Merge pull request #96 from vritser/main

feat: support minimax t2a

Author: joey-zhou

src/main/java/com/xiaozhi/communication/common/MessageHandler.java

@@ -210,7 +210,7 @@ public void handleUnboundDevice(String sessionId, SysDevice device) {
                 if (device.getDeviceName() != null && device.getModelId() == null) {
                     String message = "设备未配置对话模型，请到配置页面完成配置后开始对话";
 
-                    String audioFilePath = ttsService.getTtsService().textToSpeech(message);
+                    String audioFilePath = ttsService.getDefaultTtsService().textToSpeech(message);
                     audioService.sendAudioMessage(chatSession, new DialogueService.Sentence(message, audioFilePath), true,
                             true);
 
@@ -230,7 +230,7 @@ public void handleUnboundDevice(String sessionId, SysDevice device) {
                 String audioFilePath;
                 if (!StringUtils.hasText(codeResult.getAudioPath())) {
                     String codeMessage = "请到设备管理页面添加设备，输入验证码" + codeResult.getCode();
-                    audioFilePath = ttsService.getTtsService().textToSpeech(codeMessage);
+                    audioFilePath = ttsService.getDefaultTtsService().textToSpeech(codeMessage);
                     codeResult.setDeviceId(deviceId);
                     codeResult.setSessionId(sessionId);
                     codeResult.setAudioPath(audioFilePath);

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

@@ -8,8 +8,8 @@
 import org.slf4j.LoggerFactory;
 import org.springframework.stereotype.Component;
 import org.springframework.util.ObjectUtils;
-import org.springframework.util.StringUtils;
 
+import java.io.File;
 import java.util.Map;
 import java.util.concurrent.ConcurrentHashMap;
 
@@ -34,98 +34,49 @@ public class TtsServiceFactory {
      * 获取默认TTS服务
      */
     public TtsService getDefaultTtsService() {
-        // 如果缓存中没有默认服务，则创建一个
-        return getTtsService(DEFAULT_VOICE);
+        var config = new SysConfig().setProvider(DEFAULT_PROVIDER);
+        return getTtsService(config, TtsServiceFactory.DEFAULT_VOICE);
     }
 
-    public TtsService getTtsService() {
-        return getTtsService(DEFAULT_VOICE);
-    }
-
-    private TtsService getTtsService(String voiceName) {
-        TtsService ttsService = new EdgeTtsService(voiceName, OUT_PUT_PATH);
-        return ttsService;
-    }
-
-    private String createCacheKey(SysConfig config,String provider){
-        // 对于API服务，使用"provider:configId"作为缓存键，确保每个配置使用独立的服务实例
-        String configIdStr;
-        if (config == null) {
-            configIdStr = "default";
-        }else{
-            Integer configId = config.getConfigId();
-            configIdStr = configId != null ? String.valueOf(configId) : "default";
-        }
-        String cacheKey = provider + ":" + configIdStr;
-        return cacheKey;
+    // 对于API服务，使用"provider:configId"作为缓存键，确保每个配置使用独立的服务实例
+    private String createCacheKey(SysConfig config, String provider) {
+        var configId = config.getConfigId();
+        var configIdStr = configId != null ? String.valueOf(configId) : "default";
+        return provider + ":" + configIdStr;
     }
 
     /**
      * 根据配置获取TTS服务
      */
     public TtsService getTtsService(SysConfig config, String voiceName) {
-
-        String provider;
         // 如果提供商为空，则使用默认提供商
-        if (ObjectUtils.isEmpty(config)) {
-            provider = DEFAULT_PROVIDER;
-        } else {
-            provider = config.getProvider();
-        }
-        String cacheKey = createCacheKey(config,provider);
+        var provider = ObjectUtils.isEmpty(config) ? DEFAULT_PROVIDER : config.getProvider();
+        var cacheKey = createCacheKey(config, provider);
+
         // 检查是否已有该配置的服务实例
         if (serviceCache.containsKey(cacheKey)) {
             return serviceCache.get(cacheKey);
-        }else{
-            // 如果是默认提供商且尚未初始化，则初始化
-            if (DEFAULT_PROVIDER.equals(provider)) {
-                if(StringUtils.hasText(voiceName)){
-                    TtsService ttsService = getTtsService(voiceName);
-                    serviceCache.put(cacheKey, ttsService);
-                    return ttsService;
-                }else{
-                    TtsService ttsService = getTtsService();
-                    serviceCache.put(cacheKey, ttsService);
-                    return ttsService;
-                }
-            }
-            // 创建新的服务实例
-            try {
-                TtsService service;
-                // 创建其他API服务
-                service = createApiService(config, voiceName, OUT_PUT_PATH);
-                serviceCache.put(cacheKey, service);
-                return service;
-            } catch (Exception e) {
-                logger.error("创建{}服务失败", provider, e);
-                return getDefaultTtsService(); // 失败时返回默认服务
-            }
         }
+
+        var service = createApiService(config, voiceName, OUT_PUT_PATH);
+        serviceCache.put(cacheKey, service);
+        return service;
     }
 
     /**
      * 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();
-
-        // 如果是Edge，直接返回Edge服务
-        if (DEFAULT_PROVIDER.equals(provider)) {
-            return new EdgeTtsService(voiceName, outputPath);
-        } else if ("aliyun".equals(provider)) {
-            return new AliyunTtsService(config, voiceName, outputPath);
-        } else if ("volcengine".equals(provider)) {
-            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);
-           * }
-           */
-
-        logger.warn("不支持的TTS服务提供商: {}", provider);
-        return null;
+        // Make sure output dir exists
+        ensureOutputPath(outputPath);
+
+        return switch (config.getProvider()) {
+            case "aliyun" -> new AliyunTtsService(config, voiceName, outputPath);
+            case "volcengine" -> new VolcengineTtsService(config, voiceName, outputPath);
+            case "xfyun" -> new XfyunTtsService(config, voiceName, outputPath);
+            case "minimax" -> new MiniMaxTtsService(config, voiceName, outputPath);
+            default -> new EdgeTtsService(voiceName, outputPath);
+        };
     }
 
     public void removeCache(SysConfig config) {
@@ -136,4 +87,8 @@ 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'");
-    }
 }