@nodaro/sdk 1.23.0 → 1.25.0

package/dist/index.d.cts

@@ -222,10 +222,17 @@ declare class WorkflowsResource {
     }>;
     /**
      * Import a `WorkflowExport` bundle into the specified project.
-     * Re-creates any bundled assets (characters, objects, locations) under your account.
+     * Re-creates any bundled assets (characters, objects, creatures, locations)
+     * under your account, and re-points BOTH the entity nodes and every `@`-chip
+     * (`ConnectedReference`) bound in the graph at the rows it created.
      * Media the bundle references on other hosts is copied onto this instance's
-     * storage where reachable; `importReport` says what was copied, what could
-     * not be reached, and what was skipped (and why).
+     * storage where reachable; a bundled entity's images are copied whoever
+     * hosts them, because they are the exporter's bytes and their lifecycle is
+     * not yours. `importReport` says what was copied, what could not be reached
+     * and what was skipped (and why); `assetIdMap` maps each bundled entity id
+     * to the row created for it (for chips you hold outside the graph), and
+     * `assetsSkipped` names the entities your storage quota left uncreated —
+     * the workflow itself still lands.
      */
     import(input: WorkflowExport & {
         projectId: string;
@@ -3769,6 +3776,32 @@ declare class VoicesResource {
     dub(input: DubbingInput): Promise<{
         jobId: string;
     }>;
+    /**
+     * Voice a multi-speaker script as ONE audio file
+     * (`POST /v1/text-to-dialogue`, ElevenLabs Dialogue v3). Each `dialogue`
+     * line is `{ text, voice }` in speaking order — `voice` is a premade voice
+     * name or an ElevenLabs voice UUID (cloned/library voices work too; mixed
+     * casts are fine). At most 5,000 characters total across lines (≤2,000
+     * recommended for best quality) and 10 unique voices. Line text may carry
+     * `[audio tags]` like `[laughs]`. Costs credits and runs async — poll
+     * `jobs.get(jobId)` for `output_data.audioUrl`.
+     */
+    textToDialogue(input: {
+        dialogue: Array<{
+            text: string;
+            voice: string;
+        }>;
+        /** v3 stability: 0 (most variable) | 0.5 (balanced) | 1 (most stable). */
+        stability?: 0 | 0.5 | 1;
+        /** ISO 639-1 language hint (e.g. "en"); auto-detected when omitted. */
+        languageCode?: string;
+        /** Deterministic sampling seed (integer 0–4294967295). Omit for random. */
+        seed?: number;
+        /** Spell out numbers/dates/abbreviations: "auto" (default) | "on" | "off". */
+        applyTextNormalization?: "auto" | "on" | "off";
+    }): Promise<{
+        jobId: string;
+    }>;
     /**
      * Clone a voice from an audio FILE you hold in memory
      * (`POST /v1/voice-clones`, multipart) — the counterpart to
@@ -3795,6 +3828,18 @@ declare class VoicesResource {
 type VoiceChangerProVoice = string | {
     /** Target voice — premade name (`"Rachel"`, `"Aria"`, …) or an ElevenLabs UUID for a custom clone. */
     voiceId: string;
+    /**
+     * Which lane converts this speaker. `"sts"` (default) is the classic
+     * speech-to-speech recast; `"v3"` is RE-SPEAK — the performance is
+     * REGENERATED from the transcript with eleven_v3 (supports `[audio
