Add live audio transcription streaming support to Foundry Local JS SDK

sdk_v2/cs/src/Microsoft.AI.Foundry.Local.csproj

@@ -99,8 +99,8 @@
     <PropertyGroup>
       <!-- default version unless overridden during dotnet build/restore command -->
       <FoundryLocalCoreWinMLVersion Condition="'$(FoundryLocalCoreVersion)' != ''">$(FoundryLocalCoreVersion)</FoundryLocalCoreWinMLVersion>
-      <FoundryLocalCoreWinMLVersion Condition="'$(FoundryLocalCoreVersion)' == ''">0.9.0.8-rc3</FoundryLocalCoreWinMLVersion>
-      <FoundryLocalCoreVersion Condition="'$(FoundryLocalCoreVersion)' == ''">0.9.0.8-rc3</FoundryLocalCoreVersion>    </PropertyGroup>
+      <FoundryLocalCoreWinMLVersion Condition="'$(FoundryLocalCoreVersion)' == ''">0.9.0-dev-20260227T230631-2a3af92</FoundryLocalCoreWinMLVersion>
+      <FoundryLocalCoreVersion Condition="'$(FoundryLocalCoreVersion)' == ''">0.9.0-dev-20260227T222239-2a3af92</FoundryLocalCoreVersion>    </PropertyGroup>
 
     <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|AnyCPU'">
       <TreatWarningsAsErrors>True</TreatWarningsAsErrors>
@@ -125,4 +125,4 @@
       <!-- specify PrivateAssets to exclude from nuget dependencies -->
       <PackageReference Include="IDisposableAnalyzers" Version="4.0.8" PrivateAssets="all"/>
     </ItemGroup>
-</Project>
+</Project>

sdk_v2/cs/src/OpenAI/ChatClient.cs

@@ -171,7 +171,7 @@ private async IAsyncEnumerable<ChatCompletionCreateResponse> ChatStreamingImplAs
             {
                 var failed = false;
 
-                var response = await _coreInterop.ExecuteCommandWithCallbackAsync(
+                await _coreInterop.ExecuteCommandWithCallbackAsync(
                     "chat_completions",
                     request,
                     async (callbackData) =>
@@ -196,17 +196,6 @@ private async IAsyncEnumerable<ChatCompletionCreateResponse> ChatStreamingImplAs
                     ct
                 ).ConfigureAwait(false);
 
-                // If the native layer returned an error (e.g. missing model, invalid input)
-                // without invoking any callbacks, propagate it so the caller sees an exception
-                // instead of an empty stream.
-                if (!failed && response.Error != null)
-                {
-                    channel.Writer.TryComplete(
-                        new FoundryLocalException($"Error from chat_completions command: {response.Error}", _logger));
-                    failed = true;
-                    return;
-                }
-
                 // use TryComplete as an exception in the callback may have already closed the channel
                 _ = channel.Writer.TryComplete();
             }

sdk_v2/cs/test/FoundryLocal.Tests/Microsoft.AI.Foundry.Local.Tests.csproj

@@ -37,6 +37,10 @@
     </None>
   </ItemGroup>
 
+  <ItemGroup Condition="'$(RuntimeIdentifier)' != 'linux-x64' AND '$(UseWinML)' != 'true'">
+    <PackageReference Include="Microsoft.ML.OnnxRuntime.Gpu.Linux" Version="1.24.1" ExcludeAssets="all" PrivateAssets="all" />
+  </ItemGroup>
+
   <ItemGroup>
     <PackageReference Include="coverlet.collector" Version="6.0.4">
       <IncludeAssets>runtime; build; native; contentfiles; analyzers; buildtransitive</IncludeAssets>

sdk_v2/js/src/imodel.ts

@@ -1,5 +1,6 @@
 import { ChatClient } from './openai/chatClient.js';
 import { AudioClient } from './openai/audioClient.js';
+import { LiveAudioTranscriptionClient } from './openai/liveAudioTranscriptionClient.js';
 import { ResponsesClient } from './openai/responsesClient.js';
 
 export interface IModel {
@@ -16,6 +17,13 @@ export interface IModel {
 
     createChatClient(): ChatClient;
     createAudioClient(): AudioClient;
+
+    /**
+     * Creates a LiveAudioTranscriptionClient for real-time audio streaming ASR.
+     * The model must be loaded before calling this method.
+     * @returns A LiveAudioTranscriptionClient instance.
+     */
+    createLiveTranscriptionClient(): LiveAudioTranscriptionClient;
     /**
      * Creates a ResponsesClient for interacting with the model via the Responses API.
      * Unlike createChatClient/createAudioClient (which use FFI), the Responses API

sdk_v2/js/src/index.ts

@@ -6,6 +6,8 @@ export { ModelVariant } from './modelVariant.js';
 export type { IModel } from './imodel.js';
 export { ChatClient, ChatClientSettings } from './openai/chatClient.js';
 export { AudioClient, AudioClientSettings } from './openai/audioClient.js';
+export { LiveAudioTranscriptionClient, LiveAudioTranscriptionSettings } from './openai/liveAudioTranscriptionClient.js';
+export type { LiveAudioTranscriptionResult, CoreErrorResponse } from './openai/liveAudioTranscriptionTypes.js';
 export { ResponsesClient, ResponsesClientSettings, getOutputText } from './openai/responsesClient.js';
 export { ModelLoadManager } from './detail/modelLoadManager.js';
 /** @internal */

sdk_v2/js/src/model.ts

@@ -1,6 +1,7 @@
 import { ModelVariant } from './modelVariant.js';
 import { ChatClient } from './openai/chatClient.js';
 import { AudioClient } from './openai/audioClient.js';
+import { LiveAudioTranscriptionClient } from './openai/liveAudioTranscriptionClient.js';
 import { ResponsesClient } from './openai/responsesClient.js';
 import { IModel } from './imodel.js';
 
@@ -148,6 +149,14 @@ export class Model implements IModel {
         return this.selectedVariant.createAudioClient();
     }
 
+    /**
+     * Creates a LiveAudioTranscriptionClient for real-time audio streaming ASR.
+     * @returns A LiveAudioTranscriptionClient instance.
+     */
+    public createLiveTranscriptionClient(): LiveAudioTranscriptionClient {
+        return this.selectedVariant.createLiveTranscriptionClient();
+    }
+
     /**
      * Creates a ResponsesClient for interacting with the model via the Responses API.
      * @param baseUrl - The base URL of the Foundry Local web service.

sdk_v2/js/src/modelVariant.ts

@@ -3,6 +3,7 @@ import { ModelLoadManager } from './detail/modelLoadManager.js';
 import { ModelInfo } from './types.js';
 import { ChatClient } from './openai/chatClient.js';
 import { AudioClient } from './openai/audioClient.js';
+import { LiveAudioTranscriptionClient } from './openai/liveAudioTranscriptionClient.js';
 import { ResponsesClient } from './openai/responsesClient.js';
 import { IModel } from './imodel.js';
 
@@ -129,6 +130,14 @@ export class ModelVariant implements IModel {
         return new AudioClient(this._modelInfo.id, this.coreInterop);
     }
 
+    /**
+     * Creates a LiveAudioTranscriptionClient for real-time audio streaming ASR.
+     * @returns A LiveAudioTranscriptionClient instance.
+     */
+    public createLiveTranscriptionClient(): LiveAudioTranscriptionClient {
+        return new LiveAudioTranscriptionClient(this._modelInfo.id, this.coreInterop);
+    }
+
     /**
      * Creates a ResponsesClient for interacting with the model via the Responses API.
      * @param baseUrl - The base URL of the Foundry Local web service.