@nodaro/sdk 1.3.0 → 1.5.0

package/README.md

@@ -270,7 +270,7 @@ try {
 | `client.locations` | Location Studio: CRUD, assets, atmosphere motion |
 | `client.objects` | Object Studio: CRUD, assets, motion |
 | `client.creatures` | Creature Studio: CRUD, assets, motion |
-| `client.voices` | Voice design, clone, remix, recast |
+| `client.voices` | Voice catalog + clones (from URL or file), voice changer, multi-speaker recast + the interactive `analyze`/stems/`exportMix` flow, design, remix, dubbing |
 | `client.pipelines` | Showrunner pipelines: stages, approvals, chat, branch |
 | `client.reduce` | Fan-in reducer (pick-best, concat, vote, merge…) |
 | `client.promptHelper` | Prompt enhancement / wizard |

package/dist/index.cjs

@@ -1551,6 +1551,185 @@ var VoicesResource = class {
   recast(input) {
     return this.client.request("POST", "/v1/voice-changer-pro", { body: input });
   }
+  /**
+   * Detect the speakers in a clip WITHOUT recasting yet
+   * (`POST /v1/voice-changer-pro/analyze`) — the first step of the interactive
+   * flow. Separates voice from music once and diarizes the vocals, returning the
+   * speaker list so a user (or agent) can choose a voice per speaker before
+   * committing to a paid recast. Poll `jobs.get(jobId)`: the completed job's
+   * `output_data` carries the separated stem urls + the detected `speakers`
+   * (each with `id`, time `segments`, `firstStartSec`, `wordCount`, `snippet`)
+   * and the detected language — reshape it into a {@link VcpAnalysis} and pass it
+   * as `recast({ ..., analysis })` to skip re-detection. With `suggestTitle`,
+   * `output_data.suggestedTitle` also carries an LLM-proposed title.
+   *
+   * Cloud-only; costs credits and runs async.
+   */
+  analyze(input) {
+    return this.client.request("POST", "/v1/voice-changer-pro/analyze", { body: input });
+  }
+  /**
+   * Render a final video from a mixed set of stems
+   * (`POST /v1/voice-changer-pro/export`) — the last step of the interactive
+   * flow. After `recast({ output: "stems" })` hands back the dry per-track stems
+   * and the user has set levels / mutes / an effect in your editor, pass those
+   * `tracks` (plus the source `videoUrl`) here to mix and remux into the finished
+   * video. The video is stream-copied (never re-encoded), so the export is
+   * bit-identical to your preview. At least one track must be un-muted (all-muted
+   * is a 400); `voiceFx` is applied to the voice tracks at render time.
+   *
+   * Cloud-only; costs credits and runs async — poll `jobs.get(jobId)` for the
+   * result (`output_data.videoUrl`).
+   */
+  exportMix(input) {
+    return this.client.request("POST", "/v1/voice-changer-pro/export", { body: input });
+  }
+  /**
+   * Design a brand-new synthetic voice from a text description
+   * (`POST /v1/voice-design`) — ElevenLabs text-to-voice. `text` (100–1000 chars)
+   * is a preview line spoken in the designed voice; `voiceDescription` describes
+   * the voice to create. Costs credits and runs async — poll `jobs.get(jobId)`
+   * for the preview + the reusable voice id.
+   */
+  design(input) {
+    return this.client.request("POST", "/v1/voice-design", { body: input });
+  }
+  /**
+   * Generate speech in a voice described in natural language, without cloning
+   * (`POST /v1/voice-remix`). `text` (1–5000 chars) is spoken in a voice matching
+   * `voiceDescription`. Costs credits and runs async — poll `jobs.get(jobId)`.
+   */
+  remix(input) {
+    return this.client.request("POST", "/v1/voice-remix", { body: input });
+  }
+  /**
+   * Dub an audio clip into another language while preserving each speaker's voice
+   * (`POST /v1/dubbing`). `targetLanguage` is an ISO code (e.g. `"es"`, `"fr"`);
+   * `sourceLanguage` is auto-detected when omitted. Costs credits and runs async
+   * — poll `jobs.get(jobId)`.
+   */
+  dub(input) {
+    return this.client.request("POST", "/v1/dubbing", { body: input });
+  }
+  /**
+   * Clone a voice from an audio FILE you hold in memory
+   * (`POST /v1/voice-clones`, multipart) — the counterpart to
+   * {@link VoicesResource.createClone}, which clones from an already-uploaded
+   * URL. Pass the raw audio `file` (a `Blob`/`File` in the browser, or a
+   * `Uint8Array`/`Buffer` in Node) plus a `name`. Costs credits. Returns the new
+   * {@link VoiceClone} (`elevenlabsVoiceId` is the id to recast/synthesize with).
+   */
+  createCloneFromFile(input) {
+    const form = new FormData();
+    form.append("name", input.name);
+    const blob = input.file instanceof Blob ? input.file : new Blob([input.file], { type: input.contentType ?? "audio/mpeg" });
+    form.append("file", blob, input.filename ?? "sample");
+    return this.client.request("POST", "/v1/voice-clones", { body: form });
+  }
+};
+
+// src/resources/media.ts
+var MediaResource = class {
+  constructor(client) {
+    this.client = client;
+  }
+  client;
+  /**
+   * Download a social video (YouTube / TikTok / Instagram / X / Facebook) into
+   * your storage (`POST /v1/download-video`). `maxHeight` caps the resolution
+   * (default "best"); `sectionStartSec` + `sectionEndSec` (both-or-neither) fetch
+   * ONLY that time range instead of the whole video. Returns a `downloadId`;
+   * progress streams from `GET /v1/download-video/progress/:downloadId`
+   * (server-sent events) and the finished file lands in your library.
+   */
+  downloadVideo(input) {
+    return this.client.request("POST", "/v1/download-video", { body: input });
+  }
+  /**
+   * Copy an external media URL into your Nodaro storage (`POST /v1/save-to-storage`)
+   * — a server-side fetch, so nothing round-trips through the client. Poll
+   * `jobs.get(jobId)`.
+   */
+  saveToStorage(input) {
+    return this.client.request("POST", "/v1/save-to-storage", { body: input });
+  }
+  /**
+   * Trim a video to a range (`POST /v1/trim-video`). Give the range in whichever
+   * unit fits: `startTime`/`endTime` seconds, `trim*Frames`, `trim*Seconds`, or
+   * `keepFirst`/`keepLastSeconds`. Poll `jobs.get(jobId)`.
+   */
+  trimVideo(input) {
+    return this.client.request("POST", "/v1/trim-video", { body: input });
+  }
+  /**
+   * Trim (and extract) audio from a video or audio source
+   * (`POST /v1/trim-audio`) to `[startTime, endTime]` seconds, in `audioFormat`
+   * (`mp3` default / `wav` / `aac`). Poll `jobs.get(jobId)`.
+   */
+  trimAudio(input) {
+    return this.client.request("POST", "/v1/trim-audio", { body: input });
+  }
+  /**
+   * Probe a social video's metadata (`POST /v1/video-metadata`) — duration,
+   * dimensions, title, live status — WITHOUT downloading it. A direct read, not a
+   * job. Use it to decide whether to trim before importing.
+   */
+  videoMetadata(input) {
+    return this.client.request("POST", "/v1/video-metadata", { body: input });
+  }
+};
+
+// src/resources/audio.ts
+var AudioResource = class {
+  constructor(client) {
+    this.client = client;
+  }
+  client;
+  /**
+   * Separate an audio track into stems (`POST /v1/audio-separation`, Demucs).
+   * `mode` `"vocal_instrumental"` (default) splits voice from music/SFX;
+   * `"stems"` returns the full drums/bass/other/… breakdown. `quality`
+   * `auto` (default) / `fast` / `best`.
+   */
+  separate(input) {
+    return this.client.request("POST", "/v1/audio-separation", { body: input });
+  }
+  /** Isolate the primary voice and strip background noise (`POST /v1/audio-isolation`, ElevenLabs). */
+  isolate(input) {
+    return this.client.request("POST", "/v1/audio-isolation", { body: input });
+  }
+  /**
+   * Apply a reverb / echo / telephone / megaphone effect to an audio track
+   * (`POST /v1/audio-fx`) — the same presets VCP's `voiceFx` uses, standalone.
+   * `mix` (0–100) is the reverb wet/dry; `delayMs` + `decay` drive `echo`/`custom`;
+   * `eqLow`/`eqHigh` (dB) shape telephone/megaphone.
+   */
+  applyFx(input) {
+    return this.client.request("POST", "/v1/audio-fx", { body: input });
+  }
+  /**
+   * Layer multiple audio tracks into one (`POST /v1/mix-audio`). `audioUrls`
+   * (2–20) are summed; optional `trackVolumes` (0–200% each, positionally) set
+   * per-track level.
+   */
+  mix(input) {
+    return this.client.request("POST", "/v1/mix-audio", { body: input });
+  }
+  /**
+   * Adjust an audio (or a video's audio) level (`POST /v1/adjust-volume`):
+   * `volume` % (default 100), `normalize` to loudnorm, and `fadeIn`/`fadeOut`
+   * seconds. Provide `audioUrl` or `videoUrl`.
+   */
+  adjustVolume(input) {
+    return this.client.request("POST", "/v1/adjust-volume", { body: input });
+  }
+  /**
+   * Concatenate audio segments end-to-end (`POST /v1/combine-audio`). Each
+   * segment is a `url` with an optional `[startTime, endTime]` sub-range.
+   */
+  combine(input) {
+    return this.client.request("POST", "/v1/combine-audio", { body: input });
+  }
 };
 
