudioFormat != null) {
+      warnings.push({
+        type: "unsupported",
+        feature: "outputAudioFormat",
+        details: "The Gemini Live API always outputs 24kHz 16-bit PCM audio and does not accept an output audio format."
+      });
+    }
+    const headers = combineHeaders9(this.config.headers(), options.headers);
+    let apiKey;
+    for (const [key, value] of Object.entries(headers)) {
+      if (key.toLowerCase() === "x-goog-api-key" && value != null) {
+        apiKey = value;
+      }
+    }
+    if (apiKey == null) {
+      throw new Error(
+        "Google Generative AI API key is required for streaming translation."
+      );
+    }
+    const webSocketHeaders = Object.fromEntries(
+      Object.entries(headers).filter(
+        ([key]) => key.toLowerCase() !== "x-goog-api-key"
+      )
+    );
+    const setup = buildGoogleLiveTranslationSetup({
+      modelId: this.modelId,
+      targetLanguage: options.targetLanguage,
+      providerOptions: googleOptions
+    });
+    return {
+      request: { body: setup },
+      response: {
+        timestamp: currentDate,
+        modelId: this.modelId
+      },
+      stream: createGoogleLiveTranslationStream({
+        webSocket: this.config.webSocket,
+        url: getLiveWebSocketURL(this.config.baseURL, apiKey),
+        headers: webSocketHeaders,
+        setup,
+        inputAudioRate: (_d = options.inputAudioFormat.rate) != null ? _d : 16e3,
+        finishGraceMs: (_f = (_e = this.config._internal) == null ? void 0 : _e.finishGraceMs) != null ? _f : defaultFinishGraceMs,
+        warnings,
+        audio: options.audio,
+        abortSignal: options.abortSignal,
+        includeRawChunks: options.includeRawChunks
+      })
+    };
+  }
+};
+function createGoogleLiveTranslationStream({
+  webSocket,
+  url,
+  headers,
+  setup,
+  inputAudioRate,
+  finishGraceMs,
+  warnings,
+  audio,
+  abortSignal,
+  includeRawChunks
+}) {
+  let finished = false;
+  let cleanup = () => {
+  };
+  return new ReadableStream({
+    start: (controller) => {
+      let audioReader;
+      let connection;
+      let resolveSetupComplete;
+      const setupComplete = new Promise((resolve7) => {
+        resolveSetupComplete = resolve7;
+      });
+      let turnCounter = 0;
+      let sourceText = "";
+      let sourceTurnBuffer = "";
+      let translationText = "";
+      let translationTurnBuffer = "";
+      let audioEnded = false;
+      let usage;
+      let openTurn = false;
+      let sawTurnComplete = false;
+      let trailingSilenceMs = 0;
+      let finishTimer;
+      const itemId = () => `google-item-${turnCounter}`;
+      const cancelPendingFinish = () => {
+        if (finishTimer != null) {
+          clearTimeout(finishTimer);
+          finishTimer = void 0;
+        }
+      };
+      const schedulePendingFinish = () => {
+        if (finished || finishTimer != null) return;
+        finishTimer = setTimeout(() => {
+          finishTimer = void 0;
+          finish();
+        }, finishGraceMs);
+      };
+      const onTurnActivity = () => {
+        openTurn = true;
+        trailingSilenceMs = 0;
+        cancelPendingFinish();
+      };
+      cleanup = (closeCode) => {
+        cancelPendingFinish();
+        if (audioReader != null) {
+          void audioReader.cancel().catch(() => {
+          });
+        } else {
+          void audio.cancel().catch(() => {
+          });
+        }
+        connection == null ? void 0 : connection.close(closeCode);
+      };
+      const finishWithError = (error) => {
+        if (finished) return;
+        finished = true;
+        cleanup();
+        controller.error(error);
+      };
+      const finish = () => {
+        if (finished) return;
+        if (sourceTurnBuffer !== "" || translationTurnBuffer !== "") {
+          completeTurn();
+        }
+        finished = true;
+        controller.enqueue({
+          type: "finish",
+          sourceText,
+          outputText: translationText,
+          usage
+        });
+        controller.close();
+        cleanup(1e3);
+      };
+      const completeTurn = () => {
+        if (sourceTurnBuffer !== "") {
+          controller.enqueue({
+            type: "source-transcript-final",
+            id: itemId(),
+            text: sourceTurnBuffer
+          });
+          sourceText += sourceTurnBuffer;
+          sourceTurnBuffer = "";
+        }
+        if (translationTurnBuffer !== "") {
+          controller.enqueue({
+            type: "output-text-final",
