STT 数据流处理

.idea/misc.xml

@@ -13,5 +13,9 @@
       </set>
     </option>
   </component>
+  <component name="PWA">
+    <option name="enabled" value="true" />
+    <option name="wasEnabledAtLeastOnce" value="true" />
+  </component>
   <component name="ProjectRootManager" version="2" languageLevel="JDK_21" project-jdk-name="21" project-jdk-type="JavaSDK" />
 </project>

vetti-admin/src/main/java/com/vetti/socket/ChatWebSocketHandler.java

@@ -6,6 +6,7 @@ import cn.hutool.json.JSONUtil;
 import com.vetti.common.ai.elevenLabs.ElevenLabsClient;
 import com.vetti.common.ai.gpt.OpenAiStreamClient;
 import com.vetti.common.ai.gpt.service.OpenAiStreamListenerService;
+import com.vetti.common.ai.whisper.WhisperClient;
 import com.vetti.common.config.RuoYiConfig;
 import com.vetti.common.utils.spring.SpringUtils;
 import lombok.extern.slf4j.Slf4j;
@@ -13,17 +14,16 @@ import okhttp3.*;
 import org.apache.commons.io.FileUtils;
 import org.springframework.stereotype.Component;
 
-import javax.sound.sampled.AudioFormat;
 import javax.sound.sampled.AudioInputStream;
 import javax.sound.sampled.AudioSystem;
-import javax.sound.sampled.UnsupportedAudioFileException;
 import javax.websocket.*;
 import javax.websocket.server.PathParam;
 import javax.websocket.server.ServerEndpoint;
 import java.io.*;
 import java.nio.ByteBuffer;
-import java.util.Base64;
+import java.util.ArrayList;
 import java.util.HashMap;
+import java.util.List;
 import java.util.Map;
 import java.util.concurrent.ConcurrentHashMap;
 
@@ -70,6 +70,12 @@ public class ChatWebSocketHandler {
      */
     private final Map<String, WebSocket> cacheWebSocket = new ConcurrentHashMap<>();
 
+    /**
+     * 为每个会话维护分片缓存（线程安全，支持多用户）
+     */
+    private final ConcurrentHashMap<String, List<byte[]>> fragmentCache = new ConcurrentHashMap<>();
+
+
     // 语音文件保存目录
     private static final String VOICE_STORAGE_DIR = "/voice_files/";
 
@@ -97,6 +103,7 @@ public class ChatWebSocketHandler {
         cacheClientTts.put(clientId,new String());
         //初始化STT流式语音转换文本的socket链接
         createWhisperRealtimeSocket(clientId);
+
     }
 
     // 接收文本消息
@@ -109,16 +116,30 @@ public class ChatWebSocketHandler {
                 Map<String,String> mapResult = JSONUtil.toBean(JSONUtil.parseObj(message),Map.class);
                 String resultFlag = mapResult.get("msg");
                 if("done".equals(resultFlag)){
+                    log.info("1、开始处理时间:{}",System.currentTimeMillis()/1000);
+                    //开始合并语音流
+                    List<byte[]> fragments = fragmentCache.get(clientId);
+                    // 合并所有分片为完整语音数据
+                    byte[] fullVoiceData = mergeFragments(fragments);
+                            // 生成唯一文件名
+                    String fileName = clientId + "_" + System.currentTimeMillis() + ".webm";
+                    String pathUrl = RuoYiConfig.getProfile()+VOICE_STORAGE_DIR + fileName;
+                    log.info("文件路径为:{}", pathUrl);
+                    log.info("文件流的大小为:{}",fullVoiceData.length);
+                    saveAsWebM(fullVoiceData,pathUrl);
+                    //开始转换
+                    WhisperClient whisperClient = SpringUtils.getBean(WhisperClient.class);
+                    String cacheResultText = whisperClient.handleVoiceToText(pathUrl);
 