+     * tags]`; original delivery is replaced, and lips won't match on video).
+     * A v3 speaker needs transcript text: pass an `analysis` whose
+     * `segments[].text` carries it (analyze now emits this), or omit
+     * `analysis` and the engine re-speaks from its own transcription. For
+     * `"v3"`, `stability` accepts exactly 0 / 0.5 / 1, and
+     * `similarityBoost` / `style` / `useSpeakerBoost` are ignored.
+     */
+    engine?: "sts" | "v3";
     /** ElevenLabs stability (0–1). Higher = steadier, lower = more expressive. */
     stability?: number;
     /** ElevenLabs similarity boost (0–1) — how closely the output hugs the target voice's timbre. */
@@ -3897,10 +3942,13 @@ interface VoiceChangerProInput {
 interface VcpAnalysisSpeaker {
     /** Stable speaker id (first-appearance order). */
     id: string;
-    /** The speaker's spoken time ranges (seconds). */
+    /** The speaker's spoken time ranges (seconds). `text` is what was said in
+     *  the range — the paid input for a speaker recast with `engine: "v3"`
+     *  (editable before conversion); the STS lane ignores it. */
     segments: Array<{
         start: number;
         end: number;
+        text?: string;
     }>;
     /** When the speaker first speaks (seconds). */
     firstStartSec?: number;
@@ -3996,20 +4044,39 @@ interface VoiceRemixInput {
     /** Optional extra prompt context (≤8000 chars). */
     userPrompt?: string;
 }
-/** Input for {@link VoicesResource.dub}. */
+/** Input for {@link VoicesResource.dub}. Provide exactly ONE source:
+ *  `audioUrl` (dubbed audio out), `videoUrl` (the dubbed VIDEO out, plus the
+ *  dubbed audio track as `output_data.audioUrl`), or `sourceUrl` (a public
+ *  YouTube/TikTok/direct link ElevenLabs fetches itself). Priced per minute
+ *  of the dubbed span; the span is capped at 30 minutes — use
+ *  `startTime`/`endTime` to dub part of a longer source. */
 interface DubbingInput {
-    /** URL of the audio to dub. */
-    audioUrl: string;
+    /** URL of the audio to dub (audio mode). */
+    audioUrl?: string;
+    /** URL of the video to dub — the result is the dubbed VIDEO (+ audio track). */
+    videoUrl?: string;
+    /** Public page/media URL (YouTube, TikTok, or a direct link) — ElevenLabs fetches it directly. */
+    sourceUrl?: string;
     /** Target language ISO code (2–10 chars), e.g. `"es"`, `"pt-BR"`. */
     targetLanguage: string;
     /** Source language ISO code; auto-detected when omitted. */
     sourceLanguage?: string;
-    /** Expected number of speakers (1–20) — improves separation when known. */
+    /** Expected number of speakers (1–20; 0 = auto-detect) — improves separation when known. */
     numSpeakers?: number;
     /** Keep the original voices instead of cloning them into the target language. */
     disableVoiceCloning?: boolean;
     /** Drop the background/music bed from the dubbed output. */
     dropBackgroundAudio?: boolean;
+    /** Dub only this window of the source (seconds). */
+    startTime?: number;
+    endTime?: number;
+    /** Keep the source resolution on video dubs (slower render). */
+    highestResolution?: boolean;
+    useProfanityFilter?: boolean;
+    /** Experimental: steer dubbed voices toward an accent. */
+    targetAccent?: string;
+    /** Apply ElevenLabs' own watermark to video dubs. */
+    watermark?: boolean;
 }
 
