添加tts功能代码

src/main/java/com/supervision/controller/DigitalHumanController.java

@@ -5,9 +5,12 @@ import com.supervision.domain.LivetalkingChatDTO;
 import com.supervision.dto.DigitalHumanDTO;
 import com.supervision.dto.DigitalHumanVoiceDTO;
 import com.supervision.dto.R;
+import com.supervision.service.TTsService;
 import com.supervision.service.danmaku.DigitalHumanManageService;
 import lombok.RequiredArgsConstructor;
 import org.springframework.web.bind.annotation.*;
+import org.springframework.web.multipart.MultipartFile;
+import java.io.IOException;
 
 /**
  * 数字人控制器
@@ -19,6 +22,8 @@ public class DigitalHumanController {
 
     private final DigitalHumanManageService digitalHumanManageService;
 
+    private final TTsService tTsService;
+
     /**
      * 分页查询数字人列表
      * @param page 当前页码
@@ -49,4 +54,9 @@ public class DigitalHumanController {
         digitalHumanManageService.chatCallBack(digitalHumanVoiceDTO);
         return R.ok();
     }
+    @PostMapping("/livetalking/asr")
+    public R<String> tts(@RequestPart("file") MultipartFile file) throws IOException {
+        String stt = tTsService.stt(file.getOriginalFilename(), file.getBytes());
+        return R.ok(stt);
+    }
 }

src/main/java/com/supervision/service/TTsService.java

@@ -0,0 +1,6 @@
+package com.supervision.service;
+
+public interface TTsService {
+
+    String stt(String model,byte[] audioData);
+}

src/main/java/com/supervision/service/impl/TTsServiceImpl.java

@@ -0,0 +1,38 @@
+package com.supervision.service.impl;
+
+import cn.hutool.http.HttpRequest;
+import cn.hutool.http.HttpUtil;
+import cn.hutool.json.JSONObject;
+import cn.hutool.json.JSONUtil;
+import com.supervision.service.TTsService;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.beans.factory.annotation.Value;
+import org.springframework.stereotype.Service;
+
+@Slf4j
+@Service
+public class TTsServiceImpl implements TTsService {
+
+
+    @Value("${livetalking.tts.url}")
+    private String ttsUrl;
+    @Value("${livetalking.tts.model}")
+    private String model;
+    @Override
+    public String stt(String fileName, byte[] audioData) {
+        String url = ttsUrl + "/v1/audio/transcriptions";
+
+        HttpRequest post = HttpUtil.createPost(url);
+        post.form("model", model);
+        post.form("file", audioData, fileName);
+
+        try (var response = post.execute()) {
+            if (response.isOk()) {
+                String body = response.body();
+                JSONObject entries = JSONUtil.parseObj(body);
+                return entries.getStr("text");
+            }
+            throw new RuntimeException("STT request failed: " + response.getStatus() + " - " + response.body());
+        }
+    }
+}

src/main/resources/application.yml

@@ -43,6 +43,9 @@ livetalking:
   url: http://192.168.10.96:8010
   session:
     id: 0
+  tts:
+    url: http://192.168.10.96:9997
+    model: whisper-large-v3
 jwt:
   secret: "DlHaPUePiN6MyvpMpsMq/t6swzMHqtrRFd2YnofKz4k=" # JWT密钥 使用官方推荐方式生成 Base64.getEncoder().encodeToString(Keys.secretKeyFor(SignatureAlgorithm.HS256).getEncoded());
   expiration: 86400000 # 1小时：3600000 1天：86400000

src/test/java/com/supervision/PlatformApplicationTest.java

@@ -1,13 +1,18 @@
 package com.supervision;
 
+import cn.hutool.core.io.FileUtil;
 import com.hankcs.hanlp.seg.common.Term;
 import com.hankcs.hanlp.tokenizer.StandardTokenizer;
 import com.supervision.domain.SysByteArray;
 import com.supervision.service.SysByteArrayService;
+import com.supervision.service.TTsService;
 import lombok.extern.slf4j.Slf4j;
 import org.junit.jupiter.api.Test;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.boot.test.context.SpringBootTest;
+import java.io.File;
+import java.io.FileInputStream;
+import java.io.IOException;
 
 @Slf4j
 @SpringBootTest
@@ -36,6 +41,17 @@ public class PlatformApplicationTest {
         boolean isQuestion = isQuestion(sentence);
         System.out.println("是否是问句: " + isQuestion);
     }
+
+    @Autowired
+    private TTsService tTsService;
+    @Test
+    public void ttsTest() throws IOException {
+        File file = FileUtil.file("C:\\Users\\Administrator\\Desktop\\doubao-man.wav_0000181120_0000313600.wav");
+        FileInputStream inputStream = new FileInputStream(file);
+        byte[] bytes = inputStream.readAllBytes();
+        String stt = tTsService.stt("doubao-man.wav_0000181120_0000313600.wav", bytes);
+        System.out.println(stt);
+    }
     private boolean isQuestion(String sentence) {
         String[] questionWords = {"什么", "为什么", "如何", "哪", "谁", "多少", "是否", "能否", "是不是"};
         String[] questionEndings = {"吗", "呢", "？", "?"};