.NET: Support custom audio-to-text endpoints

dotnet/samples/Concepts/AudioToText/OpenAI_AudioToText.cs

@@ -50,4 +50,26 @@ public async Task AudioToTextAsync()
         // Output the transcribed text
         Console.WriteLine(textContent.Text);
     }
+
+    [Fact(Skip = "Requires a local OpenAI-compatible audio-to-text server, such as FunASR.")]
+    public async Task OpenAICompatibleAudioToTextAsync()
+    {
+        // FunASR exposes the OpenAI-compatible endpoint at /v1/audio/transcriptions.
+        var kernel = Kernel.CreateBuilder()
+            .AddOpenAIAudioToText(
+                modelId: "sensevoice",
+                endpoint: new Uri("http://localhost:8000/v1"))
+            .Build();
+
+        var audioToTextService = kernel.GetRequiredService<IAudioToTextService>();
+        await using var audioFileStream = EmbeddedResource.ReadStream(AudioFilename);
+        var audioFileBinaryData = await BinaryData.FromStreamAsync(audioFileStream!);
+        AudioContent audioContent = new(audioFileBinaryData, mimeType: null);
+
+        var textContent = await audioToTextService.GetTextContentAsync(
+            audioContent,
+            new OpenAIAudioToTextExecutionSettings(AudioFilename));
+
+        Console.WriteLine(textContent.Text);
+    }
 }

dotnet/src/Connectors/Connectors.OpenAI.UnitTests/Extensions/KernelBuilderExtensionsTests.cs

@@ -129,6 +129,32 @@ public void ItCanAddAudioToTextService()
         Assert.Equal("model", service.Attributes[AIServiceExtensions.ModelIdKey]);
     }
 
+    [Fact]
+    public void ItCanAddAudioToTextServiceWithCustomEndpoint()
+    {
+        // Arrange
+        var endpoint = new Uri("http://localhost:10095/v1");
+        var sut = Kernel.CreateBuilder();
+
+        // Act
+        var service = sut.AddOpenAIAudioToText("model", endpoint)
+            .Build()
+            .GetRequiredService<IAudioToTextService>();
+
+        // Assert
+        Assert.Equal(endpoint.ToString(), service.Attributes[AIServiceExtensions.EndpointKey]);
+    }
+
+    [Fact]
+    public void ItThrowsWhenAddingAudioToTextServiceWithoutCustomEndpoint()
+    {
+        // Arrange
+        var sut = Kernel.CreateBuilder();
+
+        // Act & Assert
+        Assert.Throws<ArgumentNullException>(() => sut.AddOpenAIAudioToText("model", endpoint: null!));
+    }
+
     [Fact]
     public void ItCanAddAudioToTextServiceWithOpenAIClient()
     {

dotnet/src/Connectors/Connectors.OpenAI.UnitTests/Extensions/ServiceCollectionExtensionsTests.cs

@@ -157,6 +157,32 @@ public void ItCanAddAudioToTextService()
         Assert.Equal("model", service.Attributes[AIServiceExtensions.ModelIdKey]);
     }
 
+    [Fact]
+    public void ItCanAddAudioToTextServiceWithCustomEndpoint()
+    {
+        // Arrange
+        var endpoint = new Uri("http://localhost:10095/v1");
+        var sut = new ServiceCollection();
+
+        // Act
+        var service = sut.AddOpenAIAudioToText("model", endpoint)
+            .BuildServiceProvider()
+            .GetRequiredService<IAudioToTextService>();
+
+        // Assert
+        Assert.Equal(endpoint.ToString(), service.Attributes[AIServiceExtensions.EndpointKey]);
+    }
+
+    [Fact]
+    public void ItThrowsWhenAddingAudioToTextServiceWithoutCustomEndpoint()
+    {
+        // Arrange
+        var sut = new ServiceCollection();
+
+        // Act & Assert
+        Assert.Throws<ArgumentNullException>(() => sut.AddOpenAIAudioToText("model", endpoint: null!));
+    }
+
     [Fact]
     public void ItCanAddAudioToTextServiceWithOpenAIClient()
     {

dotnet/src/Connectors/Connectors.OpenAI.UnitTests/Services/OpenAIAudioToTextServiceTests.cs

