语音播报

api/go.mod

@@ -50,6 +50,7 @@ require (
 	github.com/mattn/go-colorable v0.1.13 // indirect
 	github.com/pilu/config v0.0.0-20131214182432-3eb99e6c0b9a // indirect
 	github.com/pilu/fresh v0.0.0-20240621171608-8d1fef547a99 // indirect
+	github.com/sashabaranov/go-openai v1.38.1 // indirect
 	github.com/tklauser/go-sysconf v0.3.13 // indirect
 	github.com/tklauser/numcpus v0.7.0 // indirect
 	github.com/yusufpapurcu/wmi v1.2.4 // indirect

api/go.sum

@@ -203,6 +203,8 @@ github.com/rogpeppe/go-internal v1.8.0 h1:FCbCCtXNOY3UtUuHUYaghJg4y7Fd14rXifAYUA
 github.com/rogpeppe/go-internal v1.8.0/go.mod h1:WmiCO8CzOY8rg0OYDC4/i/2WRWAB6poM+XZ2dLUbcbE=
 github.com/rs/xid v1.5.0 h1:mKX4bl4iPYJtEIxp6CYiUuLQ/8DYMoz0PUdtGgMFRVc=
 github.com/rs/xid v1.5.0/go.mod h1:trrq9SKmegXys3aeAKXMUTdJsYXVwGY3RLcfgqegfbg=
+github.com/sashabaranov/go-openai v1.38.1 h1:TtZabbFQZa1nEni/IhVtDF/WQjVqDgd+cWR5OeddzF8=
+github.com/sashabaranov/go-openai v1.38.1/go.mod h1:lj5b/K+zjTSFxVLijLSTDZuP7adOgerWeFyZLUhAKRg=
 github.com/shirou/gopsutil v3.21.11+incompatible h1:+1+c1VGhc88SSonWP6foOcLhvnKlUeu/erjjvaPEYiI=
 github.com/shirou/gopsutil v3.21.11+incompatible/go.mod h1:5b4v6he4MtMOwMlS0TUMTu2PcXUg8+E1lC7eC3UO/RA=
 github.com/shopspring/decimal v1.3.1 h1:2Usl1nmF/WZucqkFZhnfFYxxxu8LG21F6nPQBE5gKV8=

api/handler/chat_handler.go

@@ -22,15 +22,16 @@ import (
 	"geekai/utils"
 	"geekai/utils/resp"
 	"html/template"
+	"io"
 	"net/http"
 	"net/url"
-	"regexp"
 	"strings"
 	"time"
 	"unicode/utf8"
 
 	"github.com/gin-gonic/gin"
 	"github.com/go-redis/redis/v8"
+	"github.com/sashabaranov/go-openai"
 	"gorm.io/gorm"
 )
 
