Added support for Fish Audio S2.1

__init__.py

@@ -60,6 +60,7 @@ from .modules.textInferenceInputsImages import RunwareTextInferenceInputsImages
 from .modules.textInferenceInputsVideos import RunwareTextInferenceInputsVideos
 from .modules.audioSections import RunwareAudioSections
 from .modules.audioInferenceInputs import audioInferenceInputs
+from .modules.audioInferenceReferenceVoices import RunwareAudioInferenceReferenceVoices
 from .modules.audioSettings import RunwareAudioSettings
 from .modules.audioSettingsVoiceModify import RunwareAudioSettingsVoiceModify
 from .modules.providerSettings.elevenlabsProviderSettings import RunwareElevenLabsProviderSettings
@@ -189,6 +190,7 @@ NODE_CLASS_MAPPINGS = {
     "Runware Text Inference Inputs Videos": RunwareTextInferenceInputsVideos,
     "Runware Audio Sections": RunwareAudioSections,
     "Runware Audio Inference Inputs": audioInferenceInputs,
+    "Runware Audio Inference Inputs Reference Audio": RunwareAudioInferenceReferenceVoices,
     "Runware Audio Inference Settings": RunwareAudioSettings,
     "Runware Audio Inference Settings Voice Modify": RunwareAudioSettingsVoiceModify,
     "Runware ElevenLabs Provider Settings": RunwareElevenLabsProviderSettings,