@@ -45,6 +45,35 @@ public void ConstructorWithApiKeyWorksCorrectly(bool includeLoggerFactory)
         Assert.Equal("model-id", service.Attributes["ModelId"]);
     }
 
+    [Fact]
+    public async Task ItUsesCustomEndpointWithoutApiKeyAsync()
+    {
+        // Arrange
+        var endpoint = new Uri("http://localhost:10095/v1");
+        var service = new OpenAIAudioToTextService("model-id", endpoint, httpClient: this._httpClient);
+        this._messageHandlerStub.ResponseToReturn = new HttpResponseMessage(System.Net.HttpStatusCode.OK)
+        {
+            Content = new StringContent("Test audio-to-text response")
+        };
+
+        // Act
+        await service.GetTextContentsAsync(
+            new AudioContent(new BinaryData("data"), mimeType: null),
+            new OpenAIAudioToTextExecutionSettings("file.mp3"));
+
+        // Assert
+        Assert.Equal("http://localhost:10095/v1/audio/transcriptions", this._messageHandlerStub.RequestUri!.ToString());
+        Assert.Equal(endpoint.ToString(), service.Attributes["Endpoint"]);
+    }
+
+    [Fact]
+    public void ItThrowsIfCustomEndpointIsNotProvided()
+    {
+        // Act & Assert
+        var exception = Assert.Throws<ArgumentNullException>(() => new OpenAIAudioToTextService("model-id", endpoint: null!));
+        Assert.Equal("endpoint", exception.ParamName);
+    }
+
     [Fact]
     public void ItThrowsIfModelIdIsNotProvided()
     {

dotnet/src/Connectors/Connectors.OpenAI/Extensions/OpenAIKernelBuilderExtensions.cs

@@ -270,6 +270,44 @@ public static IKernelBuilder AddOpenAIAudioToText(
         return builder;
     }
 
+    /// <summary>
+    /// Adds an OpenAI-compatible audio-to-text service with a custom endpoint to the list.
+    /// </summary>
+    /// <param name="builder">The <see cref="IKernelBuilder"/> instance to augment.</param>
+    /// <param name="modelId">Model name.</param>
+    /// <param name="endpoint">OpenAI-compatible API endpoint.</param>
+    /// <param name="apiKey">Optional API key.</param>
+    /// <param name="orgId">OpenAI organization id.</param>
+    /// <param name="serviceId">A local identifier for the given AI service.</param>
+    /// <param name="httpClient">The HttpClient to use with this service.</param>
+    /// <returns>The same instance as <paramref name="builder"/>.</returns>
+    [Experimental("SKEXP0010")]
+    public static IKernelBuilder AddOpenAIAudioToText(
+        this IKernelBuilder builder,
+        string modelId,
+        Uri endpoint,
+        string? apiKey = null,
+        string? orgId = null,
+        string? serviceId = null,
+        HttpClient? httpClient = null)
+    {
+        Verify.NotNull(builder);
+        Verify.NotNullOrWhiteSpace(modelId);
+        Verify.NotNull(endpoint);
+
+        Func<IServiceProvider, object?, OpenAIAudioToTextService> factory = (serviceProvider, _) =>
+            new(modelId,
+                endpoint,
+                apiKey,
+                orgId,
+                HttpClientProvider.GetHttpClient(httpClient, serviceProvider),
+                serviceProvider.GetService<ILoggerFactory>());
+
+        builder.Services.AddKeyedSingleton<IAudioToTextService>(serviceId, factory);
+
+        return builder;
+    }
+
     /// <summary>
     /// Adds the OpenAI audio-to-text service to the list.
     /// </summary>

dotnet/src/Connectors/Connectors.OpenAI/Extensions/OpenAIServiceCollectionExtensions.cs

@@ -186,6 +186,42 @@ public static IServiceCollection AddOpenAIAudioToText(
         return services;
     }
 