@@ -501,28 +502,50 @@ func (h *ChatHandler) saveChatHistory(
 	}
 }
 
-// 将AI回复消息中生成的图片链接下载到本地
+// 文本生成语音
-func (h *ChatHandler) extractImgUrl(text string) string {
+func (h *ChatHandler) TextToSpeech(c *gin.Context) {
-	pattern := `!\[([^\]]*)]\(([^)]+)\)`
+	var data struct {
-	re := regexp.MustCompile(pattern)
+		Text string `json:"text"`
-	matches := re.FindAllStringSubmatch(text, -1)
+	}
-
+	if err := c.ShouldBindJSON(&data); err != nil {
-	// 下载图片并替换链接地址
+		resp.ERROR(c, types.InvalidArgs)
-	for _, match := range matches {
+		return
-		imageURL := match[2]
+	}
-		logger.Debug(imageURL)
+
-		// 对于相同地址的图片，已经被替换了，就不再重复下载了
+	// 调用 DeepSeek 的 API 接口
-		if !strings.Contains(text, imageURL) {
+	var apiKey model.ApiKey
-			continue
+	h.DB.Where("type", "chat").Where("enabled", true).First(&apiKey)
-		}
+	if apiKey.Id == 0 {
-
+		resp.ERROR(c, "no available key, please import key")
-		newImgURL, err := h.uploadManager.GetUploadHandler().PutUrlFile(imageURL, false)
+		return
-		if err != nil {
+	}
-			logger.Error("error with download image: ", err)
+
-			continue
+	// 调用 openai tts api
-		}
+	config := openai.DefaultConfig(apiKey.Value)
-
+	config.BaseURL = apiKey.ApiURL
-		text = strings.ReplaceAll(text, imageURL, newImgURL)
+	client := openai.NewClientWithConfig(config)
+	req := openai.CreateSpeechRequest{
+		Model: openai.TTSModel1,
+		Input: data.Text,
+		Voice: openai.VoiceAlloy,
+	}
+
+	audioData, err := client.CreateSpeech(context.Background(), req)
+	if err != nil {
+		logger.Error("failed to create speech: ", err)
+		resp.ERROR(c, "failed to create speech")
+		return
+	}
+
+	// 设置响应头
+	c.Header("Content-Type", "audio/mpeg")
+	c.Header("Content-Disposition", "attachment; filename=speech.mp3")
+
+	// 将音频数据写入响应
+	_, err = io.Copy(c.Writer, audioData)
+	if err != nil {
+		logger.Error("failed to write audio data: ", err)
+		resp.ERROR(c, "failed to write audio data")
+		return
 	}
-	return text
 }

api/main.go

@@ -251,6 +251,7 @@ func main() {
 			group.GET("clear", h.Clear)
 			group.POST("tokens", h.Tokens)
 			group.GET("stop", h.StopGenerate)
+			group.POST("tts", h.TextToSpeech)
 		}),
 		fx.Invoke(func(s *core.AppServer, h *handler.NetHandler) {
 			s.Engine.POST("/api/upload", h.Upload)

web/src/components/ChatReply.vue

@@ -82,6 +82,7 @@
                 <i class="iconfont icon-speaker"></i>
               </el-tooltip>
             </span>
+            <img src="/images/voice.gif" />
           </span>
         </div>
       </div>
@@ -97,6 +98,8 @@ import hl from "highlight.js";
 import emoji from "markdown-it-emoji";
 import mathjaxPlugin from "markdown-it-mathjax3";
 import MarkdownIt from "markdown-it";
+import { httpPost } from "@/utils/http";
+import RippleButton from "./ui/RippleButton.vue";
 
 // eslint-disable-next-line no-undef,no-unused-vars
 const props = defineProps({
@@ -156,7 +159,26 @@ if (!props.data.icon) {
 
 const synthesis = (text) => {
   console.log(text);
-  ElMessage.info("语音合成功能暂不可用");
+  // 生成语音
+  httpPost("/api/chat/tts", { text }, { responseType: 'blob' }).then(response => {
+    // 创建 Blob 对象，明确指定 MIME 类型
+    const blob = new Blob([response], { type: 'audio/wav' });
+    // 创建 URL
+    const audioUrl = URL.createObjectURL(blob);
+    // 创建音频元素
+    const audio = new Audio(audioUrl);
+    // 播放音频
+    audio.play().then(() => {
+      // 播放完成后释放 URL
+      URL.revokeObjectURL(audioUrl);
+    }).catch(err => {
+      console.error('播放音频失败:', err);
+      ElMessage.error('音频播放失败，请检查浏览器是否支持该音频格式');
+    });
+  }).catch(err => {
+    console.error('语音合成请求失败:', err);
+    ElMessage.error('语音合成失败，请稍后重试');
+  });
 };
 
 // 重新生成
@@ -168,6 +190,7 @@ const reGenerate = (prompt) => {
 
 <style lang="stylus">
 @import '@/assets/css/markdown/vue.css';
+
 .chat-page,.chat-export {
   --font-family: Menlo,"微软雅黑","Roboto Mono","Courier New",Courier,monospace,"Inter",sans-serif;
   font-family: var(--font-family);