clientlibs/main.js

@@ -1,6 +1,6 @@
 import { app } from "../../scripts/app.js";
 import { api } from "../../scripts/api.js";
-import { promptEnhanceHandler, syncDimensionsNodeHandler, searchNodeHandler, APIKeyHandler, captionNodeHandler, saveTextHandler, mediaUUIDHandler, save3DFilepathHandler, videoTranscriptionHandler, videoOutputsHandler, handleCustomErrors, videoInferenceDimensionsHandler, videoModelSearchFilterHandler, audioModelSearchFilterHandler, textModelSearchFilterHandler, vectorizeModelSearchFilterHandler, vectorizeToggleHandler, useParameterToggleHandler, imageInferenceToggleHandler, imageInferenceAdvancedFeaturesToggleHandler, watermarkAdvancedFeatureToggleHandler, videoInferenceSpeechInputToggleHandler, regionalPromptingRegionsToggleHandler, upscalerToggleHandler, imageUpscalerSettingsToggleHandler, videoUpscalerToggleHandler, audioInferenceToggleHandler, audioInferenceSpeechToggleHandler, audioSettingsToggleHandler, textInferenceSettingsToggleHandler, videoSettingsToggleHandler, videoInferenceSettingsTtsToggleHandler, videoInferenceSettingsActiveSpeakerDetectionToggleHandler, videoInferenceSettingsActiveSpeakerBoundingBoxesToggleHandler, videoInferenceSettingsSegmentsToggleHandler, acceleratorOptionsToggleHandler, bytedanceProviderSettingsToggleHandler, xaiProviderSettingsToggleHandler, viduProviderSettingsToggleHandler, sourcefulProviderSettingsToggleHandler, sourcefulProviderSettingsFontsToggleHandler, threeDInferenceToggleHandler, threeDInferenceSettingsToggleHandler, threeDInferenceSettingsLatToggleHandler, ultralyticsProviderSettingsToggleHandler, openaiProviderSettingsToggleHandler, lightricksProviderSettingsToggleHandler, klingProviderSettingsToggleHandler, lumaProviderSettingsToggleHandler, briaProviderSettingsToggleHandler, pixverseProviderSettingsToggleHandler, alibabaProviderSettingsToggleHandler, mireloProviderSettingsToggleHandler, googleProviderSettingsToggleHandler, syncProviderSettingsToggleHandler, syncSegmentToggleHandler, settingsToggleHandler, outpaintSettingsToggleHandler, safetyInputsToggleHandler, imageInferenceSettingsColorPaletteToggleHandler, audioInputToggleHandler, speechInputToggleHandler, briaProviderMaskToggleHandler, wanAnimateAdvancedFeatureSettingsToggleHandler, videoAdvancedFeatureInputsToggleHandler, audioInferenceInputsToggleHandler, audioInferenceSpeechVoicesToggleHandler, referenceVideosToggleHandler } from "./utils.js";
+import { promptEnhanceHandler, syncDimensionsNodeHandler, searchNodeHandler, APIKeyHandler, captionNodeHandler, saveTextHandler, mediaUUIDHandler, save3DFilepathHandler, videoTranscriptionHandler, videoOutputsHandler, handleCustomErrors, videoInferenceDimensionsHandler, videoModelSearchFilterHandler, audioModelSearchFilterHandler, textModelSearchFilterHandler, vectorizeModelSearchFilterHandler, vectorizeToggleHandler, useParameterToggleHandler, imageInferenceToggleHandler, imageInferenceAdvancedFeaturesToggleHandler, watermarkAdvancedFeatureToggleHandler, videoInferenceSpeechInputToggleHandler, regionalPromptingRegionsToggleHandler, upscalerToggleHandler, imageUpscalerSettingsToggleHandler, videoUpscalerToggleHandler, audioInferenceToggleHandler, audioInferenceSpeechToggleHandler, audioSettingsToggleHandler, textInferenceSettingsToggleHandler, videoSettingsToggleHandler, videoInferenceSettingsTtsToggleHandler, videoInferenceSettingsActiveSpeakerDetectionToggleHandler, videoInferenceSettingsActiveSpeakerBoundingBoxesToggleHandler, videoInferenceSettingsSegmentsToggleHandler, acceleratorOptionsToggleHandler, bytedanceProviderSettingsToggleHandler, xaiProviderSettingsToggleHandler, viduProviderSettingsToggleHandler, sourcefulProviderSettingsToggleHandler, sourcefulProviderSettingsFontsToggleHandler, threeDInferenceToggleHandler, threeDInferenceSettingsToggleHandler, threeDInferenceSettingsLatToggleHandler, ultralyticsProviderSettingsToggleHandler, openaiProviderSettingsToggleHandler, lightricksProviderSettingsToggleHandler, klingProviderSettingsToggleHandler, lumaProviderSettingsToggleHandler, briaProviderSettingsToggleHandler, pixverseProviderSettingsToggleHandler, alibabaProviderSettingsToggleHandler, mireloProviderSettingsToggleHandler, googleProviderSettingsToggleHandler, syncProviderSettingsToggleHandler, syncSegmentToggleHandler, settingsToggleHandler, outpaintSettingsToggleHandler, safetyInputsToggleHandler, imageInferenceSettingsColorPaletteToggleHandler, audioInputToggleHandler, speechInputToggleHandler, briaProviderMaskToggleHandler, wanAnimateAdvancedFeatureSettingsToggleHandler, videoAdvancedFeatureInputsToggleHandler, audioInferenceInputsToggleHandler, audioInferenceReferenceVoiceToggleHandler, audioInferenceSpeechVoicesToggleHandler, referenceVideosToggleHandler } from "./utils.js";
 import { RUNWARE_NODE_TYPES, RUNWARE_NODE_PROPS, SEARCH_TERMS } from "./types.js";
 
 const nodeInitList = [];
@@ -163,6 +163,8 @@ app.registerExtension({
             videoAdvancedFeatureInputsToggleHandler(node);
         } else if(nodeClass === RUNWARE_NODE_TYPES.AUDIOINFERENCEINPUTS) {
             audioInferenceInputsToggleHandler(node);
+        } else if(nodeClass === RUNWARE_NODE_TYPES.AUDIOINFERENCEINPUTSREFERENCEAUDIO) {
+            audioInferenceReferenceVoiceToggleHandler(node);
         } else if(nodeClass === RUNWARE_NODE_TYPES.REFERENCEVIDEOS ||
                   nodeClass === RUNWARE_NODE_TYPES.REFERENCEVIDEOS_LEGACY) {
             referenceVideosToggleHandler(node);

