sonilo 0.12.0 → 0.14.0

package/dist/index.d.cts

@@ -110,6 +110,14 @@ interface VideoToMusicParams {
      * by `submit()`) — only meaningful via `submit()`; `stream()`/`generate()`
      * never send it, since they always request a plain stream. */
     variantsNum?: number;
+    /** How strongly the generated music follows the prompt (0-1, API default
+     * 0.5). Lower values let the video lead; higher values follow the prompt
+     * more literally. Free of charge. A generation-time knob, not a
+     * finalize-time one, so it works on the plain stream as well as async —
+     * no `mode` gating. Omitted from the wire when unset so the server
+     * default applies; `0` is a meaningful value and is sent. Out-of-range
+     * values are rejected server-side with a 422. */
+    promptInfluence?: number;
 }
 /** One service's free-trial allowance. `remaining` is already floored at 0,
  * so it is safe to compare directly. */
@@ -329,6 +337,12 @@ interface VideoToVideoMusicParams {
      * by the free trial. This endpoint is always async, so no extra `mode`
      * gating applies. The result's `videos[]` gets one entry per variant. */
     variantsNum?: number;
+    /** How strongly the generated music follows the prompt (0-1, API default
+     * 0.5). Lower values let the video lead; higher values follow the prompt
+     * more literally. Free of charge. Omitted from the wire when unset so the
+     * server default applies; `0` is a meaningful value and is sent.
+     * Out-of-range values are rejected server-side with a 422. */
+    promptInfluence?: number;
 }
 interface VideoToVideoSfxParams {
     video?: VideoInput;
@@ -438,6 +452,34 @@ interface SoundResult extends BaseTaskResult {
     /** One entry per variant; see `SoundOutputEntry`. */
     outputs?: SoundOutputEntry[];
 }
+/** Params for `audioDucking`: mix an existing music bed under an existing
+ * voice track, ducking the music wherever the voice speaks. Exactly one of
+ * `voice`/`voiceUrl` and exactly one of `music`/`musicUrl`; a local input and
+ * a URL may be mixed across the two.
+ *
+ * The voice may be audio or video (`VideoInput` is the SDK's generic media
+ * input union) — a video's own audio track becomes the voice, and the ducked
+ * mix is re-muxed back into a new video. The music must be audio: the backend
+ * never probes it for a video stream, so a video there would be silently
+ * mishandled. Each input is capped at 360 seconds server-side. */
+interface AudioDuckingParams {
+    voice?: VideoInput;
+    voiceUrl?: string;
+    music?: VideoInput;
+    musicUrl?: string;
+}
+/** Result of an `audioDucking` task (`tasks.get`) or its final state
+ * (`generate`). Same flat envelope as `SoundResult`, but a ducking task
+ * renders exactly one artifact and no stems: a `.wav` (`output_type`
+ * "audio"), or a `.mp4` with the ducked mix re-muxed in (`output_type`
+ * "video") when the voice input was a video. Pass `output_url` to
+ * `download()`. */
+interface DuckingResult extends BaseTaskResult {
+    output_url?: string;
+    output_type?: "audio" | "video";
+    output_bytes?: number;
+    duration_seconds?: number;
+}
 /**
  * A target language for /v1/dubbing. The union stays open (`string & {}`) so a
  * language added server-side still type-checks against an older SDK — the
@@ -582,6 +624,21 @@ declare class VideoToVideoSound {
     generate(params: VideoToVideoSoundParams, opts?: WaitOptions): Promise<SoundResult>;
 }
 
