Fix for non English language response from Whisper

.gitignore

@@ -160,3 +160,5 @@ OpenAPI_key.txt
 image_edit_mask_.png
 image_edit_original_.png
 instructions.txt
+/Playgrounds/Transcription/Male voices
+/Playgrounds/Transcription/Male voices.zip

Forge.OpenAI/Forge.OpenAI.csproj

@@ -9,9 +9,9 @@
 
 	<PropertyGroup>
 		<!--TargetFrameworks>net6.0</TargetFrameworks-->
-		<TargetFrameworks>net461;netstandard2.0;netcoreapp3.1;net6.0;net7.0</TargetFrameworks>
+		<TargetFrameworks>netstandard2.0;net6.0;net7.0</TargetFrameworks>
 		<SignAssembly>true</SignAssembly>
-		<AssemblyOriginatorKeyFile>..\..\.Documents\Forge.pfx</AssemblyOriginatorKeyFile>
+		<!--<AssemblyOriginatorKeyFile>..\..\.Documents\Forge.pfx</AssemblyOriginatorKeyFile>-->
 		<ProjectGuid>{D1DDE02E-1865-4173-8F3C-153D99AB394F}</ProjectGuid>
 		<DocumentationFile>Forge.OpenAI.xml</DocumentationFile>
 	</PropertyGroup>

Forge.OpenAI/Infrastructure/ApiHttpService.cs

@@ -6,6 +6,7 @@
 using Microsoft.Extensions.Options;
 using System;
 using System.Collections.Generic;
+using System.ComponentModel;
 using System.IO;
 using System.Linq;
 using System.Net.Http;
@@ -411,16 +412,20 @@ private async Task<HttpOperationResult<TResult>> ApiCall<TData, TResult>(HttpMet
                             {
                                 if (typeof(string).IsAssignableFrom(typeof(TResult)))
                                 {
+                                    //var jsonResult = await response.Content.ReadAsStringAsync().ConfigureAwait(false);
                                     result = new HttpOperationResult<TResult>(jsonResult as TResult);
                                 }
                                 else
                                 {
-                                    result = new HttpOperationResult<TResult>(JsonSerializer.Deserialize<TResult>(jsonResult, _options.JsonSerializerOptions));
+                                    var contentAsBytes = UTF8Encoding.UTF8.GetBytes(jsonResult);
+                                    var jsonObj = GetUTF8Object<TResult>(contentAsBytes);
+                                    result = new HttpOperationResult<TResult>(jsonObj);
                                     SetResponseData(response, result.Result);
                                 }
                             }
                             else
                             {
+                               // var jsonResult = await response.Content.ReadAsStringAsync().ConfigureAwait(false);
                                 _logger?.LogDebug($"ApiCall, response indicates an unsuccessful operation from {httpClient.BaseAddress}{uri}, method: {httpMethod.Method}");
 
                                 result = new HttpOperationResult<TResult>(new Exception(response.StatusCode.ToString(), new Exception(jsonResult)), response.StatusCode, jsonResult);
@@ -434,7 +439,11 @@ private async Task<HttpOperationResult<TResult>> ApiCall<TData, TResult>(HttpMet
                 return result;
             });
         }
-
+        private TResult GetUTF8Object<TResult>(byte[] responseBytes)
+        {
+            Utf8JsonReader reader = new Utf8JsonReader(responseBytes);
+            return JsonSerializer.Deserialize<TResult>(ref reader, _options.JsonSerializerOptions);
+        }
         /// <summary>Perform the API call in streaming mode the asynchronously.</summary>
         /// <typeparam name="TData">The type of the data.</typeparam>
         /// <typeparam name="TResult">The type of the result.</typeparam>

Forge.OpenAI/Interfaces/Services/ITranscriptionService.cs

@@ -17,6 +17,13 @@ public interface ITranscriptionService
         ///   TranscriptionResponse
         /// </returns>
         Task<HttpOperationResult<TranscriptionResponse>> GetAsync(TranscriptionRequest request, CancellationToken cancellationToken);
+        /// <summary>
+        /// This method is useful
+        /// </summary>
+        /// <param name="request"></param>
+        /// <param name="cancellationToken"></param>
+        /// <returns></returns>
+        //Task<HttpOperationResult<TranscriptionResponse>> GetResponseAsStringAsync(TranscriptionRequest request, CancellationToken cancellationToken);
 