clientlibs/types.js

@@ -96,6 +96,7 @@ const RUNWARE_NODE_TYPES = {
     TEXTINFERENCEINPUTSVIDEOS: "Runware Text Inference Inputs Videos",
     AUDIOSECTIONS: "Runware Audio Sections",
     AUDIOINFERENCEINPUTS: "Runware Audio Inference Inputs",
+    AUDIOINFERENCEINPUTSREFERENCEAUDIO: "Runware Audio Inference Inputs Reference Audio",
     AUDIOSETTINGS: "Runware Audio Inference Settings",
     AUDIOSETTINGSVOICEMODIFY: "Runware Audio Inference Settings Voice Modify",
     PIXVERSEPROVIDERSETTINGS: "Runware Pixverse Provider Settings",
@@ -572,6 +573,10 @@ const RUNWARE_NODE_PROPS = {
         bgColor: DEFAULT_BGCOLOR,
         colorModeOnly: true,
     },
+    [RUNWARE_NODE_TYPES.AUDIOINFERENCEINPUTSREFERENCEAUDIO]: {
+        bgColor: DEFAULT_BGCOLOR,
+        colorModeOnly: true,
+    },
     [RUNWARE_NODE_TYPES.AUDIOSETTINGS]: {
         bgColor: DEFAULT_BGCOLOR,
     },

clientlibs/utils.js

@@ -1298,6 +1298,26 @@ function audioSettingsToggleHandler(settingsNode) {
     const cfgIntervalStartWidget = settingsNode.widgets.find(w => w && w.name === "cfgIntervalStart");
     const useCfgIntervalEndWidget = settingsNode.widgets.find(w => w && w.name === "useCfgIntervalEnd");
     const cfgIntervalEndWidget = settingsNode.widgets.find(w => w && w.name === "cfgIntervalEnd");
+    const useNormalizeLoudnessWidget = settingsNode.widgets.find(w => w && w.name === "useNormalizeLoudness");
+    const normalizeLoudnessWidget = settingsNode.widgets.find(w => w && w.name === "normalizeLoudness");
+    const useTopPWidget = settingsNode.widgets.find(w => w && w.name === "useTopP");
+    const topPWidget = settingsNode.widgets.find(w => w && w.name === "topP");
+    const useChunkLengthWidget = settingsNode.widgets.find(w => w && w.name === "useChunkLength");
+    const chunkLengthWidget = settingsNode.widgets.find(w => w && w.name === "chunkLength");
+    const useMinChunkLengthWidget = settingsNode.widgets.find(w => w && w.name === "useMinChunkLength");
+    const minChunkLengthWidget = settingsNode.widgets.find(w => w && w.name === "minChunkLength");
+    const useNormalizeWidget = settingsNode.widgets.find(w => w && w.name === "useNormalize");
+    const normalizeWidget = settingsNode.widgets.find(w => w && w.name === "normalize");
+    const useLatencyWidget = settingsNode.widgets.find(w => w && w.name === "useLatency");
+    const latencyWidget = settingsNode.widgets.find(w => w && w.name === "latency");
+    const useMaxTokensWidget = settingsNode.widgets.find(w => w && w.name === "useMaxTokens");
+    const maxTokensWidget = settingsNode.widgets.find(w => w && w.name === "maxTokens");
+    const useRepetitionPenaltyWidget = settingsNode.widgets.find(w => w && w.name === "useRepetitionPenalty");
+    const repetitionPenaltyWidget = settingsNode.widgets.find(w => w && w.name === "repetitionPenalty");
+    const useConditionOnPreviousChunksWidget = settingsNode.widgets.find(w => w && w.name === "useConditionOnPreviousChunks");
+    const conditionOnPreviousChunksWidget = settingsNode.widgets.find(w => w && w.name === "conditionOnPreviousChunks");
+    const useEarlyStopThresholdWidget = settingsNode.widgets.find(w => w && w.name === "useEarlyStopThreshold");
+    const earlyStopThresholdWidget = settingsNode.widgets.find(w => w && w.name === "earlyStopThreshold");
 