+    /// <summary>
+    /// Adds an OpenAI-compatible audio-to-text service with a custom endpoint to the list.
+    /// </summary>
+    /// <param name="services">The <see cref="IServiceCollection"/> instance to augment.</param>
+    /// <param name="modelId">Model name.</param>
+    /// <param name="endpoint">OpenAI-compatible API endpoint.</param>
+    /// <param name="apiKey">Optional API key.</param>
+    /// <param name="orgId">OpenAI organization id.</param>
+    /// <param name="serviceId">A local identifier for the given AI service.</param>
+    /// <returns>The same instance as <paramref name="services"/>.</returns>
+    [Experimental("SKEXP0010")]
+    public static IServiceCollection AddOpenAIAudioToText(
+        this IServiceCollection services,
+        string modelId,
+        Uri endpoint,
+        string? apiKey = null,
+        string? orgId = null,
+        string? serviceId = null)
+    {
+        Verify.NotNull(services);
+        Verify.NotNullOrWhiteSpace(modelId);
+        Verify.NotNull(endpoint);
+
+        Func<IServiceProvider, object?, OpenAIAudioToTextService> factory = (serviceProvider, _) =>
+            new(modelId,
+                endpoint,
+                apiKey,
+                orgId,
+                HttpClientProvider.GetHttpClient(serviceProvider),
+                serviceProvider.GetService<ILoggerFactory>());
+
+        services.AddKeyedSingleton<IAudioToTextService>(serviceId, factory);
+
+        return services;
+    }
+
     /// <summary>
     /// Adds the OpenAI audio-to-text service to the list.
     /// </summary>

dotnet/src/Connectors/Connectors.OpenAI/Services/OpenAIAudioToTextService.cs

@@ -1,5 +1,6 @@
 // Copyright (c) Microsoft. All rights reserved.
 
+using System;
 using System.Collections.Generic;
 using System.Diagnostics.CodeAnalysis;
 using System.Net.Http;
@@ -12,13 +13,13 @@
 namespace Microsoft.SemanticKernel.Connectors.OpenAI;
 
 /// <summary>
-/// OpenAI text-to-audio service.
+/// OpenAI audio-to-text service.
 /// </summary>
 [Experimental("SKEXP0010")]
 public sealed class OpenAIAudioToTextService : IAudioToTextService
 {
     /// <summary>
-    /// OpenAI text-to-audio client for HTTP operations.
+    /// OpenAI audio-to-text client for HTTP operations.
     /// </summary>
     private readonly ClientCore _client;
 
@@ -44,6 +45,28 @@ public OpenAIAudioToTextService(
         this._client = new(modelId, apiKey, organization, null, httpClient, loggerFactory?.CreateLogger(typeof(OpenAIAudioToTextService)));
     }
 
+    /// <summary>
+    /// Initializes a new instance of the <see cref="OpenAIAudioToTextService"/> class for a custom OpenAI-compatible endpoint.
+    /// </summary>
+    /// <param name="modelId">Model name.</param>
+    /// <param name="endpoint">OpenAI-compatible API endpoint.</param>
+    /// <param name="apiKey">Optional API key.</param>
+    /// <param name="organization">OpenAI Organization Id (usually optional).</param>
+    /// <param name="httpClient">Custom <see cref="HttpClient"/> for HTTP requests.</param>
+    /// <param name="loggerFactory">The <see cref="ILoggerFactory"/> to use for logging. If null, no logging will be performed.</param>
+    public OpenAIAudioToTextService(
+        string modelId,
+        Uri endpoint,
+        string? apiKey = null,
+        string? organization = null,
+        HttpClient? httpClient = null,
+        ILoggerFactory? loggerFactory = null)
+    {
+        Verify.NotNullOrWhiteSpace(modelId, nameof(modelId));
+        Verify.NotNull(endpoint);
+        this._client = new(modelId, apiKey, organization, endpoint, httpClient, loggerFactory?.CreateLogger(typeof(OpenAIAudioToTextService)));
+    }
+
     /// <summary>
     /// Initializes a new instance of the <see cref="OpenAIAudioToTextService"/> class.
     /// </summary>