@fugood/bricks-project 2.21.8 → 2.21.9

package/compile/action-name-map.ts

@@ -142,6 +142,13 @@ export const templateActionNameMap = {
     },
   },
 
+  BRICK_VIDEO: {
+    BRICK_VIDEO_SEEK: {
+      seekTime: 'BRICK_VIDEO_SEEK_TIME',
+      play: 'BRICK_VIDEO_PLAY',
+    },
+  },
+
   BRICK_SLIDESHOW: {
     BRICK_SLIDESHOW_JUMP_TO_INDEX: {
       index: 'BRICK_SLIDESHOW_INDEX',
@@ -231,7 +238,12 @@ export const templateActionNameMap = {
       index: 'BRICK_ITEMS_INDEX',
     },
   },
-
+  BRICK_LOTTIE: {
+    BRICK_LOTTIE_PLAY: {
+      startFrame: 'BRICK_LOTTIE_START_FRAME',
+      endFrame: 'BRICK_LOTTIE_END_FRAME',
+    },
+  },
   BRICK_RIVE: {
     BRICK_RIVE_PLAY: {
       animationName: 'BRICK_RIVE_ANIMATION_NAME',
@@ -663,6 +675,11 @@ export const templateActionNameMap = {
       sessionCustomKey: 'GENERATOR_LLM_SESSION_CUSTOM_KEY',
     },
   },
+  GENERATOR_GGML_TTS: {
+    GENERATOR_GGML_TTS_GENERATE: {
+      text: 'GENERATOR_GGML_TTS_TEXT',
+    },
+  },
   GENERATOR_QNN_LLM: {
     GENERATOR_QNN_LLM_GENERATE: {
       prompt: 'GENERATOR_QNN_LLM_PROMPT',

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@fugood/bricks-project",
-  "version": "2.21.8",
+  "version": "2.21.9",
   "main": "index.ts",
   "scripts": {
     "build": "node scripts/build.js"

package/types/bricks.ts

@@ -865,6 +865,43 @@ export type BrickIcon = Brick &
     >
   }
 
+/* Play the video */
+export type BrickVideoActionPlay = Action & {
+  __actionName: 'BRICK_VIDEO_PLAY'
+}
+
+/* Seek the video */
+export type BrickVideoActionSeek = ActionWithParams & {
+  __actionName: 'BRICK_VIDEO_SEEK'
+  params?: Array<
+    | {
+        input: 'seekTime'
+        value?: number | DataLink | EventProperty
+        mapping?: string
+      }
+    | {
+        input: 'play'
+        value?: boolean | DataLink | EventProperty
+        mapping?: string
+      }
+  >
+}
+
+/* Pause the video */
+export type BrickVideoActionPause = Action & {
+  __actionName: 'BRICK_VIDEO_PAUSE'
+}
+
+/* Replay the video */
+export type BrickVideoActionReplay = Action & {
+  __actionName: 'BRICK_VIDEO_REPLAY'
+}
+
+/* Stop the video */
+export type BrickVideoActionStop = Action & {
+  __actionName: 'BRICK_VIDEO_STOP'
+}
+
 interface BrickVideoDef {
   /*
 Default property:
@@ -2108,6 +2145,43 @@ export type BrickItems = Brick &
     >
   }
 
+/* Play animation */
+export type BrickLottieActionPlay = ActionWithParams & {
+  __actionName: 'BRICK_LOTTIE_PLAY'
+  params?: Array<
+    | {
+        input: 'startFrame'
+        value?: number | DataLink | EventProperty
+        mapping?: string
+      }
+    | {
+        input: 'endFrame'
+        value?: number | DataLink | EventProperty
+        mapping?: string
+      }
+  >
+}
+
+/* Pause animation */
+export type BrickLottieActionPause = Action & {
+  __actionName: 'BRICK_LOTTIE_PAUSE'
+}
+
+/* Resume animation */
+export type BrickLottieActionResume = Action & {
+  __actionName: 'BRICK_LOTTIE_RESUME'
+}
+
+/* Stop animation */
+export type BrickLottieActionStop = Action & {
+  __actionName: 'BRICK_LOTTIE_STOP'
+}
+
+/* Reset animation */
+export type BrickLottieActionReset = Action & {
+  __actionName: 'BRICK_LOTTIE_RESET'
+}
+
 interface BrickLottieDef {
   /*
 Default property:
@@ -2990,6 +3064,10 @@ Default property:
     type?: 'image' | 'video' | DataLink
     /* Default image to display when no generated image is available */
     defaultImage?: string | DataLink
+    /* The hash of the default image */
+    defaultImageHash?: string | DataLink
+    /* The type of the default image hash */
+    defaultImageHashType?: 'md5' | 'sha1' | 'sha256' | DataLink
     /* The Lottie animation to show while generating */
     loadingAnimation?: string | DataLink
     /* The Lottie animation to show when an error occurs */