     function toggleWidgetState(useWidget, paramWidget, paramName) {
         if (!useWidget || !paramWidget) return;
@@ -1338,6 +1358,16 @@ function audioSettingsToggleHandler(settingsNode) {
     if (useTranscriptWidget && transcriptWidget) toggleWidgetState(useTranscriptWidget, transcriptWidget, "transcript");
     if (useCfgIntervalStartWidget && cfgIntervalStartWidget) toggleWidgetState(useCfgIntervalStartWidget, cfgIntervalStartWidget, "cfgIntervalStart");
     if (useCfgIntervalEndWidget && cfgIntervalEndWidget) toggleWidgetState(useCfgIntervalEndWidget, cfgIntervalEndWidget, "cfgIntervalEnd");
+    if (useNormalizeLoudnessWidget && normalizeLoudnessWidget) toggleWidgetState(useNormalizeLoudnessWidget, normalizeLoudnessWidget, "normalizeLoudness");
+    if (useTopPWidget && topPWidget) toggleWidgetState(useTopPWidget, topPWidget, "topP");
+    if (useChunkLengthWidget && chunkLengthWidget) toggleWidgetState(useChunkLengthWidget, chunkLengthWidget, "chunkLength");
+    if (useMinChunkLengthWidget && minChunkLengthWidget) toggleWidgetState(useMinChunkLengthWidget, minChunkLengthWidget, "minChunkLength");
+    if (useNormalizeWidget && normalizeWidget) toggleWidgetState(useNormalizeWidget, normalizeWidget, "normalize");
+    if (useLatencyWidget && latencyWidget) toggleWidgetState(useLatencyWidget, latencyWidget, "latency");
+    if (useMaxTokensWidget && maxTokensWidget) toggleWidgetState(useMaxTokensWidget, maxTokensWidget, "maxTokens");
+    if (useRepetitionPenaltyWidget && repetitionPenaltyWidget) toggleWidgetState(useRepetitionPenaltyWidget, repetitionPenaltyWidget, "repetitionPenalty");
+    if (useConditionOnPreviousChunksWidget && conditionOnPreviousChunksWidget) toggleWidgetState(useConditionOnPreviousChunksWidget, conditionOnPreviousChunksWidget, "conditionOnPreviousChunks");
+    if (useEarlyStopThresholdWidget && earlyStopThresholdWidget) toggleWidgetState(useEarlyStopThresholdWidget, earlyStopThresholdWidget, "earlyStopThreshold");
 }
 
 function textInferenceSettingsToggleHandler(settingsNode) {
@@ -3387,6 +3417,9 @@ function audioModelSearchFilterHandler(audioModelSearchNode) {
         "Google": [
             "google:gemini@3.1-flash-tts (Gemini 3.1 Flash TTS)",
         ],
+        "Fish": [
+            "fishaudio:s2.1@pro (Fish Audio S2.1 Pro)",
+        ],
     };
 
     function filterModelList() {
@@ -4691,6 +4724,43 @@ function regionalPromptingRegionsToggleHandler(regionsNode) {
     }
 }
 
