添加OpenAI TTS功能

.devcontainer/devcontainer.json

@@ -0,0 +1,27 @@
+{
+  "name": "JA Learner Development",
+  "image": "mcr.microsoft.com/dotnet/sdk:6.0",
+  "features": {
+    "ghcr.io/devcontainers/features/node:1": {
+      "version": "lts"
+    },
+    "ghcr.io/devcontainers/features/git:1": {}
+  },
+  "postCreateCommand": "dotnet restore && npm install -g @vue/cli",
+  "customizations": {
+    "vscode": {
+      "extensions": [
+        "ms-dotnettools.csharp",
+        "ms-dotnettools.vscode-dotnet-runtime",
+        "Vue.volar",
+        "dbaeumer.vscode-eslint",
+        "ms-vscode.powershell"
+      ]
+    }
+  },
+  "forwardPorts": [
+    5000,
+    5001,
+    8080
+  ]
+}

.gitignore

@@ -398,4 +398,6 @@ FodyWeavers.xsd
 *.sln.iml
 
 config.txt
-extra_prompts/
+extra_prompts/
+dist_vue
+.vscode

GUI/MainForm.cs

@@ -1,6 +1,7 @@
 using Microsoft.Web.WebView2.Core;
 using System.Diagnostics;
 using Microsoft.VisualBasic;
+using NAudio.Wave;
 
 namespace ja_learner
 {
@@ -12,6 +13,7 @@ public partial class MainForm : Form
 
         TextAnalyzer textAnalyzer = new TextAnalyzer();
         private string sentence = "";
+        private Stream audioStream = null;
         private bool immersiveMode = false;
         public bool ImmersiveMode
         {
@@ -24,7 +26,7 @@ public bool ImmersiveMode
                     tabControl.Hide();
                     panel1.Hide();
                     FormBorderStyle = FormBorderStyle.None;
-                    MinimumSize = new Size(0,0);
+                    MinimumSize = new Size(0, 0);
                 }
                 else
                 {
@@ -43,6 +45,7 @@ public string Sentence
             set
             {
                 sentence = value;
+                audioStream = null;
                 dictForm.UpdateTranslationPanelText(sentence);
                 UpdateMecabResult(RunMecab());
                 if (checkBoxAutoTranslate.Checked)
@@ -88,7 +91,7 @@ private async void MainForm_Load(object sender, EventArgs e)
             dictForm.Show();
             dictForm.Hide();
             UpdateExtraPromptCombobox();
-            
+
             // 初始化 MainForm
             if (Program.APP_SETTING.HttpProxy != string.Empty)
             {
@@ -437,5 +440,29 @@ private void checkBoxUseProxy_CheckedChanged(object sender, EventArgs e)
             UserConfig.UseProxy = checkBoxUseProxy.Checked;
             GptCaller.SetProxy(UserConfig.UseProxy);
         }
+
+        private async void buttonRead_Click(object sender, EventArgs e)
+        {
+            buttonRead.Text = "朗读(请求中)";
+            if (audioStream == null)
+            {
+                audioStream = await GptCaller.CreateTextToSpeechStream(sentence);
+            }
+
+            buttonRead.Text = "朗读(播放中)";
+            using (var mp3Reader = new Mp3FileReader(audioStream))
+            using (var waveOut = new WaveOutEvent())
+            {
+                waveOut.Init(mp3Reader);
+                waveOut.Play();
+                while (waveOut.PlaybackState == PlaybackState.Playing)
+                {
+                    await Task.Delay(100);
+                }
+            }
+
+            audioStream.Position = 0;
+            buttonRead.Text = "朗读";
+        }
     }
 }

GUI/TranslationPanel.cs

@@ -10,6 +10,8 @@
 using System.Threading.Tasks;
 using System.Windows.Forms;
 using static System.Net.Mime.MediaTypeNames;
+using Markdig;
+using System.Web;
 
 namespace ja_learner.GUI
 {
@@ -27,13 +29,14 @@ public void UpdateText(string text)
 
         private void buttonInterpret_Click(object sender, EventArgs e)
         {
-            textBoxResult.Text = "";
+            var markdownContent = "";
             buttonInterpret.Enabled = false;
             var chat = GptCaller.CreateInterpretConversation(textBoxSentence.Text);
+
             GptCaller.StreamResponse(chat, res =>
             {
-                textBoxResult.Text += res.Replace("\n", "\r\n");
-                textBoxResult.ScrollToCaret();
+                markdownContent += res;
+                webBrowserResult.Document.InvokeScript("setMarkdown", new object[] { Markdown.ToHtml(markdownContent) });
             });
 
             buttonInterpret.Enabled = true;

GptCaller.cs

@@ -7,13 +7,18 @@ namespace ja_learner
     internal class GptCaller
     {
         private static OpenAIAPI api;
-
+        private static OpenAIAPI ttsApi;
         private static IHttpClientFactory defaultFactory;
         private static IHttpClientFactory proxyFactory;
 
         public static void Initialize()
         {
             api = new(Program.APP_SETTING.GPT.ApiKey) { ApiUrlFormat = Program.APP_SETTING.GPT.ApiUrl };
+            ttsApi = api;
+            if (Program.APP_SETTING.GPT.TtsApiKey != string.Empty || Program.APP_SETTING.GPT.TtsApiUrl != string.Empty)
+            {
+                ttsApi = new(Program.APP_SETTING.GPT.TtsApiKey) { ApiUrlFormat = Program.APP_SETTING.GPT.TtsApiUrl };
+            }
             defaultFactory = api.HttpClientFactory;
             proxyFactory = new MyHttpClientFactory(Program.APP_SETTING.HttpProxy);
         }
@@ -23,16 +28,19 @@ public static void SetProxy(bool useProxy)
             if (useProxy)
             {
                 api.HttpClientFactory = proxyFactory;
+                ttsApi.HttpClientFactory = proxyFactory;
             }
             else
             {
                 api.HttpClientFactory = defaultFactory;
+                ttsApi.HttpClientFactory = defaultFactory;
             }
         }
 
