refactor(tts): add default implements

Author: ed

src/main/java/com/xiaozhi/dialogue/tts/TtsService.java

@@ -1,5 +1,6 @@
 package com.xiaozhi.dialogue.tts;
 
+import java.util.UUID;
 import java.util.function.Consumer;
 
 /**
@@ -12,17 +13,28 @@ public interface TtsService {
    */
   String getProviderName();
 
+  /**
+   * 音频格式
+   */
+  default String audioFormat() {
+    return "wav";
+  }
+
   /**
    * 生成文件名称
    * 
    * @return 文件名称
    */
-  String getAudioFileName();
+  default String getAudioFileName() {
+    return UUID.randomUUID().toString().replace("-", "") + "." + audioFormat();
+  }
 
   /**
    * 
    */
-  boolean isSupportStreamTts();
+  default boolean isSupportStreamTts() {
+    return false;
+  }
 
   /**
    * 将文本转换为语音（带自定义语音）
@@ -39,6 +51,8 @@ public interface TtsService {
    * @param audioDataConsumer 音频数据消费者，接收PCM格式的音频数据块
    * @throws Exception 转换过程中可能发生的异常
    */
-  void streamTextToSpeech(String text, Consumer<byte[]> audioDataConsumer) throws Exception;
+  default void streamTextToSpeech(String text, Consumer<byte[]> audioDataConsumer) throws Exception {
+    throw new UnsupportedOperationException("Unimplemented method 'streamTextToSpeech'");
+  }
 
 }

src/main/java/com/xiaozhi/dialogue/tts/factory/TtsServiceFactory.java

@@ -10,6 +10,7 @@
 import org.springframework.util.ObjectUtils;
 import org.springframework.util.StringUtils;
 
+import java.io.File;
 import java.util.Map;
 import java.util.concurrent.ConcurrentHashMap;
 
@@ -43,8 +44,7 @@ public TtsService getTtsService() {
     }
 
     private TtsService getTtsService(String voiceName) {
-        TtsService ttsService = new EdgeTtsService(voiceName, OUT_PUT_PATH);
-        return ttsService;
+        return getTtsService(null, voiceName);
     }
 
     private String createCacheKey(SysConfig config,String provider){
@@ -56,8 +56,7 @@ private String createCacheKey(SysConfig config,String provider){
             Integer configId = config.getConfigId();
             configIdStr = configId != null ? String.valueOf(configId) : "default";
         }
-        String cacheKey = provider + ":" + configIdStr;
-        return cacheKey;
+        return provider + ":" + configIdStr;
     }
 
     /**
@@ -107,8 +106,10 @@ public TtsService getTtsService(SysConfig config, String voiceName) {
      * 根据配置创建API类型的TTS服务
      */
     private TtsService createApiService(SysConfig config, String voiceName, String outputPath) {
-        String provider = config.getProvider();
+        // Make sure output dir exists
+        ensureOutputPath(outputPath);
 
+        var provider = config.getProvider();
         // 如果是Edge，直接返回Edge服务
         if (DEFAULT_PROVIDER.equals(provider)) {
             return new EdgeTtsService(voiceName, outputPath);
@@ -118,11 +119,9 @@ private TtsService createApiService(SysConfig config, String voiceName, String o
             return new VolcengineTtsService(config, voiceName, outputPath);
         } else if ("xfyun".equals(provider)) {
             return new XfyunTtsService(config, voiceName, outputPath);
-        }/*
-           * else if ("tencent".equals(provider)) {
-           * return new TencentTtsService(config, voiceName, outputPath);
-           * }
-           */
+        } else if ("minimax".equals(provider)) {
+            return new MiniMaxTtsService(config, voiceName, outputPath);
+        }
 
         logger.warn("不支持的TTS服务提供商: {}", provider);
         return null;
@@ -136,4 +135,8 @@ public void removeCache(SysConfig config) {
         serviceCache.remove(cacheKey);
     }
 
+    private void ensureOutputPath(String outputPath) {
+        File dir = new File(outputPath);
+        if (!dir.exists()) dir.mkdirs();
+    }
 }

src/main/java/com/xiaozhi/dialogue/tts/providers/AliyunTtsService.java

@@ -22,13 +22,11 @@
 import java.io.InputStream;
 import java.net.URL;
 import java.nio.ByteBuffer;
-import java.util.UUID;
 import java.util.concurrent.ExecutorService;
 import java.util.concurrent.Executors;
 import java.util.concurrent.Future;
 import java.util.concurrent.TimeUnit;
 import java.util.concurrent.TimeoutException;
-import java.util.function.Consumer;
 
 public class AliyunTtsService implements TtsService {
     private static final Logger logger = LoggerFactory.getLogger(AliyunTtsService.class);
@@ -58,24 +56,9 @@ public String getProviderName() {
         return PROVIDER_NAME;
     }
 
-    @Override
-    public boolean isSupportStreamTts() {
-        return false;
-    }
-
-    @Override
-    public String getAudioFileName() {
-        String uuid = UUID.randomUUID().toString().replace("-", "");
-        return uuid + ".wav";
-    }
-
     @Override
     public String textToSpeech(String text) throws Exception {
         try {
-            File outputDir = new File(outputPath);
-            if (!outputDir.exists()) {
-                outputDir.mkdirs();
-            }
             if (voiceName.contains("sambert")) {
                 return ttsSambert(text);
             } else if (getVoiceByName(voiceName) != null) {
@@ -394,9 +377,4 @@ public String ttsSambert(String text) {
         return StrUtil.EMPTY;
     }
 