+/** Duck an existing music bed under a voice track. Async only (202 + poll).
+ *
+ * Both inputs are user-supplied — nothing is generated here. The voice may be
+ * audio OR a video: the backend extracts a video's audio track, ducks the
+ * music under it, and re-muxes the ducked mix back into a new video (the
+ * result's `output_type` announces which came back). The music must be audio —
+ * the backend never probes it for a video stream, so a video there would be
+ * silently mishandled. */
+declare class AudioDucking {
+    private readonly client;
+    constructor(client: SoniloClient);
+    submit(params: AudioDuckingParams): Promise<SfxTask>;
+    generate(params: AudioDuckingParams, opts?: WaitOptions): Promise<DuckingResult>;
+}
+
 /** Dub a video into one or more target languages. Async only; the result
  * carries a language → dubbed-video-URL map under `outputs`. */
 declare class Dubbing {
@@ -628,6 +685,7 @@ declare class SoniloClient {
     readonly videoToVideoSfx: VideoToVideoSfx;
     readonly videoToSound: VideoToSound;
     readonly videoToVideoSound: VideoToVideoSound;
+    readonly audioDucking: AudioDucking;
     readonly dubbing: Dubbing;
     constructor(options?: SoniloClientOptions);
     /**
@@ -710,6 +768,6 @@ declare class RequestTimeoutError extends SoniloError {
 declare function download(media: SfxMedia | string | undefined, fetchFn?: typeof globalThis.fetch, timeout?: number): Promise<Uint8Array>;
 
 /** The SDK's own version. Generated by scripts/sync-versions.mjs — do not edit. */
-declare const VERSION = "0.12.0";
+declare const VERSION = "0.14.0";
 
-export { APIError, type AccountServices, type AudioChunkEvent, AuthenticationError, BadRequestError, type BaseTaskResult, type CompleteEvent, type CostEvent, type CostInfo, DEFAULT_TIMEOUT_MS, type DailyUsage, type DubbingLanguage, type DubbingParams, type DubbingResult, type ErrorEvent, GenerationError, type MusicMediaEntry, type MusicMuxEntry, type MusicTaskResult, type MusicTitle, PaymentRequiredError, RateLimitError, RequestTimeoutError, type Segment, type SegmentLabel, type SfxAudioFormat, type SfxError, type SfxMedia, type SfxResult, type SfxSegment, type SfxTask, SoniloClient, type SoniloClientOptions, SoniloError, type SoundOutputEntry, type SoundResult, type StreamEvent, TaskFailedError, TaskTimeoutError, type TextToMusicParams, type TextToSfxParams, type TitleEvent, type Track, TrialExhaustedError, type TrialQuota, type UnknownEvent, type UsageResponse, type UsageSummary, VERSION, type VideoInput, type VideoResult, type VideoToMusicParams, type VideoToSfxParams, type VideoToSoundParams, type VideoToVideoMusicParams, type VideoToVideoSfxParams, type VideoToVideoSoundParams, type WaitOptions, download, isAudioChunkEvent, isErrorEvent };
+export { APIError, type AccountServices, type AudioChunkEvent, type AudioDuckingParams, AuthenticationError, BadRequestError, type BaseTaskResult, type CompleteEvent, type CostEvent, type CostInfo, DEFAULT_TIMEOUT_MS, type DailyUsage, type DubbingLanguage, type DubbingParams, type DubbingResult, type DuckingResult, type ErrorEvent, GenerationError, type MusicMediaEntry, type MusicMuxEntry, type MusicTaskResult, type MusicTitle, PaymentRequiredError, RateLimitError, RequestTimeoutError, type Segment, type SegmentLabel, type SfxAudioFormat, type SfxError, type SfxMedia, type SfxResult, type SfxSegment, type SfxTask, SoniloClient, type SoniloClientOptions, SoniloError, type SoundOutputEntry, type SoundResult, type StreamEvent, TaskFailedError, TaskTimeoutError, type TextToMusicParams, type TextToSfxParams, type TitleEvent, type Track, TrialExhaustedError, type TrialQuota, type UnknownEvent, type UsageResponse, type UsageSummary, VERSION, type VideoInput, type VideoResult, type VideoToMusicParams, type VideoToSfxParams, type VideoToSoundParams, type VideoToVideoMusicParams, type VideoToVideoSfxParams, type VideoToVideoSoundParams, type WaitOptions, download, isAudioChunkEvent, isErrorEvent };