+function audioInferenceReferenceVoiceToggleHandler(referenceVoiceNode) {
+    if (!referenceVoiceNode?.widgets) return;
+    if (referenceVoiceNode._audioInferenceReferenceVoiceToggleHandlerRegistered) return;
+    referenceVoiceNode._audioInferenceReferenceVoiceToggleHandlerRegistered = true;
+
+    function toggleWidgetState(useWidget, paramWidget, paramName) {
+        if (!useWidget || !paramWidget) return;
+
+        function applyState() {
+            const enabled = useWidget.value === true;
+            toggleWidgetEnabled(paramWidget, enabled, referenceVoiceNode);
+            if (paramWidget.options && paramWidget.options.element) {
+                paramWidget.options.element.disabled = !enabled;
+                paramWidget.options.element.style.opacity = enabled ? "1" : "0.5";
+                paramWidget.options.element.style.pointerEvents = enabled ? "auto" : "none";
+            }
+            referenceVoiceNode.setDirtyCanvas(true);
+        }
+
+        appendWidgetCB(useWidget, () => setTimeout(applyState, 50));
+        setTimeout(applyState, 100);
+    }
+
+    for (let i = 1; i <= 4; i++) {
+        const useReferenceVoiceWidget = referenceVoiceNode.widgets.find((w) => w && w.name === `useReferenceVoice${i}`);
+        const audioWidget = referenceVoiceNode.widgets.find((w) => w && w.name === `audio${i}`);
+        const textWidget = referenceVoiceNode.widgets.find((w) => w && w.name === `text${i}`);
+
+        if (useReferenceVoiceWidget && audioWidget) {
+            toggleWidgetState(useReferenceVoiceWidget, audioWidget, `audio${i}`);
+        }
+        if (useReferenceVoiceWidget && textWidget) {
+            toggleWidgetState(useReferenceVoiceWidget, textWidget, `text${i}`);
+        }
+    }
+}
+
 function audioInferenceInputsToggleHandler(audioInputsNode) {
     if (!audioInputsNode?.widgets) return;
 
@@ -5058,6 +5128,7 @@ export {
     wanAnimateAdvancedFeatureSettingsToggleHandler,
     videoAdvancedFeatureInputsToggleHandler,
     audioInferenceInputsToggleHandler,
+    audioInferenceReferenceVoiceToggleHandler,
     audioInferenceSpeechVoicesToggleHandler,
     referenceVideosToggleHandler,
 };

modules/audioInferenceInputs.py

@@ -52,13 +52,21 @@ class audioInferenceInputs:
                 "tooltip": f"Audio URL or mediaUUID for the {ordinal} audio. Only used when 'Use Audios' is enabled.",
                 "default": "",
             })
+
+        optionalInputs["Reference Voice"] = ("RUNWAREAUDIOINFERENCEREFERENCEVOICES", {
+            "tooltip": "Connect Runware Audio Inference Inputs Reference Audio for zero-shot voice cloning (inputs.referenceVoices).",
+        })
         
         return {
             "required": {},
             "optional": optionalInputs
         }
 
-    DESCRIPTION = "Configure custom inputs for Runware Audio Inference, including optional single or multiple audio URL/mediaUUID (inputs.audio or inputs.audios), and single or multiple video inputs for audio extraction or generation."
+    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."
+    )
     FUNCTION = "createInputs"
     RETURN_TYPES = ("RUNWAREAUDIOINFERENCEINPUTS",)
     RETURN_NAMES = ("Audio Inference Inputs",)
@@ -72,6 +80,7 @@ class audioInferenceInputs:
         useVideo = kwargs.get("useVideo", False)
         video = kwargs.get("Video", None)
         useVideos = kwargs.get("useVideos", False)
+        referenceVoices = kwargs.get("Reference Voice", None)
         
         inputs = {}
 
@@ -106,6 +115,9 @@ class audioInferenceInputs:
             
             if len(videoList) > 0:
                 inputs["videos"] = videoList
+
+        if referenceVoices is not None and isinstance(referenceVoices, list) and len(referenceVoices) > 0:
+            inputs["referenceVoices"] = referenceVoices
         
         return (inputs,)
 