-    @Override
-    public void streamTextToSpeech(String text, Consumer<byte[]> audioDataConsumer) throws Exception {
-        // TODO Auto-generated method stub
-        throw new UnsupportedOperationException("Unimplemented method 'streamTextToSpeech'");
-    }
 }

src/main/java/com/xiaozhi/dialogue/tts/providers/EdgeTtsService.java

@@ -6,8 +6,6 @@
 
 import java.nio.file.Files;
 import java.nio.file.Paths;
-import java.util.UUID;
-import java.util.function.Consumer;
 import java.util.stream.Collectors;
 
 import org.slf4j.Logger;
@@ -38,14 +36,8 @@ public String getProviderName() {
     }
 
     @Override
-    public boolean isSupportStreamTts() {
-        return false;
-    }
-
-    @Override
-    public String getAudioFileName() {
-        String uuid = UUID.randomUUID().toString().replace("-", "");
-        return uuid + ".mp3";
+    public String audioFormat() {
+        return "mp3";
     }
 
     @Override
@@ -80,9 +72,4 @@ public String textToSpeech(String text) throws Exception {
         return AudioUtils.AUDIO_PATH + resampledFileName;
     }
 
-    @Override
-    public void streamTextToSpeech(String text, Consumer<byte[]> audioDataConsumer) throws Exception {
-        // TODO Auto-generated method stub
-        throw new UnsupportedOperationException("Unimplemented method 'streamTextToSpeech'");
-    }
 }

src/main/java/com/xiaozhi/dialogue/tts/providers/VolcengineTtsService.java

@@ -15,7 +15,6 @@
 import java.io.FileOutputStream;
 import java.util.Base64;
 import java.util.UUID;
-import java.util.function.Consumer;
 
 public class VolcengineTtsService implements TtsService {
     private static final Logger logger = LoggerFactory.getLogger(VolcengineTtsService.class);
@@ -34,7 +33,7 @@ public class VolcengineTtsService implements TtsService {
     private String appId;
     private String accessToken; // 对应 apiKey
 
-    private final OkHttpClient client = HttpUtil.client;;
+    private final OkHttpClient client = HttpUtil.client;
 
     public VolcengineTtsService(SysConfig config, String voiceName, String outputPath) {
         this.voiceName = voiceName;
@@ -48,17 +47,6 @@ public String getProviderName() {
         return PROVIDER_NAME;
     }
 
-    @Override
-    public boolean isSupportStreamTts() {
-        return false;
-    }
-
-    @Override
-    public String getAudioFileName() {
-        String uuid = UUID.randomUUID().toString().replace("-", "");
-        return uuid + ".wav";
-    }
-
     @Override
     public String textToSpeech(String text) throws Exception {
         if (text == null || text.isEmpty()) {
@@ -163,12 +151,6 @@ private boolean sendRequest(String text, String audioFilePath) throws Exception
                         String base64Audio = jsonResponse.get("data").getAsString();
                         byte[] audioData = Base64.getDecoder().decode(base64Audio);
 
-                        // 确保目录存在
-                        File audioFileDir = new File(outputPath);
-                        if (!audioFileDir.exists()) {
-                            audioFileDir.mkdirs();
-                        }
-
                         // 保存音频文件
                         File audioFile = new File(audioFilePath);
                         try (FileOutputStream fout = new FileOutputStream(audioFile)) {
@@ -191,10 +173,4 @@ private boolean sendRequest(String text, String audioFilePath) throws Exception
         }
     }
 
-    @Override
-    public void streamTextToSpeech(String text, Consumer<byte[]> audioDataConsumer) throws Exception {
-        // TODO Auto-generated method stub
-        throw new UnsupportedOperationException("Unimplemented method 'streamTextToSpeech'");
-    }
-
 }

src/main/java/com/xiaozhi/dialogue/tts/providers/XfyunTtsService.java

@@ -13,10 +13,8 @@
 import java.io.File;
 import java.io.FileOutputStream;
 import java.io.IOException;
-import java.util.UUID;
 import java.util.concurrent.CountDownLatch;
 import java.util.concurrent.TimeUnit;
-import java.util.function.Consumer;
 
 public class XfyunTtsService implements TtsService {
     private static final Logger logger = LoggerFactory.getLogger(XfyunTtsService.class);
@@ -50,14 +48,8 @@ public String getProviderName() {
     }
 
     @Override
-    public boolean isSupportStreamTts() {
-        return false;
-    }
-
-    @Override
-    public String getAudioFileName() {
-        String uuid = UUID.randomUUID().toString().replace("-", "");
-        return uuid + ".mp3";
+    public String audioFormat() {
+        return "mp3";
     }
 
     @Override
@@ -139,10 +131,4 @@ public void onBusinessFail(WebSocket webSocket, TtsResponse ttsResponse) {
         return true;
     }
 
-    @Override
-    public void streamTextToSpeech(String text, Consumer<byte[]> audioDataConsumer) throws Exception {
-        // TODO Auto-generated method stub
-        throw new UnsupportedOperationException("Unimplemented method 'streamTextToSpeech'");
-    }
-
 }