 /**

package/dist/index.d.ts

@@ -222,10 +222,17 @@ declare class WorkflowsResource {
     }>;
     /**
      * Import a `WorkflowExport` bundle into the specified project.
-     * Re-creates any bundled assets (characters, objects, locations) under your account.
+     * Re-creates any bundled assets (characters, objects, creatures, locations)
+     * under your account, and re-points BOTH the entity nodes and every `@`-chip
+     * (`ConnectedReference`) bound in the graph at the rows it created.
      * Media the bundle references on other hosts is copied onto this instance's
-     * storage where reachable; `importReport` says what was copied, what could
-     * not be reached, and what was skipped (and why).
+     * storage where reachable; a bundled entity's images are copied whoever
+     * hosts them, because they are the exporter's bytes and their lifecycle is
+     * not yours. `importReport` says what was copied, what could not be reached
+     * and what was skipped (and why); `assetIdMap` maps each bundled entity id
+     * to the row created for it (for chips you hold outside the graph), and
+     * `assetsSkipped` names the entities your storage quota left uncreated —
+     * the workflow itself still lands.
      */
     import(input: WorkflowExport & {
         projectId: string;
@@ -3769,6 +3776,32 @@ declare class VoicesResource {
     dub(input: DubbingInput): Promise<{
         jobId: string;
     }>;
+    /**
+     * Voice a multi-speaker script as ONE audio file
+     * (`POST /v1/text-to-dialogue`, ElevenLabs Dialogue v3). Each `dialogue`
+     * line is `{ text, voice }` in speaking order — `voice` is a premade voice
+     * name or an ElevenLabs voice UUID (cloned/library voices work too; mixed
+     * casts are fine). At most 5,000 characters total across lines (≤2,000
+     * recommended for best quality) and 10 unique voices. Line text may carry
+     * `[audio tags]` like `[laughs]`. Costs credits and runs async — poll
+     * `jobs.get(jobId)` for `output_data.audioUrl`.
+     */
+    textToDialogue(input: {
+        dialogue: Array<{
+            text: string;
+            voice: string;
+        }>;
+        /** v3 stability: 0 (most variable) | 0.5 (balanced) | 1 (most stable). */
+        stability?: 0 | 0.5 | 1;
+        /** ISO 639-1 language hint (e.g. "en"); auto-detected when omitted. */
+        languageCode?: string;
+        /** Deterministic sampling seed (integer 0–4294967295). Omit for random. */
+        seed?: number;
+        /** Spell out numbers/dates/abbreviations: "auto" (default) | "on" | "off". */
+        applyTextNormalization?: "auto" | "on" | "off";
+    }): Promise<{
+        jobId: string;
+    }>;
     /**
      * Clone a voice from an audio FILE you hold in memory
      * (`POST /v1/voice-clones`, multipart) — the counterpart to
@@ -3795,6 +3828,18 @@ declare class VoicesResource {
 type VoiceChangerProVoice = string | {
     /** Target voice — premade name (`"Rachel"`, `"Aria"`, …) or an ElevenLabs UUID for a custom clone. */
     voiceId: string;
+    /**
+     * Which lane converts this speaker. `"sts"` (default) is the classic
+     * speech-to-speech recast; `"v3"` is RE-SPEAK — the performance is
+     * REGENERATED from the transcript with eleven_v3 (supports `[audio
+     * tags]`; original delivery is replaced, and lips won't match on video).
+     * A v3 speaker needs transcript text: pass an `analysis` whose
+     * `segments[].text` carries it (analyze now emits this), or omit
+     * `analysis` and the engine re-speaks from its own transcription. For
+     * `"v3"`, `stability` accepts exactly 0 / 0.5 / 1, and
+     * `similarityBoost` / `style` / `useSpeakerBoost` are ignored.
+     */
+    engine?: "sts" | "v3";
     /** ElevenLabs stability (0–1). Higher = steadier, lower = more expressive. */
     stability?: number;
     /** ElevenLabs similarity boost (0–1) — how closely the output hugs the target voice's timbre. */
@@ -3897,10 +3942,13 @@ interface VoiceChangerProInput {
 interface VcpAnalysisSpeaker {
     /** Stable speaker id (first-appearance order). */
     id: string;
-    /** The speaker's spoken time ranges (seconds). */
+    /** The speaker's spoken time ranges (seconds). `text` is what was said in
+     *  the range — the paid input for a speaker recast with `engine: "v3"`
+     *  (editable before conversion); the STS lane ignores it. */
     segments: Array<{
         start: number;
         end: number;
+        text?: string;
     }>;
     /** When the speaker first speaks (seconds). */
     firstStartSec?: number;
@@ -3996,20 +4044,39 @@ interface VoiceRemixInput {
     /** Optional extra prompt context (≤8000 chars). */
     userPrompt?: string;
 }
-/** Input for {@link VoicesResource.dub}. */
+/** Input for {@link VoicesResource.dub}. Provide exactly ONE source:
+ *  `audioUrl` (dubbed audio out), `videoUrl` (the dubbed VIDEO out, plus the
+ *  dubbed audio track as `output_data.audioUrl`), or `sourceUrl` (a public
+ *  YouTube/TikTok/direct link ElevenLabs fetches itself). Priced per minute
+ *  of the dubbed span; the span is capped at 30 minutes — use
+ *  `startTime`/`endTime` to dub part of a longer source. */
 interface DubbingInput {
-    /** URL of the audio to dub. */
-    audioUrl: string;
+    /** URL of the audio to dub (audio mode). */
+    audioUrl?: string;
+    /** URL of the video to dub — the result is the dubbed VIDEO (+ audio track). */
+    videoUrl?: string;
+    /** Public page/media URL (YouTube, TikTok, or a direct link) — ElevenLabs fetches it directly. */
+    sourceUrl?: string;
     /** Target language ISO code (2–10 chars), e.g. `"es"`, `"pt-BR"`. */
     targetLanguage: string;
     /** Source language ISO code; auto-detected when omitted. */
     sourceLanguage?: string;
-    /** Expected number of speakers (1–20) — improves separation when known. */
+    /** Expected number of speakers (1–20; 0 = auto-detect) — improves separation when known. */
     numSpeakers?: number;
     /** Keep the original voices instead of cloning them into the target language. */
     disableVoiceCloning?: boolean;
     /** Drop the background/music bed from the dubbed output. */
     dropBackgroundAudio?: boolean;
+    /** Dub only this window of the source (seconds). */
+    startTime?: number;
+    endTime?: number;
+    /** Keep the source resolution on video dubs (slower render). */
+    highestResolution?: boolean;
+    useProfanityFilter?: boolean;
+    /** Experimental: steer dubbed voices toward an accent. */
+    targetAccent?: string;
+    /** Apply ElevenLabs' own watermark to video dubs. */
+    watermark?: boolean;
 }
 
