feat(model/cosyvoice): support hot_fix params and max_prompt_audio_length (#191)

Author: songguocola

src/main/java/com/alibaba/dashscope/aigc/multimodalconversation/MultiModalConversationParam.java

@@ -182,6 +182,10 @@ public JsonObject getInput() {
       jsonObject.addProperty(ApiKeywords.VOICE, voice.getValue());
     }
 
+    if (parameters != null && !parameters.isEmpty() && parameters.containsKey(ApiKeywords.VOICE)) {
+      jsonObject.addProperty(ApiKeywords.VOICE, (String) parameters.get(ApiKeywords.VOICE));
+    }
+
     if (languageType != null) {
       jsonObject.addProperty(ApiKeywords.LANGUAGE_TYPE, languageType);
     }

src/main/java/com/alibaba/dashscope/audio/tts/SpeechSynthesisApiKeywords.java

@@ -6,7 +6,9 @@ public class SpeechSynthesisApiKeywords {
   public static final String TEXT_TYPE = "text_type";
 
   public static final String FORMAT = "format";
+
   public static final String BIT_RATE = "bit_rate";
+
   public static final String VOICE = "voice";
 
   public static final String SAMPLE_RATE = "sample_rate";
@@ -21,6 +23,12 @@ public class SpeechSynthesisApiKeywords {
 
   public static final String PHONEME_TIMESTAMP = "phoneme_timestamp_enabled";
 
+  public static final String HOT_FIX = "hot_fix";
+
+  public static final String PRONUNCIATION = "pronunciation";
+
+  public static final String REPLACE = "replace";
+
   public static final String SENTENCE = "sentence";
 
   public static final String WORDS = "words";

src/main/java/com/alibaba/dashscope/audio/ttsv2/ParamHotFix.java

@@ -0,0 +1,62 @@
+// Copyright (c) Alibaba, Inc. and its affiliates.
+
+package com.alibaba.dashscope.audio.ttsv2;
+
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import lombok.AllArgsConstructor;
+import lombok.Data;
+
+/** Hot fix configuration for speech synthesis, including pronunciation and replace rules. */
+@Data
+public class ParamHotFix {
+
+  /** Pronunciation rules to customize specific words. */
+  private List<PronunciationItem> pronunciation;
+
+  /** Replace rules to replace specific words with others. */
+  private List<ReplaceItem> replace;
+
+  public ArrayList<Object> getPronunciation() {
+    if (pronunciation == null || pronunciation.isEmpty()) {
+      return null;
+    }
+    ArrayList<Object> pronunciationList = new ArrayList<>();
+    for (PronunciationItem item : pronunciation) {
+      HashMap<String, String> pronunciationItem = new HashMap<>();
+      pronunciationItem.put(item.getText(), item.getPinyin());
+      pronunciationList.add(pronunciationItem);
+    }
+
+    return pronunciationList;
+  }
+
+  public ArrayList<Object> getReplace() {
+    if (replace == null || replace.isEmpty()) {
+      return null;
+    }
+    ArrayList<Object> replaceList = new ArrayList<>();
+    for (ReplaceItem item : replace) {
+      HashMap<String, String> replaceItem = new HashMap<>();
+      replaceItem.put(item.getText(), item.getReplacement());
+      replaceList.add(replaceItem);
+    }
+
+    return replaceList;
+  }
+
+  @Data
+  @AllArgsConstructor
+  public static class PronunciationItem {
+    private String text;
+    private String pinyin;
+  }
+
+  @Data
+  @AllArgsConstructor
+  public static class ReplaceItem {
+    private String text;
+    private String replacement;
+  }
+}

src/main/java/com/alibaba/dashscope/audio/ttsv2/SpeechSynthesisParam.java

@@ -55,6 +55,8 @@ public class SpeechSynthesisParam extends FullDuplexServiceParam {
   @Builder.Default private List<String> languageHints = null;
   /** synthesis style */
   @Builder.Default private int style = 0;
+  /** Hot fix configuration for pronunciation and replace rules. */
+  @Builder.Default private ParamHotFix hotFix = null;
 