+            id: itemId(),
+            text: translationTurnBuffer
+          });
+          translationText += translationTurnBuffer;
+          translationTurnBuffer = "";
+        }
+        turnCounter++;
+      };
+      const sendAudio = async (socket) => {
+        audioReader = audio.getReader();
+        try {
+          while (true) {
+            const { done, value } = await audioReader.read();
+            if (done || finished) break;
+            socket.send(
+              JSON.stringify({
+                realtimeInput: {
+                  audio: {
+                    data: convertToBase644(value),
+                    mimeType: `audio/pcm;rate=${inputAudioRate}`
+                  }
+                }
+              })
+            );
+            await waitForWebSocketBufferDrain(socket);
+          }
+        } finally {
+          audioReader.releaseLock();
+          audioReader = void 0;
+        }
+        if (!finished) {
+          socket.send(
+            JSON.stringify({ realtimeInput: { audioStreamEnd: true } })
+          );
+          audioEnded = true;
+          if (sawTurnComplete && !openTurn) {
+            schedulePendingFinish();
+          }
+        }
+      };
+      connection = connectToWebSocket({
+        url,
+        headers,
+        webSocket,
+        abortSignal,
+        onAbort: finishWithError,
+        onProcessingError: finishWithError,
+        onOpen: (socket) => {
+          controller.enqueue({ type: "stream-start", warnings });
+          socket.send(JSON.stringify({ setup }));
+          void setupComplete.then(() => finished ? void 0 : sendAudio(socket)).catch(finishWithError);
+        },
+        onMessageText: async (text) => {
+          var _a, _b, _c, _d, _e, _f, _g, _h, _i;
+          if (finished) return;
+          const parsed = await safeParseJSON2({ text });
+          if (!parsed.success) return;
+          const message = parsed.value;
+          if (includeRawChunks) {
+            controller.enqueue({ type: "raw", rawValue: message });
+          }
+          if (message.setupComplete != null) {
+            resolveSetupComplete();
+          }
+          if (message.usageMetadata != null) {
+            usage = accumulateGoogleLiveUsage(usage, message.usageMetadata);
+          }
+          if (message.error != null) {
+            finishWithError(
+              new Error((_a = message.error.message) != null ? _a : "Google Live API error")
+            );
+            return;
+          }
+          const inputTranscriptionText = (_e = (_c = (_b = message.serverContent) == null ? void 0 : _b.inputTranscription) == null ? void 0 : _c.text) != null ? _e : (_d = message.inputTranscription) == null ? void 0 : _d.text;
+          if (inputTranscriptionText) {
+            onTurnActivity();
+            sourceTurnBuffer += inputTranscriptionText;
+            controller.enqueue({
+              type: "source-transcript-delta",
+              id: itemId(),
+              delta: inputTranscriptionText
+            });
+          }
+          const serverContent = message.serverContent;
+          if (serverContent == null) {
+            return;
+          }
+          for (const part of (_g = (_f = serverContent.modelTurn) == null ? void 0 : _f.parts) != null ? _g : []) {
+            if ((_h = part.inlineData) == null ? void 0 : _h.data) {
+              controller.enqueue({
+                type: "audio",
+                id: itemId(),
+                audio: part.inlineData.data
+              });
+              const silenceDurationMs = getPcm16SilenceDurationMs(
+                part.inlineData.data
+              );
+              if (audioEnded && silenceDurationMs != null) {
+                trailingSilenceMs += silenceDurationMs;
+                if (trailingSilenceMs >= finishGraceMs) {
+                  finish();
+                  return;
+                }
+              } else {
+                onTurnActivity();
+              }
+            }
+          }
+          if ((_i = serverContent.outputTranscription) == null ? void 0 : _i.text) {
+            onTurnActivity();
+            translationTurnBuffer += serverContent.outputTranscription.text;
+            controller.enqueue({
+              type: "output-text-delta",
+              id: itemId(),
+              delta: serverContent.outputTranscription.text
+            });
+          }
+          if (serverContent.turnComplete) {
+            completeTurn();
+            openTurn = false;