                     //发送消息
-                    WebSocket webSocket = cacheWebSocket.get(clientId);
+//                    WebSocket webSocket = cacheWebSocket.get(clientId);
 //                    webSocket.send("{\"type\": \"input_audio_buffer.commit\"}");
 //                    webSocket.send("{\"type\": \"response.create\"}");
 //                    if(webSocket != null){
 //                        webSocket.close(1000,null);
 //                    }
                     //语音结束,开始进行回答解析
-                    String cacheResultText = cacheClientTts.get(clientId);
+//                    String cacheResultText = cacheClientTts.get(clientId);
                     log.info("返回的结果为:{}",cacheResultText);
                     if(StrUtil.isEmpty(cacheResultText)){
                         cacheResultText = "Hello , How are you?";
@@ -179,6 +200,51 @@ public class ChatWebSocketHandler {
         }
     }
 
+//    // 接收二进制消息（流数据）
+//    @OnMessage
+//    public void onBinaryMessage(Session session, @PathParam("clientId") String clientId, ByteBuffer byteBuffer) {
+//        log.info("1、开始接收数据流时间:{}",System.currentTimeMillis()/1000);
+//        log.info("客户端ID为:{}", clientId);
+//        // 处理二进制流数据
+//        byte[] bytes = new byte[byteBuffer.remaining()];
+//        //从缓冲区中读取数据并存储到指定的字节数组中
+//        byteBuffer.get(bytes);
+//        log.info("2、开始接收数据流时间:{}",System.currentTimeMillis()/1000);
+//        // 生成唯一文件名
+//        String fileName = clientId + "_" + System.currentTimeMillis() + ".wav";
+//        String pathUrl = RuoYiConfig.getProfile()+VOICE_STORAGE_DIR + fileName;
+//        log.info("文件路径为:{}", pathUrl);
+//        log.info("3、开始接收数据流时间:{}",System.currentTimeMillis()/1000);
+//        try{
+//            log.info("文件流的大小为:{}",bytes.length);
+//            saveAsWebM(bytes,pathUrl);
+//            //接收到数据流后直接就进行SST处理
+//            //发送消息
+//            WebSocket webSocket = cacheWebSocket.get(clientId);
+//            log.info("获取的socket对象为:{}",webSocket);
+//            if(webSocket != null){
+////                 1. 启动音频缓冲
+////                webSocket.send("{\"type\": \"input_audio_buffer.start\"}");
+//                log.info("3.1 开始发送数据音频流啦");
+//                // 将音频数据转换为 Base64 编码的字符串
+//                //进行转换
+//                // 转换音频格式
+//                AudioFormat format = new AudioFormat(SAMPLE_RATE, BITS_PER_SAMPLE, 1, true, false);
+//                byte[] outputAudioBytes = convertAudio(bytes, format);
+//                String base64Audio = Base64.getEncoder().encodeToString(outputAudioBytes);
+//                String message = "{ \"type\": \"input_audio_buffer.append\", \"audio\": \"" + base64Audio + "\" }";
+//                webSocket.send(message);
+//                log.info("4、开始接收数据流时间:{}",System.currentTimeMillis()/1000);
+//                // 3. 提交音频并请求转录
+////                webSocket.send("{\"type\": \"input_audio_buffer.commit\"}");
+////                webSocket.send("{\"type\": \"response.create\"}");
+//            }
+//        }catch (Exception e){
+//            e.printStackTrace();
+//        }
+//
+//    }
+
     // 接收二进制消息（流数据）
     @OnMessage
     public void onBinaryMessage(Session session, @PathParam("clientId") String clientId, ByteBuffer byteBuffer) {
@@ -188,40 +254,15 @@ public class ChatWebSocketHandler {
         byte[] bytes = new byte[byteBuffer.remaining()];
         //从缓冲区中读取数据并存储到指定的字节数组中
         byteBuffer.get(bytes);
-        log.info("2、开始接收数据流时间:{}",System.currentTimeMillis()/1000);
-        // 生成唯一文件名
-        String fileName = clientId + "_" + System.currentTimeMillis() + ".wav";
-        String pathUrl = RuoYiConfig.getProfile()+VOICE_STORAGE_DIR + fileName;
-        log.info("文件路径为:{}", pathUrl);
-        log.info("3、开始接收数据流时间:{}",System.currentTimeMillis()/1000);
-        try{
-            log.info("文件流的大小为:{}",bytes.length);
-            saveAsWebM(bytes,pathUrl);
-            //接收到数据流后直接就进行SST处理
-            //发送消息
-            WebSocket webSocket = cacheWebSocket.get(clientId);
-            log.info("获取的socket对象为:{}",webSocket);
-            if(webSocket != null){
-//                 1. 启动音频缓冲
-//                webSocket.send("{\"type\": \"input_audio_buffer.start\"}");
-                log.info("3.1 开始发送数据音频流啦");
-                // 将音频数据转换为 Base64 编码的字符串
-                //进行转换
-                // 转换音频格式
-                AudioFormat format = new AudioFormat(SAMPLE_RATE, BITS_PER_SAMPLE, 1, true, false);
-                byte[] outputAudioBytes = convertAudio(bytes, format);
-                String base64Audio = Base64.getEncoder().encodeToString(outputAudioBytes);
-                String message = "{ \"type\": \"input_audio_buffer.append\", \"audio\": \"" + base64Audio + "\" }";
-                webSocket.send(message);
-                log.info("4、开始接收数据流时间:{}",System.currentTimeMillis()/1000);
-                // 3. 提交音频并请求转录
-//                webSocket.send("{\"type\": \"input_audio_buffer.commit\"}");
-//                webSocket.send("{\"type\": \"response.create\"}");
-            }
-        }catch (Exception e){
-            e.printStackTrace();
-        }
 
