feat: support multi speaker for Vertex AI

jscud · copybara-github · commit c50e47bfba44 · 2025-12-12T15:41:29.000-08:00
PiperOrigin-RevId: 843858005
diff --git a/src/main/java/com/google/genai/LiveConverters.java b/src/main/java/com/google/genai/LiveConverters.java
@@ -307,10 +307,7 @@ ObjectNode generationConfigToVertex(JsonNode fromObject, ObjectNode parentObject
       Common.setValueByPath(
           toObject,
           new String[] {"speechConfig"},
-          speechConfigToVertex(
-              JsonSerializable.toJsonNode(
-                  Common.getValueByPath(fromObject, new String[] {"speechConfig"})),
-              toObject));
+          Common.getValueByPath(fromObject, new String[] {"speechConfig"}));
     }
 
     if (Common.getValueByPath(fromObject, new String[] {"stopSequences"}) != null) {
@@ -1038,11 +1035,8 @@ ObjectNode liveConnectConfigToVertex(JsonNode fromObject, ObjectNode parentObjec
       Common.setValueByPath(
           parentObject,
           new String[] {"setup", "generationConfig", "speechConfig"},
-          speechConfigToVertex(
-              JsonSerializable.toJsonNode(
-                  Transformers.tLiveSpeechConfig(
-                      Common.getValueByPath(fromObject, new String[] {"speechConfig"}))),
-              toObject));
+          Transformers.tLiveSpeechConfig(
+              Common.getValueByPath(fromObject, new String[] {"speechConfig"})));
     }
 
     if (Common.getValueByPath(fromObject, new String[] {"thinkingConfig"}) != null) {
@@ -1480,32 +1474,6 @@ ObjectNode sessionResumptionConfigToMldev(JsonNode fromObject, ObjectNode parent
     return toObject;
   }
 
-  @ExcludeFromGeneratedCoverageReport
-  ObjectNode speechConfigToVertex(JsonNode fromObject, ObjectNode parentObject) {
-    ObjectNode toObject = JsonSerializable.objectMapper().createObjectNode();
-    if (Common.getValueByPath(fromObject, new String[] {"voiceConfig"}) != null) {
-      Common.setValueByPath(
-          toObject,
-          new String[] {"voiceConfig"},
-          Common.getValueByPath(fromObject, new String[] {"voiceConfig"}));
-    }
-
-    if (Common.getValueByPath(fromObject, new String[] {"languageCode"}) != null) {
-      Common.setValueByPath(
-          toObject,
-          new String[] {"languageCode"},
-          Common.getValueByPath(fromObject, new String[] {"languageCode"}));
-    }
-
-    if (!Common.isZero(
-        Common.getValueByPath(fromObject, new String[] {"multiSpeakerVoiceConfig"}))) {
-      throw new IllegalArgumentException(
-          "multiSpeakerVoiceConfig parameter is not supported in Vertex AI.");
-    }
-
-    return toObject;
-  }
-
   @ExcludeFromGeneratedCoverageReport
   ObjectNode toolToMldev(JsonNode fromObject, ObjectNode parentObject) {
     ObjectNode toObject = JsonSerializable.objectMapper().createObjectNode();
diff --git a/src/main/java/com/google/genai/Models.java b/src/main/java/com/google/genai/Models.java
@@ -1567,11 +1567,8 @@ ObjectNode generateContentConfigToVertex(
       Common.setValueByPath(
           toObject,
           new String[] {"speechConfig"},
-          speechConfigToVertex(
-              JsonSerializable.toJsonNode(
-                  Transformers.tSpeechConfig(
-                      Common.getValueByPath(fromObject, new String[] {"speechConfig"}))),
-              toObject));
+          Transformers.tSpeechConfig(
+              Common.getValueByPath(fromObject, new String[] {"speechConfig"})));
     }
 
     if (Common.getValueByPath(fromObject, new String[] {"audioTimestamp"}) != null) {
@@ -2980,10 +2977,7 @@ ObjectNode generationConfigToVertex(JsonNode fromObject, ObjectNode parentObject
       Common.setValueByPath(
           toObject,
           new String[] {"speechConfig"},
-          speechConfigToVertex(
-              JsonSerializable.toJsonNode(
-                  Common.getValueByPath(fromObject, new String[] {"speechConfig"})),
-              toObject));
+          Common.getValueByPath(fromObject, new String[] {"speechConfig"}));
     }
 
     if (Common.getValueByPath(fromObject, new String[] {"stopSequences"}) != null) {
@@ -4208,32 +4202,6 @@ ObjectNode segmentImageSourceToVertex(JsonNode fromObject, ObjectNode parentObje
     return toObject;
   }
 