 /**

package/dist/index.js

@@ -238,10 +238,17 @@ var WorkflowsResource = class {
   }
   /**
    * Import a `WorkflowExport` bundle into the specified project.
-   * Re-creates any bundled assets (characters, objects, locations) under your account.
+   * Re-creates any bundled assets (characters, objects, creatures, locations)
+   * under your account, and re-points BOTH the entity nodes and every `@`-chip
+   * (`ConnectedReference`) bound in the graph at the rows it created.
    * Media the bundle references on other hosts is copied onto this instance's
-   * storage where reachable; `importReport` says what was copied, what could
-   * not be reached, and what was skipped (and why).
+   * storage where reachable; a bundled entity's images are copied whoever
+   * hosts them, because they are the exporter's bytes and their lifecycle is
+   * not yours. `importReport` says what was copied, what could not be reached
+   * and what was skipped (and why); `assetIdMap` maps each bundled entity id
+   * to the row created for it (for chips you hold outside the graph), and
+   * `assetsSkipped` names the entities your storage quota left uncreated —
+   * the workflow itself still lands.
    */
   import(input) {
     const { projectId, ...workflowJson } = input;
@@ -1756,6 +1763,19 @@ var VoicesResource = class {
   dub(input) {
     return this.client.request("POST", "/v1/dubbing", { body: input });
   }
+  /**
+   * Voice a multi-speaker script as ONE audio file
+   * (`POST /v1/text-to-dialogue`, ElevenLabs Dialogue v3). Each `dialogue`
+   * line is `{ text, voice }` in speaking order — `voice` is a premade voice
+   * name or an ElevenLabs voice UUID (cloned/library voices work too; mixed
+   * casts are fine). At most 5,000 characters total across lines (≤2,000
+   * recommended for best quality) and 10 unique voices. Line text may carry
+   * `[audio tags]` like `[laughs]`. Costs credits and runs async — poll
+   * `jobs.get(jobId)` for `output_data.audioUrl`.
+   */
+  textToDialogue(input) {
+    return this.client.request("POST", "/v1/text-to-dialogue", { body: input });
+  }
   /**
    * Clone a voice from an audio FILE you hold in memory
    * (`POST /v1/voice-clones`, multipart) — the counterpart to
@@ -2615,7 +2635,7 @@ var WorkspacesResource = class {
     return this.client.request("POST", "/v1/workspaces/join", { body: { code } });
   }
 };
-var SDK_VERSION = "1.23.0" ;
+var SDK_VERSION = "1.25.0" ;
 var CLIENT_HEADER = "X-Nodaro-Client";
 var isBrowser = () => typeof window !== "undefined" && typeof window.document !== "undefined";
 var NodaroClient = class _NodaroClient {