package/dist/index.d.ts

@@ -110,6 +110,14 @@ interface VideoToMusicParams {
      * by `submit()`) — only meaningful via `submit()`; `stream()`/`generate()`
      * never send it, since they always request a plain stream. */
     variantsNum?: number;
+    /** How strongly the generated music follows the prompt (0-1, API default
+     * 0.5). Lower values let the video lead; higher values follow the prompt
+     * more literally. Free of charge. A generation-time knob, not a
+     * finalize-time one, so it works on the plain stream as well as async —
+     * no `mode` gating. Omitted from the wire when unset so the server
+     * default applies; `0` is a meaningful value and is sent. Out-of-range
+     * values are rejected server-side with a 422. */
+    promptInfluence?: number;
 }
 /** One service's free-trial allowance. `remaining` is already floored at 0,
  * so it is safe to compare directly. */
@@ -329,6 +337,12 @@ interface VideoToVideoMusicParams {
      * by the free trial. This endpoint is always async, so no extra `mode`
      * gating applies. The result's `videos[]` gets one entry per variant. */
     variantsNum?: number;
+    /** How strongly the generated music follows the prompt (0-1, API default
+     * 0.5). Lower values let the video lead; higher values follow the prompt
+     * more literally. Free of charge. Omitted from the wire when unset so the
+     * server default applies; `0` is a meaningful value and is sent.
+     * Out-of-range values are rejected server-side with a 422. */
+    promptInfluence?: number;
 }
 interface VideoToVideoSfxParams {
     video?: VideoInput;
@@ -438,6 +452,34 @@ interface SoundResult extends BaseTaskResult {
     /** One entry per variant; see `SoundOutputEntry`. */
     outputs?: SoundOutputEntry[];
 }
+/** Params for `audioDucking`: mix an existing music bed under an existing
+ * voice track, ducking the music wherever the voice speaks. Exactly one of
+ * `voice`/`voiceUrl` and exactly one of `music`/`musicUrl`; a local input and
+ * a URL may be mixed across the two.
+ *
+ * The voice may be audio or video (`VideoInput` is the SDK's generic media
+ * input union) — a video's own audio track becomes the voice, and the ducked
+ * mix is re-muxed back into a new video. The music must be audio: the backend
+ * never probes it for a video stream, so a video there would be silently
+ * mishandled. Each input is capped at 360 seconds server-side. */
+interface AudioDuckingParams {
+    voice?: VideoInput;
+    voiceUrl?: string;
+    music?: VideoInput;
+    musicUrl?: string;
+}
+/** Result of an `audioDucking` task (`tasks.get`) or its final state
+ * (`generate`). Same flat envelope as `SoundResult`, but a ducking task
+ * renders exactly one artifact and no stems: a `.wav` (`output_type`
+ * "audio"), or a `.mp4` with the ducked mix re-muxed in (`output_type`
+ * "video") when the voice input was a video. Pass `output_url` to
+ * `download()`. */
+interface DuckingResult extends BaseTaskResult {
+    output_url?: string;
+    output_type?: "audio" | "video";
+    output_bytes?: number;
+    duration_seconds?: number;
+}
 /**
  * A target language for /v1/dubbing. The union stays open (`string & {}`) so a
  * language added server-side still type-checks against an older SDK — the
@@ -582,6 +624,21 @@ declare class VideoToVideoSound {
     generate(params: VideoToVideoSoundParams, opts?: WaitOptions): Promise<SoundResult>;
 }
 