   @Override
   public Map<String, Object> getParameters() {
@@ -83,6 +85,20 @@ public Map<String, Object> getParameters() {
     if (getStyle() != 0) {
       params.put(SpeechSynthesisApiKeywords.STYLE, getStyle());
     }
+    // Add hot fix parameters if present
+    if (getHotFix() != null) {
+      Map<String, Object> hotFixParams = new HashMap<>();
+      if (getHotFix().getPronunciation() != null && !getHotFix().getPronunciation().isEmpty()) {
+        hotFixParams.put(SpeechSynthesisApiKeywords.PRONUNCIATION, getHotFix().getPronunciation());
+      }
+      if (getHotFix().getReplace() != null && !getHotFix().getReplace().isEmpty()) {
+        hotFixParams.put(SpeechSynthesisApiKeywords.REPLACE, getHotFix().getReplace());
+      }
+      if (!hotFixParams.isEmpty()) {
+        params.put(SpeechSynthesisApiKeywords.HOT_FIX, hotFixParams);
+      }
+    }
+
     params.putAll(parameters);
     return params;
   }

src/main/java/com/alibaba/dashscope/audio/ttsv2/enrollment/VoiceEnrollmentParam.java

@@ -25,6 +25,8 @@ public class VoiceEnrollmentParam extends HalfDuplexServiceParam {
 
   private int pageIndex;
   private int pageSize;
+  /** Maximum length of prompt audio in seconds. */
+  private float maxPromptAudioLength = 10.0f;
 
   protected VoiceEnrollmentParam(HalfDuplexServiceParamBuilder<?, ?> b) {
     super(b);
@@ -50,6 +52,9 @@ public JsonObject getInput() {
         if (languageHints != null) {
           input.add("language_hints", JsonUtils.toJsonArray(languageHints));
         }
+        if (maxPromptAudioLength > 0) {
+          input.addProperty("max_prompt_audio_length", maxPromptAudioLength);
+        }
         break;
       case LIST:
         input.addProperty(ApiKeywords.ACTION, operationType.getValue());

src/main/java/com/alibaba/dashscope/audio/ttsv2/enrollment/VoiceEnrollmentService.java

@@ -144,6 +144,7 @@ public Voice createVoice(
             .languageHints(customParam.getLanguageHints())
             .headers(customParam.getHeaders())
             .resources(customParam.getResources())
+            .maxPromptAudioLength(customParam.getMaxPromptAudioLength())
             .parameters(customParam.getParameters())
             .workspace(customParam.getWorkspace())
             .build();

src/test/java/com/alibaba/dashscope/TestTtsV2SpeechSynthesizer.java

@@ -5,6 +5,7 @@
 import static org.junit.Assert.assertEquals;
 
 import com.alibaba.dashscope.audio.tts.SpeechSynthesisResult;
+import com.alibaba.dashscope.audio.ttsv2.ParamHotFix;
 import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisAudioFormat;
 import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;
 import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;
@@ -129,6 +130,16 @@ public void testStreamingCall() {
 
     // 获取 URL
     String url = mockServer.url("/binary").toString();
+    ParamHotFix hotFix = new ParamHotFix();
+    ArrayList<ParamHotFix.PronunciationItem> pronunciations = new ArrayList<>();
+    pronunciations.add(new ParamHotFix.PronunciationItem("今天", "jin1 tian1"));
+    pronunciations.add(new ParamHotFix.PronunciationItem("草地", "cao3 di4"));
+    hotFix.setPronunciation(pronunciations);
+
+    ArrayList<ParamHotFix.ReplaceItem> replaces = new ArrayList<>();
+    replaces.add(new ParamHotFix.ReplaceItem("草地", "草弟"));
+    replaces.add(new ParamHotFix.ReplaceItem("惠州", "汇州"));
+    hotFix.setReplace(replaces);
 
     // 在真实世界中，你会在这里做 HTTP 请求，并得到响应
     System.out.println("Mock Server is running at: " + url);
@@ -138,6 +149,7 @@ public void testStreamingCall() {
             .model("cosyvoice-v1")
             .voice("longxiaochun")
             .format(SpeechSynthesisAudioFormat.MP3_16000HZ_MONO_128KBPS)
+            .hotFix(hotFix)
             .build();
     SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, callback);
     synthesizer.setStartedTimeout(1000);