modules/audioInferenceReferenceVoices.py

@@ -0,0 +1,74 @@
+"""
+Runware Audio Inference Inputs Reference Audio node.
+Builds inputs.referenceVoices for zero-shot voice cloning (up to 4 entries).
+"""
+
+from typing import Any, Dict, List
+
+from .utils import runwareUtils as rwUtils
+
+
+class RunwareAudioInferenceReferenceVoices:
+    """Build inputs.referenceVoices[] for Fish Audio and other TTS models."""
+
+    MAX_REFERENCE_VOICES = 4
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        optional_inputs = {}
+        for i in range(1, cls.MAX_REFERENCE_VOICES + 1):
+            ordinal = rwUtils.getOrdinal(i)
+            optional_inputs[f"useReferenceVoice{i}"] = ("BOOLEAN", {
+                "default": False,
+                "tooltip": f"Enable to include the {ordinal} reference voice in inputs.referenceVoices.",
+            })
+            optional_inputs[f"audio{i}"] = ("STRING", {
+                "default": "",
+                "tooltip": f"Reference audio clip ({ordinal}) as media UUID, URL, or base64. Required when enabled.",
+            })
+            optional_inputs[f"text{i}"] = ("STRING", {
+                "multiline": True,
+                "default": "",
+                "tooltip": f"Transcript of the {ordinal} reference audio clip (1–1000 characters). Required when enabled.",
+            })
+
+        return {
+            "required": {},
+            "optional": optional_inputs,
+        }
+
+    RETURN_TYPES = ("RUNWAREAUDIOINFERENCEREFERENCEVOICES",)
+    RETURN_NAMES = ("referenceVoices",)
+    FUNCTION = "createReferenceVoices"
+    CATEGORY = "Runware/Audio"
+    DESCRIPTION = (
+        "Configure inputs.referenceVoices for zero-shot voice cloning (up to 4 entries). "
+        "Each entry: { \"audio\": \"<UUID/URL/base64>\", \"text\": \"<transcript>\" }. "
+        "Connect to Runware Audio Inference Inputs."
+    )
+
+    def createReferenceVoices(self, **kwargs) -> tuple[List[Dict[str, Any]]]:
+        reference_voices: List[Dict[str, Any]] = []
+
+        for i in range(1, self.MAX_REFERENCE_VOICES + 1):
+            if not kwargs.get(f"useReferenceVoice{i}", False):
+                continue
+
+            audio = (kwargs.get(f"audio{i}") or "").strip()
+            text = (kwargs.get(f"text{i}") or "").strip()
+
+            if not audio or not text:
+                continue
+
+            reference_voices.append({"audio": audio, "text": text})
+
+        return (reference_voices,)
+
+
+NODE_CLASS_MAPPINGS = {
+    "RunwareAudioInferenceReferenceVoices": RunwareAudioInferenceReferenceVoices,
+}
+
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "RunwareAudioInferenceReferenceVoices": "Runware Audio Inference Inputs Reference Audio",
+}

modules/audioModelSearch.py

@@ -47,6 +47,9 @@ class RunwareAudioModelSearch:
         "Google": [
             "google:gemini@3.1-flash-tts (Gemini 3.1 Flash TTS)",
         ],
+        "Fish": [
+            "fishaudio:s2.1@pro (Fish Audio S2.1 Pro)",
+        ],
     }
     
     MODEL_PROVIDERS = [
@@ -61,6 +64,7 @@ class RunwareAudioModelSearch:
         "MiniMax",
         "Inworld",
         "Google",
+        "Fish",
     ]
     
     @classmethod

modules/audioSettings.py

@@ -2,8 +2,9 @@
 Runware Audio Inference Settings Node
 Provides lyrics, lyricsOptimizer, instrumental, guidanceType, languageBoost, turbo, temperature, textNormalization,
 bpm, keyScale, timeSignature, vocalLanguage, coverConditioningScale, repaintingStart, repaintingEnd,
-cfgIntervalStart, cfgIntervalEnd, xVectorOnly, maxNewTokens, transcript,
-and more for Runware Audio Inference.
+cfgIntervalStart, cfgIntervalEnd, xVectorOnly, maxNewTokens, maxTokens, transcript,
+normalizeLoudness, topP, chunkLength, minChunkLength, normalize, latency, repetitionPenalty,
+conditionOnPreviousChunks, earlyStopThreshold, and more for Runware Audio Inference.
 """
 