+            sawTurnComplete = true;
+            if (audioEnded) {
+              schedulePendingFinish();
+            }
+          }
+        },
+        onSocketError: () => {
+          finishWithError(new Error("Google Live translation error"));
+        },
+        onClose: ({ code, reason }) => {
+          if (finished) return;
+          if (finishTimer != null) {
+            finish();
+            return;
+          }
+          finishWithError(
+            new Error(
+              `Google Live translation WebSocket closed unexpectedly before finishing (code ${code != null ? code : "unknown"}${reason ? `, reason: ${reason}` : ""}).`
+            )
+          );
+        }
+      });
+    },
+    cancel: () => {
+      if (finished) return;
+      finished = true;
+      cleanup();
+    }
+  });
+}
+function accumulateGoogleLiveUsage(usage, usageMetadata) {
+  var _a, _b;
+  let inputAudioTokens = usage == null ? void 0 : usage.inputAudioTokens;
+  let outputAudioTokens = usage == null ? void 0 : usage.outputAudioTokens;
+  for (const detail of (_a = usageMetadata.promptTokensDetails) != null ? _a : []) {
+    if (detail.modality === "AUDIO" && detail.tokenCount != null) {
+      inputAudioTokens = (inputAudioTokens != null ? inputAudioTokens : 0) + detail.tokenCount;
+    }
+  }
+  for (const detail of (_b = usageMetadata.responseTokensDetails) != null ? _b : []) {
+    if (detail.modality === "AUDIO" && detail.tokenCount != null) {
+      outputAudioTokens = (outputAudioTokens != null ? outputAudioTokens : 0) + detail.tokenCount;
+    }
+  }
+  if (inputAudioTokens == null && outputAudioTokens == null) {
+    return usage;
+  }
+  return {
+    ...usage,
+    ...inputAudioTokens != null ? { inputAudioTokens } : {},
+    ...outputAudioTokens != null ? { outputAudioTokens } : {}
+  };
+}
+function getPcm16SilenceDurationMs(audio) {
+  let bytes;
+  try {
+    bytes = convertBase64ToUint8Array2(audio);
+  } catch (e) {
+    return void 0;
+  }
+  if (bytes.byteLength < 2) {
+    return void 0;
+  }
+  const view = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);
+  const sampleCount = Math.floor(bytes.byteLength / 2);
+  for (let i = 0; i < sampleCount; i++) {
+    if (Math.abs(view.getInt16(i * 2, true)) > pcm16SilenceAmplitudeThreshold) {
+      return void 0;
+    }
+  }
+  return sampleCount / googleLiveOutputAudioRate * 1e3;
+}
+function buildGoogleLiveTranslationSetup({
+  modelId,
+  targetLanguage,
+  providerOptions
+}) {
+  return {
+    model: getModelPath(modelId),
+    generationConfig: {
+      responseModalities: ["AUDIO"],
+      translationConfig: {
+        targetLanguageCode: targetLanguage,
+        ...(providerOptions == null ? void 0 : providerOptions.echoTargetLanguage) != null ? { echoTargetLanguage: providerOptions.echoTargetLanguage } : {}
+      }
+    },
+    inputAudioTranscription: {},
+    outputAudioTranscription: {}
+  };
+}
+function validateGoogleTranslationInputAudioFormat(inputAudioFormat) {
+  if (inputAudioFormat.type !== "audio/pcm" || inputAudioFormat.rate != null && inputAudioFormat.rate !== 16e3) {
+    throw new InvalidArgumentError({
+      argument: "inputAudioFormat",
+      message: "The Gemini Live translation API only supports 16kHz 16-bit PCM input audio."
+    });
+  }
+}
+
 // src/google-provider.ts
 var supportedExternalUrlMediaTypes = [
   "text/html",
@@ -7457,6 +7915,12 @@ function createGoogle(options = {}) {
     headers: getHeaders,
     fetch: options.fetch
   });
+  const createTranslationModel = (modelId) => new GoogleTranslationModel(modelId, {
+    provider: `${providerName}.translation`,
+    baseURL,
+    headers: getHeaders,
+    webSocket: options.webSocket
+  });
   const createSpeechModel = (modelId) => new GoogleSpeechModel(modelId, {
     provider: `${providerName}.speech`,
     baseURL,
@@ -7517,6 +7981,8 @@ function createGoogle(options = {}) {
   provider.files = createFiles;
   provider.speech = createSpeechModel;
   provider.speechModel = createSpeechModel;
+  provider.translation = createTranslationModel;
+  provider.speechTranslationModel = createTranslationModel;
   provider.interactions = createInteractionsModel;
   provider.tools = googleTools;
   return provider;
@@ -7524,6 +7990,7 @@ function createGoogle(options = {}) {
 var google = createGoogle();
 export {
   GoogleRealtimeModel as Experimental_GoogleRealtimeModel,
+  GoogleTranslationModel as Experimental_GoogleTranslationModel,
   VERSION,
   createGoogle,
   createGoogle as createGoogleGenerativeAI,