TTS 返回语音优化

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

@@ -145,7 +145,9 @@ public class ChatWebSocketHandler {
                             //持续返回数据流给客户端
                             try {
                                 String resultOutPathUrl = RuoYiConfig.getProfile() + VOICE_STORAGE_RESULT_DIR + "123_"+resultFileName;
-                                removeSilence(resultPathUrl, resultOutPathUrl);
+                                File inputFile = new File(resultPathUrl);
+                                File outputFile = new File(resultOutPathUrl);
+                                trimSilence(inputFile, outputFile);
                                 //文件转换成文件流
                                 ByteBuffer outByteBuffer = convertFileToByteBuffer(resultOutPathUrl);
                                 //发送文件流数据
@@ -406,26 +408,60 @@ public class ChatWebSocketHandler {
         }
     }
 
-    public void removeSilence(String inputFile, String outputFile) {
-        try {
-            // FFmpeg 命令，去除音频后面的静音部分
-            String command = "ffmpeg -i " + inputFile +
-                    " -af silenceremove=start_periods=1:start_duration=0.2:start_threshold=-40dB " +
-                    "-af silenceremove=end_periods=1:end_duration=0.2:end_threshold=-40dB " +
-                    outputFile;
 
-            // 执行命令
-            Process process = Runtime.getRuntime().exec(command);
-            int exitCode = process.waitFor();
-            if (exitCode == 0) {
-                System.out.println("静音部分已去除");
-            } else {
-                System.out.println("音频处理失败");
-            }
-        } catch (IOException | InterruptedException e) {
+    // 截取音频流，去除静音部分
+    public void trimSilence(File inputFile, File outputFile) {
+        try {
+            // 获取音频流
+            AudioInputStream audioInputStream = AudioSystem.getAudioInputStream(inputFile);
+            AudioFormat format = audioInputStream.getFormat();
+
+            // 将音频流转为字节数组
+            byte[] audioBytes = audioInputStream.readAllBytes();
+
+            // 去除前后静音部分
+            int start = findNonSilenceStart(audioBytes, format);
+            int end = findNonSilenceEnd(audioBytes, format);
+
+            // 截取音频流
+            byte[] trimmedAudio = new byte[end - start];
+            System.arraycopy(audioBytes, start, trimmedAudio, 0, trimmedAudio.length);
+
+            // 创建新的音频输入流
+            ByteArrayInputStream byteArrayInputStream = new ByteArrayInputStream(trimmedAudio);
+            AudioInputStream trimmedAudioStream = new AudioInputStream(byteArrayInputStream, format, trimmedAudio.length / format.getFrameSize());
+
+            // 保存裁剪后的音频文件
+            AudioSystem.write(trimmedAudioStream, AudioFileFormat.Type.WAVE, outputFile);
+            System.out.println("裁剪后的音频已保存到: " + outputFile.getAbsolutePath());
+
+            audioInputStream.close();
+        } catch (Exception e) {
             e.printStackTrace();
         }
     }
 
+    // 找到音频流中非静音部分的开始位置
+    private int findNonSilenceStart(byte[] audioBytes, AudioFormat format) {
+        int threshold = 10; // 假设小于这个值的幅度为静音
+        for (int i = 0; i < audioBytes.length; i++) {
+            if (Math.abs(audioBytes[i]) > threshold) {
+                return i;
+            }
+        }
+        return 0; // 如果音频全是静音，返回 0
+    }
+
+    // 找到音频流中非静音部分的结束位置
+    private int findNonSilenceEnd(byte[] audioBytes, AudioFormat format) {
+        int threshold = 10; // 假设小于这个值的幅度为静音
+        for (int i = audioBytes.length - 1; i >= 0; i--) {
+            if (Math.abs(audioBytes[i]) > threshold) {
+                return i + 1;
+            }
+        }
+        return audioBytes.length; // 如果音频全是静音，返回音频的最后位置
+    }
+
 }