 from typing import Dict, Any
@@ -256,6 +257,110 @@ class RunwareAudioSettings:
                     "max": 1.0,
                     "step": 0.01,
                 }),
+                "useNormalizeLoudness": ("BOOLEAN", {
+                    "tooltip": "Enable to include normalizeLoudness in settings.",
+                    "default": False,
+                }),
+                "normalizeLoudness": ("BOOLEAN", {
+                    "tooltip": "Normalize output loudness for consistent perceived volume. Only used when 'Use Normalize Loudness' is enabled.",
+                    "default": True,
+                    "label_on": "true",
+                    "label_off": "false",
+                }),
+                "useTopP": ("BOOLEAN", {
+                    "tooltip": "Enable to include topP in settings.",
+                    "default": False,
+                }),
+                "topP": ("FLOAT", {
+                    "tooltip": "Controls diversity via nucleus sampling. Must be between 0.0001 and 1. Only used when 'Use Top P' is enabled.",
+                    "default": 0.7,
+                    "min": 0.0001,
+                    "max": 1.0,
+                    "step": 0.0001,
+                }),
+                "useChunkLength": ("BOOLEAN", {
+                    "tooltip": "Enable to include chunkLength in settings.",
+                    "default": False,
+                }),
+                "chunkLength": ("INT", {
+                    "tooltip": "Text segment size for processing. Only used when 'Use Chunk Length' is enabled.",
+                    "default": 300,
+                    "min": 100,
+                    "max": 300,
+                    "step": 1,
+                }),
+                "useMinChunkLength": ("BOOLEAN", {
+                    "tooltip": "Enable to include minChunkLength in settings.",
+                    "default": False,
+                }),
+                "minChunkLength": ("INT", {
+                    "tooltip": "Minimum characters before splitting into a new chunk. Only used when 'Use Min Chunk Length' is enabled.",
+                    "default": 50,
+                    "min": 0,
+                    "max": 100,
+                    "step": 1,
+                }),
+                "useNormalize": ("BOOLEAN", {
+                    "tooltip": "Enable to include normalize in settings (text normalization for English and Chinese).",
+                    "default": False,
+                }),
+                "normalize": ("BOOLEAN", {
+                    "tooltip": "Normalizes text for English and Chinese, improving stability for numbers. Only used when 'Use Normalize' is enabled.",
+                    "default": True,
+                    "label_on": "true",
+                    "label_off": "false",
+                }),
+                "useLatency": ("BOOLEAN", {
+                    "tooltip": "Enable to include latency in settings.",
+                    "default": False,
+                }),
+                "latency": (["low", "normal", "balanced"], {
+                    "tooltip": "Latency / quality trade-off. normal = best quality, balanced = reduced latency, low = lowest latency. Only used when 'Use Latency' is enabled.",
+                    "default": "normal",
+                }),
+                "useMaxTokens": ("BOOLEAN", {
+                    "tooltip": "Enable to include maxTokens in settings (e.g. Fish Audio TTS).",
+                    "default": False,
+                }),
+                "maxTokens": ("INT", {
+                    "tooltip": "Maximum audio tokens to generate per text chunk. Only used when 'Use Max Tokens' is enabled.",
+                    "default": 1024,
+                    "min": 1,
+                    "max": 4294967295,
+                    "step": 1,
+                }),
+                "useRepetitionPenalty": ("BOOLEAN", {
+                    "tooltip": "Enable to include repetitionPenalty in settings.",
+                    "default": False,
+                }),
+                "repetitionPenalty": ("FLOAT", {
+                    "tooltip": "Penalty for repeating audio patterns. Values above 1.0 reduce repetition. Only used when 'Use Repetition Penalty' is enabled.",
+                    "default": 1.2,
+                    "min": 0.0,
+                    "max": 2.0,
+                    "step": 0.01,
+                }),
+                "useConditionOnPreviousChunks": ("BOOLEAN", {
+                    "tooltip": "Enable to include conditionOnPreviousChunks in settings.",
+                    "default": False,
+                }),
+                "conditionOnPreviousChunks": ("BOOLEAN", {
+                    "tooltip": "Use previous audio as context for voice consistency across chunks. Only used when 'Use Condition On Previous Chunks' is enabled.",
+                    "default": True,
+                    "label_on": "true",
+                    "label_off": "false",
+                }),
+                "useEarlyStopThreshold": ("BOOLEAN", {
+                    "tooltip": "Enable to include earlyStopThreshold in settings.",
+                    "default": False,
+                }),
+                "earlyStopThreshold": ("FLOAT", {
+                    "tooltip": "Early stopping threshold for batch processing. Only used when 'Use Early Stop Threshold' is enabled.",
+                    "default": 1.0,
+                    "min": 0.0,
+                    "max": 1.0,
+                    "step": 0.01,
+                }),
             }
         }
 