+        // 1. 获取当前会话的缓存
+        List<byte[]> fragments = fragmentCache.get(clientId);
+        if (fragments == null) {
+            fragments = new ArrayList<>();
+            fragmentCache.put(clientId, fragments);
+        }
+        fragments.add(bytes);
+        fragmentCache.put(clientId, fragments);
     }
 
     // 连接关闭时调用
@@ -237,27 +278,27 @@ public class ChatWebSocketHandler {
         throwable.printStackTrace();
     }
 
-    public static byte[] convertAudio(byte[] inputAudioBytes, AudioFormat targetFormat) throws Exception {
-        // 将 byte[] 转换为 AudioInputStream
-        ByteArrayInputStream byteArrayInputStream = new ByteArrayInputStream(inputAudioBytes);
-        AudioInputStream inputAudioStream = new AudioInputStream(byteArrayInputStream, targetFormat, inputAudioBytes.length);
-
-        // 创建目标格式的 AudioInputStream
-        AudioInputStream outputAudioStream = AudioSystem.getAudioInputStream(targetFormat, inputAudioStream);
-
-        // 获取输出音频的 byte[]
-        ByteArrayOutputStream byteArrayOutputStream = new ByteArrayOutputStream();
-        byte[] buffer = new byte[1024];
-        int bytesRead;
-
-        // 从 AudioInputStream 读取数据并写入 ByteArrayOutputStream
-        while ((bytesRead = outputAudioStream.read(buffer)) != -1) {
-            byteArrayOutputStream.write(buffer, 0, bytesRead);
-        }
-
-        // 返回转换后的 byte[]
-        return byteArrayOutputStream.toByteArray();
-    }
+//    public static byte[] convertAudio(byte[] inputAudioBytes, AudioFormat targetFormat) throws Exception {
+//        // 将 byte[] 转换为 AudioInputStream
+//        ByteArrayInputStream byteArrayInputStream = new ByteArrayInputStream(inputAudioBytes);
+//        AudioInputStream inputAudioStream = new AudioInputStream(byteArrayInputStream, targetFormat, inputAudioBytes.length);
+//
+//        // 创建目标格式的 AudioInputStream
+//        AudioInputStream outputAudioStream = AudioSystem.getAudioInputStream(targetFormat, inputAudioStream);
+//
+//        // 获取输出音频的 byte[]
+//        ByteArrayOutputStream byteArrayOutputStream = new ByteArrayOutputStream();
+//        byte[] buffer = new byte[1024];
+//        int bytesRead;
+//
+//        // 从 AudioInputStream 读取数据并写入 ByteArrayOutputStream
+//        while ((bytesRead = outputAudioStream.read(buffer)) != -1) {
+//            byteArrayOutputStream.write(buffer, 0, bytesRead);
+//        }
+//
+//        // 返回转换后的 byte[]
+//        return byteArrayOutputStream.toByteArray();
+//    }
 
