@ai-sdk/google 4.0.0-beta.83 → 4.0.0-beta.84

package/docs/15-google.mdx

@@ -1091,6 +1091,45 @@ Google realtime models may require provider-specific audio formats, depending
 on the model and modality. See [Realtime](/docs/ai-sdk-core/realtime) for the
 complete setup and tool calling pattern.
 
+### Live Translation
+
+Use `gemini-3.5-live-translate-preview` for low-latency speech-to-speech
+translation. Configure the target language with
+`providerOptions.google.translationConfig` in the realtime session config you
+use when creating the token:
+
+```ts
+import {
+  google,
+  type Experimental_GoogleRealtimeModelOptions as GoogleRealtimeModelOptions,
+} from '@ai-sdk/google';
+
+const token = await google.experimental_realtime.getToken({
+  model: 'gemini-3.5-live-translate-preview',
+  sessionConfig: {
+    outputModalities: ['audio'],
+    inputAudioTranscription: {},
+    outputAudioTranscription: {},
+    providerOptions: {
+      google: {
+        translationConfig: {
+          targetLanguageCode: 'pl',
+          echoTargetLanguage: true,
+        },
+      } satisfies GoogleRealtimeModelOptions,
+    },
+  },
+});
+```
+
+Set `targetLanguageCode` to a BCP-47 language code (defaults to `en`). When
+`echoTargetLanguage` is `true`, input audio already in the target language is
+echoed instead of producing silence.
+
+Gemini Live Translation accepts audio input and produces translated audio
+output. Text input, tools, and custom instructions are not supported by this
+model.
+
 ## Interactions API
 
 The [Gemini Interactions API](https://ai.google.dev/gemini-api/docs/interactions)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ai-sdk/google",
-  "version": "4.0.0-beta.83",
+  "version": "4.0.0-beta.84",
   "type": "module",
   "license": "Apache-2.0",
   "sideEffects": false,

package/src/index.ts

@@ -56,5 +56,9 @@ export type {
 } from './google-provider';
 export { GoogleRealtimeModel as Experimental_GoogleRealtimeModel } from './realtime/google-realtime-model';
 export type { GoogleRealtimeModelConfig as Experimental_GoogleRealtimeModelConfig } from './realtime/google-realtime-model';
+export type {
+  GoogleRealtimeModelId as Experimental_GoogleRealtimeModelId,
+  GoogleRealtimeModelOptions as Experimental_GoogleRealtimeModelOptions,
+} from './realtime/google-realtime-model-options';
 
 export { VERSION } from './version';

package/src/realtime/google-realtime-event-mapper.ts

@@ -8,6 +8,7 @@ import type {
 import { safeParseJSON } from '@ai-sdk/provider-utils';
 import { convertJSONSchemaToOpenAPISchema } from '../convert-json-schema-to-openapi-schema';
 import { getModelPath } from '../get-model-path';
+import type { GoogleRealtimeModelOptions } from './google-realtime-model-options';
 
 type GoogleRealtimeFunctionCall = {
   id: string;
@@ -38,6 +39,10 @@ type GoogleRealtimeWireEvent = {
   inputTranscription?: { text?: string };
 };
 
+function isRecord(value: unknown): value is Record<string, unknown> {
+  return value != null && typeof value === 'object' && !Array.isArray(value);
+}
+
 /**
  * Stateful event mapper for Google's Gemini Live API.
  *
@@ -268,6 +273,12 @@ export class GoogleRealtimeEventMapper {
         };
 
       case 'input-audio-commit':
+        return {
+          realtimeInput: {
+            audioStreamEnd: true,
+          },
+        };
+
       case 'input-audio-clear':
       case 'response-create':
       case 'response-cancel':
@@ -375,8 +386,25 @@ export function buildGoogleSessionConfig(
     setup.outputAudioTranscription = {};
   }
 
-  if (config?.providerOptions != null) {
-    Object.assign(setup, config.providerOptions);
+  if (config?.providerOptions == null) {
+    return setup;
+  }
+
+  const { google, ...providerOptions } = config.providerOptions;
+  Object.assign(setup, providerOptions);
+
+  const googleOptions = isRecord(google)
+    ? (google as GoogleRealtimeModelOptions)
+    : undefined;
+
+  if (googleOptions?.translationConfig != null) {
+    const target = isRecord(setup.generationConfig)
+      ? setup.generationConfig
+      : generationConfig;
+    setup.generationConfig = {
+      ...target,
+      translationConfig: googleOptions.translationConfig,
+    };
   }
 
   return setup;

package/src/realtime/google-realtime-model-options.ts

@@ -1,3 +1,23 @@
 export type GoogleRealtimeModelId = string;
 
-export type GoogleRealtimeModelOptions = Record<string, never>;
+export type GoogleRealtimeModelOptions = {
+  /**
+   * Gemini Live Translation configuration.
+   *
+   * Required for `gemini-3.5-live-translate-preview` when translating speech
+   * to a target language.
+   */
+  translationConfig?: {
+    /**
+     * BCP-47 language code of the language to translate into.
+     * Defaults to `en` in the Gemini API.
+     */
+    targetLanguageCode?: string;
+
+    /**
+     * Whether input audio already in the target language should be echoed
+     * instead of producing silence.
+     */
+    echoTargetLanguage?: boolean;
+  };
+};

package/src/realtime/index.ts

@@ -1,2 +1,6 @@
 export { GoogleRealtimeModel as Experimental_GoogleRealtimeModel } from './google-realtime-model';
 export type { GoogleRealtimeModelConfig as Experimental_GoogleRealtimeModelConfig } from './google-realtime-model';
+export type {
+  GoogleRealtimeModelId as Experimental_GoogleRealtimeModelId,
+  GoogleRealtimeModelOptions as Experimental_GoogleRealtimeModelOptions,
+} from './google-realtime-model-options';