     }
 

Forge.OpenAI/Services/TranscriptionService.cs

@@ -92,7 +92,7 @@ private async Task<HttpContent> TranscriptHttpContentFactoryAsync(TranscriptionR
 
             if (!string.IsNullOrWhiteSpace(request.Prompt)) content.Add(new StringContent(request.Prompt), "prompt");
             if (!string.IsNullOrWhiteSpace(request.ResponseFormat)) content.Add(new StringContent(request.ResponseFormat), "response_format");
-            if (!string.IsNullOrWhiteSpace(request.Language)) content.Add(new StringContent(request.ResponseFormat), "language");
+            if (!string.IsNullOrWhiteSpace(request.Language)) content.Add(new StringContent(request.Language), "language");
             if (request.Temperature.HasValue) content.Add(new StringContent(request.Temperature.Value.ToString()), "temperature");
 
             return content;

Forge.OpenAI/Settings/OpenAIDefaultOptions.cs

@@ -160,7 +160,9 @@ public static class OpenAIDefaultOptions
         /// <summary>The default json serializer options</summary>
         public static JsonSerializerOptions DefaultJsonSerializerOptions { get; set; } = new JsonSerializerOptions
         {
-            DefaultIgnoreCondition = JsonIgnoreCondition.WhenWritingNull
+            DefaultIgnoreCondition = JsonIgnoreCondition.WhenWritingNull,
+            Encoder = System.Text.Encodings.Web.JavaScriptEncoder.UnsafeRelaxedJsonEscaping
+
         };
 
     }

Playgrounds/Transcription/Program.cs

@@ -5,12 +5,22 @@
 using Forge.OpenAI.Models.Common;
 using Microsoft.Extensions.DependencyInjection;
 using Microsoft.Extensions.Hosting;
+using System.Text;
 
 namespace Transcription
 {
 
     internal class Program
     {
+        public static string DecodeFromUtf16ToUtf8(string utf16String)
+        {
+            // copy the string as UTF-8 bytes.
+            byte[] utf8Bytes = new byte[utf16String.Length];
+            for (int i = 0; i < utf16String.Length; ++i)
+                utf8Bytes[i] = (byte)utf16String[i];
+
+            return Encoding.UTF8.GetString(utf8Bytes, 0, utf8Bytes.Length);
+        }
 
         static async Task Main(string[] args)
         {
@@ -27,10 +37,11 @@ static async Task Main(string[] args)
 
             TranscriptionRequest request = new TranscriptionRequest();
             request.AudioFile = new BinaryContentData() { ContentName = "audio.mp3", SourceStream = File.OpenRead("audio.mp3") };
-
+            //request.Language = "ta";
             HttpOperationResult<TranscriptionResponse> response = await openAi.TranscriptionService.GetAsync(request, CancellationToken.None).ConfigureAwait(false);
             if (response.IsSuccess)
             {
+                Console.OutputEncoding = Encoding.UTF8;
                 Console.WriteLine(response.Result?.Text);
             }
             else

Playgrounds/Transcription/Transcription.csproj

@@ -15,6 +15,7 @@
 	</PropertyGroup>
 
 	<ItemGroup>
+		<!--<PackageReference Include="Forge.OpenAI" Version="1.1.1" />-->
 		<PackageReference Include="Microsoft.Extensions.Configuration" Version="7.0.0" />
 		<PackageReference Include="Microsoft.Extensions.Configuration.Binder" Version="7.0.2" />
 		<PackageReference Include="Microsoft.Extensions.Configuration.Json" Version="7.0.0" />
@@ -24,16 +25,21 @@
 	</ItemGroup>
 
 	<ItemGroup>
-		<ProjectReference Include="..\..\Forge.OpenAI\Forge.OpenAI.csproj" />
+	  <ProjectReference Include="..\..\Forge.OpenAI\Forge.OpenAI.csproj" />
 	</ItemGroup>
 
+
+
 	<ItemGroup>
 		<None Update="appsettings.json">
 			<CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
 		</None>
 		<None Update="audio.mp3">
 			<CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
 		</None>
+		<None Update="Sample.mp3">
+		  <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+		</None>
 	</ItemGroup>
 
 </Project>