Added support for Seed Audio 1

modules/audioInferenceInputs.py

@@ -7,6 +7,8 @@ class audioInferenceInputs:
     
     MAX_VIDEOS = 4
     MAX_AUDIOS = 4
+    MAX_REFERENCE_IMAGES = 4
+    MAX_REFERENCE_AUDIOS = 4
     
     @classmethod
     def INPUT_TYPES(cls):
@@ -53,6 +55,23 @@ class audioInferenceInputs:
                 "default": "",
             })
 
+        for i in range(1, cls.MAX_REFERENCE_IMAGES + 1):
+            ordinal = rwUtils.getOrdinal(i)
+            optionalInputs[f"Reference Image {i}"] = ("IMAGE", {
+                "tooltip": f"Optional {ordinal} reference image for audio inference (inputs.referenceImages).",
+            })
+
+        optionalInputs["useReferenceAudios"] = ("BOOLEAN", {
+            "tooltip": "Enable to include reference audios in API request (inputs.referenceAudios array).",
+            "default": False,
+        })
+        for i in range(1, cls.MAX_REFERENCE_AUDIOS + 1):
+            ordinal = rwUtils.getOrdinal(i)
+            optionalInputs[f"Reference Audio {i}"] = ("STRING", {
+                "tooltip": f"Audio URL or mediaUUID for the {ordinal} reference audio (inputs.referenceAudios).",
+                "default": "",
+            })
+
         optionalInputs["Reference Voice"] = ("RUNWAREAUDIOINFERENCEREFERENCEVOICES", {
             "tooltip": "Connect Runware Audio Inference Inputs Reference Audio for zero-shot voice cloning (inputs.referenceVoices).",
         })
@@ -65,7 +84,8 @@ class audioInferenceInputs:
     DESCRIPTION = (
         "Configure custom inputs for Runware Audio Inference, including optional single or multiple audio URL/mediaUUID "
         "(inputs.audio or inputs.audios), reference voice for cloning (inputs.referenceVoices), "
-        "and single or multiple video inputs for audio extraction or generation."
+        "single or multiple video inputs for audio extraction or generation, up to 4 reference images "
+        "(inputs.referenceImages), and up to 4 reference audios (inputs.referenceAudios)."
     )
     FUNCTION = "createInputs"
     RETURN_TYPES = ("RUNWAREAUDIOINFERENCEINPUTS",)
@@ -80,6 +100,7 @@ class audioInferenceInputs:
         useVideo = kwargs.get("useVideo", False)
         video = kwargs.get("Video", None)
         useVideos = kwargs.get("useVideos", False)
+        useReferenceAudios = kwargs.get("useReferenceAudios", False)
         referenceVoices = kwargs.get("Reference Voice", None)
         
         inputs = {}
@@ -116,6 +137,23 @@ class audioInferenceInputs:
             if len(videoList) > 0:
                 inputs["videos"] = videoList
 
+        referenceImages = []
+        for i in range(1, self.MAX_REFERENCE_IMAGES + 1):
+            image = kwargs.get(f"Reference Image {i}", None)
+            if image is not None:
+                referenceImages.append(rwUtils.convertTensor2IMG(image))
+        if len(referenceImages) > 0:
+            inputs["referenceImages"] = referenceImages
+
+        if useReferenceAudios:
+            referenceAudios = []
+            for i in range(1, self.MAX_REFERENCE_AUDIOS + 1):
+                audioUrl = kwargs.get(f"Reference Audio {i}", None)
+                if audioUrl and isinstance(audioUrl, str) and audioUrl.strip() != "":
+                    referenceAudios.append(audioUrl.strip())
+            if len(referenceAudios) > 0:
+                inputs["referenceAudios"] = referenceAudios
+
         if referenceVoices is not None and isinstance(referenceVoices, list) and len(referenceVoices) > 0:
             inputs["referenceVoices"] = referenceVoices
         

modules/audioInferenceSpeech.py

@@ -23,8 +23,8 @@ class RunwareAudioInferenceSpeech:
             "required": {
                 "text": ("STRING", {
                     "multiline": True,
-                    "default": "Hello, this is a test of the speech synthesis model.",
-                    "tooltip": "Text to synthesize as speech (speech.text)",
+                    "default": "",
+                    "tooltip": "Text to synthesize as speech (speech.text). Optional when positivePrompt on Runware Audio Inference carries the script (e.g. Seed Audio).",
                 }),
             },
             "optional": {
@@ -38,19 +38,19 @@ class RunwareAudioInferenceSpeech:
                 }),
                 "useSpeed": ("BOOLEAN", {"default": False, "tooltip": "Include speed (speech.speed)"}),
                 "speed": ("FLOAT", {
-                    "default": 1.0,
-                    "min": 0.5,
+                    "default": 0.0,
+                    "min": 0.0,
                     "max": 2.0,
                     "step": 0.01,
-                    "tooltip": "Speech rate multiplier (speech.speed)",
+                    "tooltip": "Speech rate adjustment (speech.speed). 0 = no change for Seed Audio.",
                 }),
                 "useVolume": ("BOOLEAN", {"default": False, "tooltip": "Include volume (speech.volume)"}),
                 "volume": ("FLOAT", {
-                    "default": 1.0,
-                    "min": 1.0,
+                    "default": 0.0,
+                    "min": 0.0,
                     "max": 10.0,
                     "step": 0.1,
-                    "tooltip": "Volume level 1-10 (speech.volume)",
+                    "tooltip": "Volume adjustment (speech.volume). 0 = no change.",
                 }),
                 "usePitch": ("BOOLEAN", {"default": False, "tooltip": "Include pitch (speech.pitch)"}),
                 "pitch": ("FLOAT", {
@@ -111,10 +111,12 @@ class RunwareAudioInferenceSpeech:
         language = (kwargs.get("language") or "Auto").strip()
         voices = kwargs.get("voices")
 
-        speech: Dict[str, Any] = {"text": text}
+        speech: Dict[str, Any] = {}
+        if text:
+            speech["text"] = text
         if isinstance(voices, list) and len(voices) > 0:
             speech["voices"] = voices
-        elif use_voice:
+        elif use_voice and voice:
             speech["voice"] = voice
         if use_speed:
             speech["speed"] = speed
@@ -131,8 +133,6 @@ class RunwareAudioInferenceSpeech:
         if use_language:
             speech["language"] = language
 
-        if not text:
-            return ({},)
         return (speech,)
 
 

modules/audioModelSearch.py

@@ -27,6 +27,9 @@ class RunwareAudioModelSearch:
             "alibaba:qwen@3-tts-1.7b-customvoice (Qwen3 TTS 1.7B Custom Voice)",
             "alibaba:qwen@3-tts-1.7b-base (Qwen3 TTS 1.7B Base)",
         ],
+        "Bytedance": [
+            "bytedance:seed-audio@1.0 (Seed Audio 1.0)",
+        ],
         "Dia": [
             "runware:dia@v1.0 (Dia 1.6B)",
             "runware:dia2@2b (Dia2 2B)",
@@ -59,6 +62,7 @@ class RunwareAudioModelSearch:
         "Mirelo",
         "Ace",
         "Alibaba",
+        "Bytedance",
         "Dia",
         "xAI",
         "MiniMax",

modules/version.py

@@ -1 +1 @@
-__version__ = "1.9.1"
+__version__ = "1.9.2"