-  @ExcludeFromGeneratedCoverageReport
-  ObjectNode speechConfigToVertex(JsonNode fromObject, ObjectNode parentObject) {
-    ObjectNode toObject = JsonSerializable.objectMapper().createObjectNode();
-    if (Common.getValueByPath(fromObject, new String[] {"voiceConfig"}) != null) {
-      Common.setValueByPath(
-          toObject,
-          new String[] {"voiceConfig"},
-          Common.getValueByPath(fromObject, new String[] {"voiceConfig"}));
-    }
-
-    if (Common.getValueByPath(fromObject, new String[] {"languageCode"}) != null) {
-      Common.setValueByPath(
-          toObject,
-          new String[] {"languageCode"},
-          Common.getValueByPath(fromObject, new String[] {"languageCode"}));
-    }
-
-    if (!Common.isZero(
-        Common.getValueByPath(fromObject, new String[] {"multiSpeakerVoiceConfig"}))) {
-      throw new IllegalArgumentException(
-          "multiSpeakerVoiceConfig parameter is not supported in Vertex AI.");
-    }
-
-    return toObject;
-  }
-
   @ExcludeFromGeneratedCoverageReport
   ObjectNode toolConfigToMldev(JsonNode fromObject, ObjectNode parentObject) {
     ObjectNode toObject = JsonSerializable.objectMapper().createObjectNode();
diff --git a/src/main/java/com/google/genai/types/MultiSpeakerVoiceConfig.java b/src/main/java/com/google/genai/types/MultiSpeakerVoiceConfig.java
@@ -30,11 +30,14 @@
 import java.util.List;
 import java.util.Optional;
 
-/** The configuration for the multi-speaker setup. This data type is not supported in Vertex AI. */
+/** Configuration for a multi-speaker text-to-speech request. */
 @AutoValue
 @JsonDeserialize(builder = MultiSpeakerVoiceConfig.Builder.class)
 public abstract class MultiSpeakerVoiceConfig extends JsonSerializable {
-  /** Required. All the enabled speaker voices. */
+  /**
+   * Required. A list of configurations for the voices of the speakers. Exactly two speaker voice
+   * configurations must be provided.
+   */
   @JsonProperty("speakerVoiceConfigs")
   public abstract Optional<List<SpeakerVoiceConfig>> speakerVoiceConfigs();
 
@@ -59,15 +62,17 @@ private static Builder create() {
     /**
      * Setter for speakerVoiceConfigs.
      *
-     * <p>speakerVoiceConfigs: Required. All the enabled speaker voices.
+     * <p>speakerVoiceConfigs: Required. A list of configurations for the voices of the speakers.
+     * Exactly two speaker voice configurations must be provided.
      */
     @JsonProperty("speakerVoiceConfigs")
     public abstract Builder speakerVoiceConfigs(List<SpeakerVoiceConfig> speakerVoiceConfigs);
 
     /**
      * Setter for speakerVoiceConfigs.
      *
-     * <p>speakerVoiceConfigs: Required. All the enabled speaker voices.
+     * <p>speakerVoiceConfigs: Required. A list of configurations for the voices of the speakers.
+     * Exactly two speaker voice configurations must be provided.
      */
     @CanIgnoreReturnValue
     public Builder speakerVoiceConfigs(SpeakerVoiceConfig... speakerVoiceConfigs) {
@@ -77,7 +82,8 @@ public Builder speakerVoiceConfigs(SpeakerVoiceConfig... speakerVoiceConfigs) {
     /**
      * Setter for speakerVoiceConfigs builder.
      *
-     * <p>speakerVoiceConfigs: Required. All the enabled speaker voices.
+     * <p>speakerVoiceConfigs: Required. A list of configurations for the voices of the speakers.
+     * Exactly two speaker voice configurations must be provided.
      */
     @CanIgnoreReturnValue
     public Builder speakerVoiceConfigs(SpeakerVoiceConfig.Builder... speakerVoiceConfigsBuilders) {
diff --git a/src/main/java/com/google/genai/types/SpeechConfig.java b/src/main/java/com/google/genai/types/SpeechConfig.java
@@ -39,8 +39,8 @@ public abstract class SpeechConfig extends JsonSerializable {
   public abstract Optional<String> languageCode();
 
   /**
-   * Optional. The configuration for the multi-speaker setup. It is mutually exclusive with the
-   * voice_config field. This field is not supported in Vertex AI.
+   * The configuration for a multi-speaker text-to-speech request. This field is mutually exclusive
+   * with `voice_config`.
    */
   @JsonProperty("multiSpeakerVoiceConfig")
   public abstract Optional<MultiSpeakerVoiceConfig> multiSpeakerVoiceConfig();
@@ -112,8 +112,8 @@ public Builder clearLanguageCode() {
     /**
      * Setter for multiSpeakerVoiceConfig.
      *
-     * <p>multiSpeakerVoiceConfig: Optional. The configuration for the multi-speaker setup. It is
-     * mutually exclusive with the voice_config field. This field is not supported in Vertex AI.
+     * <p>multiSpeakerVoiceConfig: The configuration for a multi-speaker text-to-speech request.
+     * This field is mutually exclusive with `voice_config`.
      */
     @JsonProperty("multiSpeakerVoiceConfig")
     public abstract Builder multiSpeakerVoiceConfig(
@@ -122,8 +122,8 @@ public abstract Builder multiSpeakerVoiceConfig(
     /**
      * Setter for multiSpeakerVoiceConfig builder.
      *
-     * <p>multiSpeakerVoiceConfig: Optional. The configuration for the multi-speaker setup. It is
-     * mutually exclusive with the voice_config field. This field is not supported in Vertex AI.
+     * <p>multiSpeakerVoiceConfig: The configuration for a multi-speaker text-to-speech request.
+     * This field is mutually exclusive with `voice_config`.
      */
     @CanIgnoreReturnValue
     public Builder multiSpeakerVoiceConfig(