     /**
      * 将字节数组保存为WebM文件
@@ -405,56 +446,26 @@ public class ChatWebSocketHandler {
         }
     }
 
-    private void handleVoice(String inputPath,String outputPath){
-        double trimMs = 270; // 要去掉的尾部时长（毫秒）
-        try {
-            // 1. 解析音频格式和总长度
-            AudioInputStream audioIn = AudioSystem.getAudioInputStream(new File(inputPath));
-            AudioFormat format = audioIn.getFormat();
-            long totalBytes = audioIn.getFrameLength() * format.getFrameSize(); // 总字节数
 
-            // 2. 计算300毫秒对应的字节数
-            float sampleRate = format.getSampleRate(); // 采样率（Hz）
-            int frameSize = format.getFrameSize(); // 每帧字节数（位深/8 * 声道数）
-            double trimSeconds = trimMs / 1000.0; // 转换为秒
-            long trimBytes = (long) (sampleRate * trimSeconds * frameSize); // 要去掉的字节数
-
-            // 3. 计算需要保留的字节数（避免负数）
-            long keepBytes = Math.max(0, totalBytes - trimBytes);
-            if (keepBytes == 0) {
-                System.out.println("音频长度小于300毫秒，无法截断");
-                return;
+    /**
+     * 合并分片数组为完整字节数组
+     */
+    private byte[] mergeFragments(List<byte[]> fragments) {
+        // 计算总长度
+        int totalLength = 0;
+        for (byte[] fragment : fragments) {
+            totalLength += fragment.length;
         }
-            File file = new File(outputPath);
-            // 创建空文件
-            boolean isCreated = file.createNewFile();
-            if (isCreated) {
-                System.out.println("空文件创建成功：" + file.getAbsolutePath());
-            } else {
-                System.out.println("文件已存在：" + file.getAbsolutePath());
+        // 拼接所有分片
+        byte[] result = new byte[totalLength];
+        int offset = 0;
+        for (byte[] fragment : fragments) {
+            System.arraycopy(fragment, 0, result, offset, fragment.length);
+            offset += fragment.length;
         }
-            // 4. 读取并保留前半部分（去掉最后300毫秒）
-            try (InputStream in = new FileInputStream(inputPath);
-                 OutputStream out = new FileOutputStream(outputPath)) {
-
-                byte[] buffer = new byte[4096];
-                long totalRead = 0;
-                int bytesRead;
-
-                while (totalRead < keepBytes && (bytesRead = in.read(buffer)) != -1) {
-                    long remaining = keepBytes - totalRead;
-                    int writeBytes = (remaining < bytesRead) ? (int) remaining : bytesRead;
-                    out.write(buffer, 0, writeBytes);
-                    totalRead += writeBytes;
+        return result;
     }
 
-                System.out.println("处理完成，去掉了最后" + trimMs + "毫秒，保留了" + totalRead + "字节");
-            }
-
-        } catch (UnsupportedAudioFileException | IOException e) {
-            e.printStackTrace();
-        }
-    }
 
 }