package/types/generators.ts

@@ -4848,6 +4848,10 @@ Default property:
   "model": "BricksDisplay/vits-eng",
   "modelType": "auto",
   "vocoderModel": "speecht5_hifigan",
+  "maxLength": 4096,
+  "temperature": 0.1,
+  "repetitionPenalty": 1.1,
+  "doSample": true,
   "outputType": "play",
   "cacheGenerated": true,
   "autoInferEnable": false,
@@ -4866,6 +4870,7 @@ Default property:
       | 'BricksDisplay/vits-eng'
       | 'BricksDisplay/vits-cmn'
       | 'BricksDisplay/ellie-Bert-VITS2'
+      | 'onnx-community/OuteTTS-1.0-0.6B-ONNX'
       | 'mms-tts-ara (NC)'
       | 'mms-tts-deu (NC)'
       | 'mms-tts-eng (NC)'
@@ -4881,7 +4886,7 @@ Default property:
       | 'speecht5_tts'
       | DataLink
     /* Model type */
-    modelType?: 'auto' | 'vits' | 'bert_vits2' | 'speecht5' | DataLink
+    modelType?: 'auto' | 'vits' | 'bert_vits2' | 'speecht5' | 'outetts-1.0' | DataLink
     /* Load quantized model (deprecated, use `quantizeType` instead) */
     quantized?: boolean | DataLink
     /* Quantize type */
@@ -4908,6 +4913,16 @@ Default property:
     speakerEmbedUrl?: string | DataLink
     /* MD5 checksum of `speakerEmbedUrl` */
     speakerEmbedMd5?: string | DataLink
+    /* Speaker config, for OuteTTS model */
+    speakerConfig?: {} | DataLink
+    /* Audio token generation max length */
+    maxLength?: number | DataLink
+    /* Audio token generation temperature */
+    temperature?: number | DataLink
+    /* Audio token generation repetition penalty */
+    repetitionPenalty?: number | DataLink
+    /* Use greedy sampling for audio token generation */
+    doSample?: boolean | DataLink
     /* Output mode */
     outputType?: 'play' | 'file' | DataLink
     /* Enable cache for generated audio */
@@ -6385,7 +6400,7 @@ Default property:
     /* Prompt (text mode) */
     completionPrompt?: string | DataLink
     /* Media paths to be used in the prompt template (PREVIEW FEATURE)  
- In prompt, use `<__image__>` for position of media content */
+ In prompt, use `<__media__>` for position of media content */
     completionPromptMediaPaths?: Array<string | DataLink> | DataLink
     /* Data to be used in the prompt template (e.g. `Hello ${name}`). Supports nested data, such as `Hello ${user.name}`. */
     completionPromptTemplateData?: {} | DataLink
@@ -6542,6 +6557,165 @@ export type GeneratorLLM = Generator &
     >
   }
 