 // src/resources/credits.ts
@@ -1853,6 +2032,8 @@ var NodaroClient = class {
   reduce;
   promptHelper;
   voices;
+  media;
+  audio;
   credits;
   uploads;
   library;
@@ -1880,6 +2061,8 @@ var NodaroClient = class {
     this.reduce = new ReduceResource(this);
     this.promptHelper = new PromptHelperResource(this);
     this.voices = new VoicesResource(this);
+    this.media = new MediaResource(this);
+    this.audio = new AudioResource(this);
     this.credits = new CreditsResource(this);
     this.uploads = new UploadsResource(this);
     this.library = new LibraryResource(this);
@@ -2046,6 +2229,7 @@ Object.defineProperty(exports, "SURROUND_DIRECTIONS", {
   get: function () { return shared.SURROUND_DIRECTIONS; }
 });
 exports.AppsResource = AppsResource;
+exports.AudioResource = AudioResource;
 exports.CREATURE_ASSET_TYPES = CREATURE_ASSET_TYPES;
 exports.CallbackAuth = CallbackAuth;
 exports.CharactersResource = CharactersResource;
@@ -2062,6 +2246,7 @@ exports.JobTimeoutError = JobTimeoutError;
 exports.JobsResource = JobsResource;
 exports.LibraryResource = LibraryResource;
 exports.LocationsResource = LocationsResource;
+exports.MediaResource = MediaResource;
 exports.NodaroClient = NodaroClient;
 exports.NodaroError = NodaroError;
 exports.NodesResource = NodesResource;