STT 数据流处理

vetti-common/src/main/java/com/vetti/common/ai/whisper/AudioToPCM.java

@@ -0,0 +1,54 @@
+package com.vetti.common.ai.whisper;
+
+import javax.sound.sampled.*;
+import java.io.*;
+
+public class AudioToPCM {
+    public static void main(String[] args) {
+        File inputFile = new File("/Users/wangxiangshun/Desktop/临时文件/110/buffer.wav");  // 输入音频文件
+        File outputFile = new File("/Users/wangxiangshun/Desktop/临时文件/110/output_pcm_audio.pcm");  // 输出PCM格式文件
+
+        try {
+            // 读取音频文件
+            AudioInputStream inputAudioStream = AudioSystem.getAudioInputStream(inputFile);
+
+            // 获取音频文件的格式信息
+            AudioFormat sourceFormat = inputAudioStream.getFormat();
+            System.out.println("Input Audio Format: " + sourceFormat);
+
+            // 设置目标PCM格式 (可以是16-bit, 8kHz, Mono, Linear PCM)
+            AudioFormat pcmFormat = new AudioFormat(
+                    AudioFormat.Encoding.PCM_SIGNED,
+                    sourceFormat.getSampleRate(),
+                    16,  // 16-bit samples
+                    1,    // 单声道
+                    2,    // 每个样本2字节（16位）
+                    sourceFormat.getSampleRate(),
+                    false // 大端模式
+            );
+
+            // 获取PCM格式的音频流
+            AudioInputStream pcmAudioStream = AudioSystem.getAudioInputStream(pcmFormat, inputAudioStream);
+
+            // 创建输出文件流
+            FileOutputStream fos = new FileOutputStream(outputFile);
+            byte[] buffer = new byte[1024];
+            int bytesRead;
+
+            // 将PCM音频数据写入输出文件
+            while ((bytesRead = pcmAudioStream.read(buffer)) != -1) {
+                fos.write(buffer, 0, bytesRead);
+            }
+
+            // 关闭流
+            pcmAudioStream.close();
+            fos.close();
+
+            System.out.println("Audio has been converted to PCM format and saved at: " + outputFile.getAbsolutePath());
+
+        } catch (Exception e) {
+            e.printStackTrace();
+        }
+    }
+}
+

vetti-common/src/main/java/com/vetti/common/ai/whisper/RealtimeTranscriptionMicrophone.java

@@ -2,11 +2,14 @@ package com.vetti.common.ai.whisper;
 
 import cn.hutool.json.JSONObject;
 import okhttp3.*;
+import org.apache.commons.io.FileUtils;
 
 import javax.sound.sampled.AudioFormat;
 import javax.sound.sampled.AudioSystem;
 import javax.sound.sampled.DataLine;
 import javax.sound.sampled.TargetDataLine;
+import java.io.File;
+import java.nio.ByteBuffer;
 import java.util.Base64;
 import java.util.concurrent.CountDownLatch;
 
@@ -63,23 +66,30 @@ public class RealtimeTranscriptionMicrophone {
                 new Thread(() -> {
                     try {
                         // 设置麦克风输入流
-                        AudioFormat format = new AudioFormat(SAMPLE_RATE, BITS_PER_SAMPLE, 1, true, false);
-                        DataLine.Info info = new DataLine.Info(TargetDataLine.class, format);
-                        TargetDataLine line = (TargetDataLine) AudioSystem.getLine(info);
-                        line.open(format);
-                        line.start();
-
-                        byte[] buffer = new byte[BUFFER_SIZE];
-                        int bytesRead;
-                        while ((bytesRead = line.read(buffer, 0, buffer.length)) > 0) {
-                            // 将音频数据转换为 Base64 编码的字符串
-                            byte[] audioData = new byte[bytesRead];
-                            System.arraycopy(buffer, 0, audioData, 0, bytesRead);
-                            String base64Audio = Base64.getEncoder().encodeToString(audioData);
-                            String message = "{ \"type\": \"input_audio_buffer.append\", \"audio\": \"" + base64Audio + "\" }";
-                            webSocket.send(message);
-                        }
-
+//                        AudioFormat format = new AudioFormat(SAMPLE_RATE, BITS_PER_SAMPLE, 1, true, false);
+//                        DataLine.Info info = new DataLine.Info(TargetDataLine.class, format);
+//                        TargetDataLine line = (TargetDataLine) AudioSystem.getLine(info);
+//                        line.open(format);
+//                        line.start();
+//
+//                        byte[] buffer = new byte[BUFFER_SIZE];
+//                        int bytesRead;
+//                        while ((bytesRead = line.read(buffer, 0, buffer.length)) > 0) {
+//                            // 将音频数据转换为 Base64 编码的字符串
+//                            byte[] audioData = new byte[bytesRead];
+//                            System.arraycopy(buffer, 0, audioData, 0, bytesRead);
+//                            String base64Audio = Base64.getEncoder().encodeToString(audioData);
+//                            String message = "{ \"type\": \"input_audio_buffer.append\", \"audio\": \"" + base64Audio + "\" }";
+//                            webSocket.send(message);
+//                        }
+                        File outputFile = new File("/Users/wangxiangshun/Desktop/临时文件/110/output_pcm_audio.pcm");  // 输出PCM格式文件
+                        ByteBuffer byteBuffer = ByteBuffer.wrap(FileUtils.readFileToByteArray(outputFile));
+                        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);
+                        String base64Audio = Base64.getEncoder().encodeToString(bytes);
+                        String message = "{ \"type\": \"input_audio_buffer.append\", \"audio\": \"" + base64Audio + "\" }";
+                        webSocket.send(message);
                         // 3. 提交音频并请求转录
 //                        webSocket.send("{\"type\": \"input_audio_buffer.commit\"}");
 //                        webSocket.send("{\"type\": \"response.create\"}");