@@ -266,7 +371,9 @@ class RunwareAudioSettings:
     DESCRIPTION = (
         "Configure audio generation settings (lyrics, lyricsOptimizer, instrumental, guidanceType, languageBoost, turbo, temperature, textNormalization, "
         "bpm, keyScale, timeSignature, vocalLanguage, coverConditioningScale, repaintingStart, repaintingEnd, "
-        "cfgIntervalStart, cfgIntervalEnd, xVectorOnly, maxNewTokens, transcript, etc.) "
+        "cfgIntervalStart, cfgIntervalEnd, xVectorOnly, maxNewTokens, maxTokens, transcript, "
+        "normalizeLoudness, topP, chunkLength, minChunkLength, normalize, latency, repetitionPenalty, "
+        "conditionOnPreviousChunks, earlyStopThreshold, etc.) "
         "for Runware Audio Inference. Connect to Runware Audio Inference node."
     )
 
@@ -318,6 +425,24 @@ class RunwareAudioSettings:
         cfg_interval_start = kwargs.get("cfgIntervalStart", 0.0)
         use_cfg_interval_end = kwargs.get("useCfgIntervalEnd", False)
         cfg_interval_end = kwargs.get("cfgIntervalEnd", 1.0)
+        use_normalize_loudness = kwargs.get("useNormalizeLoudness", False)
+        use_top_p = kwargs.get("useTopP", False)
+        top_p = kwargs.get("topP", 0.7)
+        use_chunk_length = kwargs.get("useChunkLength", False)
+        chunk_length = kwargs.get("chunkLength", 300)
+        use_min_chunk_length = kwargs.get("useMinChunkLength", False)
+        min_chunk_length = kwargs.get("minChunkLength", 50)
+        use_normalize = kwargs.get("useNormalize", False)
+        normalize = kwargs.get("normalize", True)
+        use_latency = kwargs.get("useLatency", False)
+        latency = kwargs.get("latency", "normal")
+        use_max_tokens = kwargs.get("useMaxTokens", False)
+        max_tokens = kwargs.get("maxTokens", 1024)
+        use_repetition_penalty = kwargs.get("useRepetitionPenalty", False)
+        repetition_penalty = kwargs.get("repetitionPenalty", 1.2)
+        use_condition_on_previous_chunks = kwargs.get("useConditionOnPreviousChunks", False)
+        use_early_stop_threshold = kwargs.get("useEarlyStopThreshold", False)
+        early_stop_threshold = kwargs.get("earlyStopThreshold", 1.0)
 
         settings: Dict[str, Any] = {}
 
@@ -385,6 +510,36 @@ class RunwareAudioSettings:
         if use_cfg_interval_end:
             settings["cfgIntervalEnd"] = float(cfg_interval_end)
 
+        if use_normalize_loudness:
+            settings["normalizeLoudness"] = bool(kwargs.get("normalizeLoudness", True))
+
+        if use_top_p:
+            settings["topP"] = float(top_p)
+
+        if use_chunk_length:
+            settings["chunkLength"] = int(chunk_length)
+
+        if use_min_chunk_length:
+            settings["minChunkLength"] = int(min_chunk_length)
+
+        if use_normalize:
+            settings["normalize"] = bool(normalize)
+
+        if use_latency:
+            settings["latency"] = str(latency)
+
+        if use_max_tokens:
+            settings["maxTokens"] = int(max_tokens)
+
+        if use_repetition_penalty:
+            settings["repetitionPenalty"] = float(repetition_penalty)
+
+        if use_condition_on_previous_chunks:
+            settings["conditionOnPreviousChunks"] = bool(kwargs.get("conditionOnPreviousChunks", True))
+
+        if use_early_stop_threshold:
+            settings["earlyStopThreshold"] = float(early_stop_threshold)
+
         return (settings,)