Added support for Gemini TTS

modules/audioInferenceSpeech.py

@@ -14,20 +14,6 @@ class RunwareAudioInferenceSpeech:
         "calm", "fluent", "whisper",
     ]
 
-    SPEECH_LANGUAGES = [
-        "Auto",
-        "Chinese",
-        "English",
-        "Japanese",
-        "Korean",
-        "German",
-        "French",
-        "Russian",
-        "Portuguese",
-        "Spanish",
-        "Italian",
-    ]
-
     def __init__(self):
         pass
 
@@ -50,7 +36,7 @@ class RunwareAudioInferenceSpeech:
                     "default": "",
                     "tooltip": "Voice identifier.",
                 }),
-                "useSpeed": ("BOOLEAN", {"default": True, "tooltip": "Include speed (speech.speed)"}),
+                "useSpeed": ("BOOLEAN", {"default": False, "tooltip": "Include speed (speech.speed)"}),
                 "speed": ("FLOAT", {
                     "default": 1.0,
                     "min": 0.5,
@@ -58,7 +44,7 @@ class RunwareAudioInferenceSpeech:
                     "step": 0.01,
                     "tooltip": "Speech rate multiplier (speech.speed)",
                 }),
-                "useVolume": ("BOOLEAN", {"default": True, "tooltip": "Include volume (speech.volume)"}),
+                "useVolume": ("BOOLEAN", {"default": False, "tooltip": "Include volume (speech.volume)"}),
                 "volume": ("FLOAT", {
                     "default": 1.0,
                     "min": 1.0,
@@ -87,9 +73,12 @@ class RunwareAudioInferenceSpeech:
                     "default": False,
                     "tooltip": "Enable to include language (speech.language).",
                 }),
-                "language": (cls.SPEECH_LANGUAGES, {
+                "language": ("STRING", {
                     "default": "Auto",
-                    "tooltip": "Language of the input text. Options: Auto, Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, Italian. Applies to all variants. Only used when 'Use Language' is enabled.",
+                    "tooltip": "Language of the input text (for example: en-US). Applies to all variants. Only used when 'Use Language' is enabled.",
+                }),
+                "voices": ("RUNWAREAUDIOINFERENCESPEECHVOICES", {
+                    "tooltip": "Optional multi-speaker voices array (speech.voices[]). If provided, it takes precedence over speech.voice.",
                 }),
             }
         }
@@ -119,10 +108,13 @@ class RunwareAudioInferenceSpeech:
         use_tone = kwargs.get("useTone", False)
         tone_str = kwargs.get("tone", "") or ""
         use_language = kwargs.get("useLanguage", False)
-        language = kwargs.get("language", "Auto")
+        language = (kwargs.get("language") or "Auto").strip()
+        voices = kwargs.get("voices")
 
         speech: Dict[str, Any] = {"text": text}
-        if use_voice:
+        if isinstance(voices, list) and len(voices) > 0:
+            speech["voices"] = voices
+        elif use_voice:
             speech["voice"] = voice
         if use_speed:
             speech["speed"] = speed

modules/audioInferenceSpeechVoices.py

@@ -0,0 +1,87 @@
+"""
+Runware Audio Inference Speech Voices node:
+Builds speech.voices[] for multi-speaker audio inference speech payloads.
+"""
+
+import re
+from typing import Any, Dict, List
+
+
+class RunwareAudioInferenceSpeechVoices:
+    """Build speech.voices[] list for Runware Audio Inference."""
+
+    MAX_VOICES = 4
+    _SPEAKER_ALIAS_PATTERN = re.compile(r"^[A-Za-z0-9]+$")
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        optional_inputs = {}
+        for i in range(1, cls.MAX_VOICES + 1):
+            optional_inputs[f"useVoice{i}"] = ("BOOLEAN", {
+                "default": i == 1,
+                "tooltip": f"Enable to include speaker/voice pair #{i} in speech.voices.",
+            })
+            optional_inputs[f"speaker{i}"] = ("STRING", {
+                "default": f"Speaker{i}",
+                "tooltip": "Unique alphanumeric speaker alias (speech.voices[].speaker).",
+            })
+            optional_inputs[f"voice{i}"] = ("STRING", {
+                "default": "",
+                "tooltip": "Voice name for this speaker (speech.voices[].voice).",
+            })
+
+        return {
+            "required": {},
+            "optional": optional_inputs,
+        }
+
+    RETURN_TYPES = ("RUNWAREAUDIOINFERENCESPEECHVOICES",)
+    RETURN_NAMES = ("voices",)
+    FUNCTION = "createVoices"
+    CATEGORY = "Runware/Audio"
+    DESCRIPTION = (
+        "Build speech.voices[] for multi-speaker synthesis. "
+        "Each item contains speaker (unique alphanumeric alias) and voice name."
+    )
+
+    def createVoices(self, **kwargs) -> tuple[List[Dict[str, str]]]:
+        voices: List[Dict[str, str]] = []
+        seen_speakers = set()
+
+        for i in range(1, self.MAX_VOICES + 1):
+            use_voice = bool(kwargs.get(f"useVoice{i}", i == 1))
+            if not use_voice:
+                continue
+
+            speaker = (kwargs.get(f"speaker{i}") or "").strip()
+            voice = (kwargs.get(f"voice{i}") or "").strip()
+
+            if not speaker or not voice:
+                raise Exception(
+                    f"Voice #{i} is enabled, but speaker{i} and voice{i} must both be provided."
+                )
+            if not self._SPEAKER_ALIAS_PATTERN.match(speaker):
+                raise Exception(
+                    f"speaker{i} must be alphanumeric only (got: '{speaker}')."
+                )
+            if speaker in seen_speakers:
+                raise Exception(
+                    f"Duplicate speaker alias '{speaker}'. Each speaker must be unique."
+                )
+
+            seen_speakers.add(speaker)
+            voices.append({
+                "speaker": speaker,
+                "voice": voice,
+            })
+
+        return (voices,)
+
+
+NODE_CLASS_MAPPINGS = {
+    "RunwareAudioInferenceSpeechVoices": RunwareAudioInferenceSpeechVoices,
+}
+
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "RunwareAudioInferenceSpeechVoices": "Runware Audio Inference Speech Voices",
+}

modules/audioModelSearch.py

@@ -40,6 +40,9 @@ class RunwareAudioModelSearch:
             "inworld:tts@1.5-mini (Inworld TTS-1.5 Mini)",
             "inworld:tts@1.5-max (Inworld TTS-1.5 Max)",
         ],
+        "Google": [
+            "google:gemini@3.1-flash-tts (Gemini 3.1 Flash TTS)",
+        ],
     }
     
     MODEL_PROVIDERS = [
@@ -53,6 +56,7 @@ class RunwareAudioModelSearch:
         "xAI",
         "MiniMax",
         "Inworld",
+        "Google",
     ]
     
     @classmethod