         public static Conversation CreateTranslateConversation(string text)
         {
             Conversation conversation = api.Chat.CreateConversation();
+            conversation.Model = Program.APP_SETTING.GPT.Model;
             conversation.AppendSystemMessage(Program.APP_SETTING.GPT.TranslatePrompt);
             if (UserConfig.useExtraPrompt)
             {
@@ -45,6 +53,7 @@ public static Conversation CreateTranslateConversation(string text)
         public static Conversation CreateInterpretConversation(string text)
         {
             Conversation conversation = api.Chat.CreateConversation();
+            conversation.Model = Program.APP_SETTING.GPT.Model;
             conversation.AppendSystemMessage(Program.APP_SETTING.GPT.ExplainPrompt);
             if (UserConfig.useExtraPrompt)
             {
@@ -54,6 +63,12 @@ public static Conversation CreateInterpretConversation(string text)
             return conversation;
         }
 
+        public static Task<Stream> CreateTextToSpeechStream(string text)
+        {
+            return ttsApi.TextToSpeech.GetSpeechAsStreamAsync(
+                text, Program.APP_SETTING.GPT.Voice, Program.APP_SETTING.GPT.VoiceSpeed);
+        }
+
         private static void AddExtraSystemPrompt(Conversation conversation)
         {
             if (UserConfig.ExtraPrompt.Length > 0)

Model/AppSetting.cs

@@ -13,6 +13,11 @@ public class GPTOptions
 {
     public string ApiKey { get; set; } = string.Empty;
     public string ApiUrl { get; set; } = string.Empty;
+    public string TtsApiKey { get; set; } = string.Empty;
+    public string TtsApiUrl { get; set; } = string.Empty;
+    public string Voice { get; set; } = string.Empty;
+    public double VoiceSpeed { get; set; } = 1.0;
+    public string Model { get; set; } = string.Empty;
     public string ExtraPromptDir { get; set; } = string.Empty;
     public string TranslatePrompt { get; set; } = string.Empty;
     public string ExplainPrompt { get; set; } = string.Empty;

README.md

@@ -76,6 +76,11 @@
 "GPT": {
     "ApiKey": "sk-xxx",
     "ApiUrl": "https://api.openai.com/{0}/{1}", // 实际调用为https://api.openai.com/v1/chat/completions
+    "Model": "gpt-4o-mini", // 设置模型
+    "TtsApiKey": "", // 可以单独设置OpenAI AI文字转语音key,适用于不同key价格更便宜的场景,不设置的话仍用上面的配置
+    "TtsApiUrl": "", // 单独的TTS API URL
+    "Voice": "shimmer", // 发声角色,可选值为alloy,ash,coral,echo,fable,onyx,nova,sage,shimmer
+    "VoiceSpeed": "1.0", // OpenAI TTS语速
     "ExtraPromptDir": "extra_prompts",		// 额外的Prompt，比如指定某些角色名字怎么翻译
     "TranslatePrompt": "...",			// 翻译Prompt
     "ExplainPrompt": "..."			// 分析Prompt

appsettings.json

@@ -3,8 +3,13 @@
 		"Port": 8080,
 		"HttpProxy": "",
 		"GPT": {
-		    "ApiKey": "sk-xxx",
+			"ApiKey": "sk-xxx",
 			"ApiUrl": "https://api.openai.com/{0}/{1}",
+			"Model": "gpt-4o-mini",
+		    "TtsApiKey": "",
+			"TtsApiUrl": "",
+			"Voice": "shimmer",
+			"VoiceSpeed": "1.0",
 			"ExtraPromptDir": "extra_prompts",
 			"TranslatePrompt": "Translate the text to Simplified Chinese. Don't output anything other than translation results.",
 			"ExplainPrompt": "List and explain the vocabulary (except prepositions) and grammar of the given Japanese text in Simplified Chinese. Your output consists of three parts: translation, vocabulary, grammar. Don't use English and romaji."

ja-learner.csproj

@@ -10,14 +10,17 @@
     <_SuppressWinFormsTrimError>true</_SuppressWinFormsTrimError>
     <ImplicitUsings>enable</ImplicitUsings>
     <GenerateDocumentationFile>False</GenerateDocumentationFile>
-	<MeCabUseDefaultDictionary>True</MeCabUseDefaultDictionary>
+    <MeCabUseDefaultDictionary>True</MeCabUseDefaultDictionary>
+    <BuiltInComInteropSupport>true</BuiltInComInteropSupport>
   </PropertyGroup>
 
   <ItemGroup>
     <PackageReference Include="MeCab.DotNet" Version="1.2.0" />
     <PackageReference Include="Microsoft.Extensions.Hosting" Version="8.0.0" />
     <PackageReference Include="Microsoft.Web.WebView2" Version="1.0.1774.30" />
-    <PackageReference Include="OpenAI" Version="1.7.2" />
+    <PackageReference Include="NAudio" Version="2.2.1" />
+    <PackageReference Include="OpenAI" Version="1.11.0" />
+    <PackageReference Include="Markdig" Version="0.30.4" />
   </ItemGroup>
 
   <ItemGroup>