+/* Load the model */
+export type GeneratorGGMLTTSActionLoadModel = Action & {
+  __actionName: 'GENERATOR_GGML_TTS_LOAD_MODEL'
+}
+
+/* Generate audio */
+export type GeneratorGGMLTTSActionGenerate = ActionWithParams & {
+  __actionName: 'GENERATOR_GGML_TTS_GENERATE'
+  params?: Array<{
+    input: 'text'
+    value?: string | DataLink | EventProperty
+    mapping?: string
+  }>
+}
+
+/* Clean cache */
+export type GeneratorGGMLTTSActionCleanCache = Action & {
+  __actionName: 'GENERATOR_GGML_TTS_CLEAN_CACHE'
+}
+
+/* Release context */
+export type GeneratorGGMLTTSActionReleaseContext = Action & {
+  __actionName: 'GENERATOR_GGML_TTS_RELEASE_CONTEXT'
+}
+
+interface GeneratorGGMLTTSDef {
+  /*
+Default property:
+{
+  "vocoderUrl": "https://huggingface.co/ggml-org/WavTokenizer/resolve/main/WavTokenizer-Large-75-F16.gguf",
+  "vocoderHashType": "sha256",
+  "vocoderHash": "2356baa8631cc2995ea3465196a017a2733600d849a91180c0f97fa7fb375bbe",
+  "outputType": "play",
+  "cacheGenerated": true,
+  "autoInferEnable": false,
+  "softBreakRegex": "^[^\\r\\n\\t\\f\\v]*([\\r\\n]+|[。！？!?.]\\B)",
+  "hardBreakTime": 500,
+  "completionTemperature": 0.1,
+  "completionRepetitionPenalty": 1.1,
+  "completionTopK": 40,
+  "completionTopP": 0.9,
+  "completionMinP": 0.05,
+  "useGuideToken": false,
+  "contextSize": 4096,
+  "batchSize": 512,
+  "maxThreads": 1,
+  "accelVariant": "default",
+  "mainGpu": 0,
+  "gpuLayers": 0,
+  "useMlock": true,
+  "useMmap": true,
+  "useFlashAttn": false
+}
+  */
+  property?: {
+    /* Initialize the TTS context on generator initialization */
+    init?: boolean | DataLink
+    /* The URL or path of model  
+ We used GGUF format model, please refer to https://github.com/ggerganov/llama.cpp/tree/master#description */
+    modelUrl?: string | DataLink
+    /* Hash type of model */
+    modelHashType?: 'md5' | 'sha256' | 'sha1' | DataLink
+    /* Hash of model */
+    modelHash?: string | DataLink
+    /* The URL or path of vocoder model */
+    vocoderUrl?: string | DataLink
+    /* Hash type of vocoder model */
+    vocoderHashType?: 'md5' | 'sha256' | 'sha1' | DataLink
+    /* Hash of vocoder model */
+    vocoderHash?: string | DataLink
+    /* Output mode */
+    outputType?: 'play' | 'file' | DataLink
+    /* Enable cache for generated audio */
+    cacheGenerated?: boolean | DataLink
+    /* Text to generate */
+    prompt?: string | DataLink
+    /* Speaker JSON */
+    speaker?: {} | DataLink
+    /* Auto inference when prompt changes */
+    autoInferEnable?: boolean | DataLink
+    /* Segmentation rule for auto inference */
+    softBreakRegex?: string | DataLink
+    /* Time to force inference when softBreakRegex is not satisfied */
+    hardBreakTime?: number | DataLink
+    /* Temperature */
+    completionTemperature?: number | DataLink
+    /* Repetition Penalty */
+    completionRepetitionPenalty?: number | DataLink
+    /* Top K sampling */
+    completionTopK?: number | DataLink
+    /* Top P sampling */
+    completionTopP?: number | DataLink
+    /* Min P sampling */
+    completionMinP?: number | DataLink
+    /* Set the random number generator (RNG) seed (default: -1, -1 = random seed) */
+    completionSeed?: number | DataLink
+    /* Enable guide token to help prevent hallucinations by forcing the TTS to use the correct words. */
+    useGuideToken?: boolean | DataLink
+    /* Context size, for OutTTS recommended 4096 ~ 8192 (Default to 4096) */
+    contextSize?: number | DataLink
+    /* Logical batch size for prompt processing */
+    batchSize?: number | DataLink
+    /* Number of threads */
+    maxThreads?: number | DataLink
+    /* Accelerator variant (Only for desktop)  
+ `default` - CPU / Metal (macOS)  
+ `vulkan` - Use Vulkan  
+ `cuda` - Use CUDA */
+    accelVariant?: 'default' | 'vulkan' | 'cuda' | DataLink
+    /* Main GPU index */
+    mainGpu?: number | DataLink
+    /* Number of GPU layers (NOTE: Currently not supported for Android) */
+    gpuLayers?: number | DataLink
+    /* Use memory lock */
+    useMlock?: boolean | DataLink
+    /* Use mmap */
+    useMmap?: boolean | DataLink
+    /* Use Flash Attention for inference (Recommended with GPU enabled) */
+    useFlashAttn?: boolean | DataLink
+  }
+  events?: {
+    /* Event triggered when state change */
+    onContextStateChange?: Array<EventAction>
+    /* Event triggered when error occurs */
+    onError?: Array<EventAction>
+  }
+  outlets?: {
+    /* Context state */
+    contextState?: () => Data
+    /* Generated audio file */
+    generatedAudio?: () => Data
+    /* Generated audio file is playing */
+    generatedAudioPlaying?: () => Data
+  }
+}
+
+/* Local Text-to-Speech (TTS) inference based on GGML and [llama.cpp](https://github.com/ggerganov/llama.cpp)  
+ You can use any converted model on HuggingFace. */
+export type GeneratorGGMLTTS = Generator &
+  GeneratorGGMLTTSDef & {
+    templateKey: 'GENERATOR_GGML_TTS'
+    switches: Array<
+      SwitchDef &
+        GeneratorGGMLTTSDef & {
+          conds?: Array<{
+            method: '==' | '!=' | '>' | '<' | '>=' | '<='
+            cond:
+              | SwitchCondInnerStateCurrentCanvas
+              | SwitchCondData
+              | {
+                  __typename: 'SwitchCondInnerStateOutlet'
+                  outlet: 'contextState' | 'generatedAudio' | 'generatedAudioPlaying'
+                  value: any
+                }
+          }>
+        }
+    >
+  }
+
 /* Load the model */
 export type GeneratorQnnLlmActionLoadModel = Action & {
   __actionName: 'GENERATOR_QNN_LLM_LOAD_MODEL'

package/utils/event-props.ts

@@ -806,6 +806,14 @@ export const templateEventPropsMap = {
       'GENERATOR_LLM_COMPLETION_FUNCTION_DETAILS', // type: object
     ],
   },
+  GENERATOR_GGML_TTS: {
+    onContextStateChange: [
+      'GENERATOR_GGML_TTS_CONTEXT_STATE', // type: string
+    ],
+    onError: [
+      'GENERATOR_GGML_TTS_ERROR', // type: string
+    ],
+  },
   GENERATOR_QNN_LLM: {
     onContextStateChange: [
       'GENERATOR_QNN_LLM_CONTEXT_STATE', // type: string