+/** Duck an existing music bed under a voice track. Async only (202 + poll).
+ *
+ * Both inputs are user-supplied — nothing is generated here. The voice may be
+ * audio OR a video: the backend extracts a video's audio track, ducks the
+ * music under it, and re-muxes the ducked mix back into a new video (the
+ * result's `output_type` announces which came back). The music must be audio —
+ * the backend never probes it for a video stream, so a video there would be
+ * silently mishandled. */
+declare class AudioDucking {
+    private readonly client;
+    constructor(client: SoniloClient);
+    submit(params: AudioDuckingParams): Promise<SfxTask>;
+    generate(params: AudioDuckingParams, opts?: WaitOptions): Promise<DuckingResult>;
+}
+
 /** Dub a video into one or more target languages. Async only; the result
  * carries a language → dubbed-video-URL map under `outputs`. */
 declare class Dubbing {
@@ -628,6 +685,7 @@ declare class SoniloClient {
     readonly videoToVideoSfx: VideoToVideoSfx;
     readonly videoToSound: VideoToSound;
     readonly videoToVideoSound: VideoToVideoSound;
+    readonly audioDucking: AudioDucking;
     readonly dubbing: Dubbing;
     constructor(options?: SoniloClientOptions);
     /**
@@ -710,6 +768,6 @@ declare class RequestTimeoutError extends SoniloError {
 declare function download(media: SfxMedia | string | undefined, fetchFn?: typeof globalThis.fetch, timeout?: number): Promise<Uint8Array>;
 
 /** The SDK's own version. Generated by scripts/sync-versions.mjs — do not edit. */
-declare const VERSION = "0.12.0";
+declare const VERSION = "0.14.0";
 
-export { APIError, type AccountServices, type AudioChunkEvent, AuthenticationError, BadRequestError, type BaseTaskResult, type CompleteEvent, type CostEvent, type CostInfo, DEFAULT_TIMEOUT_MS, type DailyUsage, type DubbingLanguage, type DubbingParams, type DubbingResult, type ErrorEvent, GenerationError, type MusicMediaEntry, type MusicMuxEntry, type MusicTaskResult, type MusicTitle, PaymentRequiredError, RateLimitError, RequestTimeoutError, type Segment, type SegmentLabel, type SfxAudioFormat, type SfxError, type SfxMedia, type SfxResult, type SfxSegment, type SfxTask, SoniloClient, type SoniloClientOptions, SoniloError, type SoundOutputEntry, type SoundResult, type StreamEvent, TaskFailedError, TaskTimeoutError, type TextToMusicParams, type TextToSfxParams, type TitleEvent, type Track, TrialExhaustedError, type TrialQuota, type UnknownEvent, type UsageResponse, type UsageSummary, VERSION, type VideoInput, type VideoResult, type VideoToMusicParams, type VideoToSfxParams, type VideoToSoundParams, type VideoToVideoMusicParams, type VideoToVideoSfxParams, type VideoToVideoSoundParams, type WaitOptions, download, isAudioChunkEvent, isErrorEvent };
+export { APIError, type AccountServices, type AudioChunkEvent, type AudioDuckingParams, AuthenticationError, BadRequestError, type BaseTaskResult, type CompleteEvent, type CostEvent, type CostInfo, DEFAULT_TIMEOUT_MS, type DailyUsage, type DubbingLanguage, type DubbingParams, type DubbingResult, type DuckingResult, type ErrorEvent, GenerationError, type MusicMediaEntry, type MusicMuxEntry, type MusicTaskResult, type MusicTitle, PaymentRequiredError, RateLimitError, RequestTimeoutError, type Segment, type SegmentLabel, type SfxAudioFormat, type SfxError, type SfxMedia, type SfxResult, type SfxSegment, type SfxTask, SoniloClient, type SoniloClientOptions, SoniloError, type SoundOutputEntry, type SoundResult, type StreamEvent, TaskFailedError, TaskTimeoutError, type TextToMusicParams, type TextToSfxParams, type TitleEvent, type Track, TrialExhaustedError, type TrialQuota, type UnknownEvent, type UsageResponse, type UsageSummary, VERSION, type VideoInput, type VideoResult, type VideoToMusicParams, type VideoToSfxParams, type VideoToSoundParams, type VideoToVideoMusicParams, type VideoToVideoSfxParams, type VideoToVideoSoundParams, type WaitOptions, download, isAudioChunkEvent, isErrorEvent };

package/dist/index.js

@@ -392,6 +392,9 @@ var VideoToMusic = class {
     if (params.segments !== void 0) {
       form.set("segments", JSON.stringify(params.segments));
     }
+    if (params.promptInfluence !== void 0) {
+      form.set("prompt_influence", String(params.promptInfluence));
+    }
     const res = await this.client.request(
       "/v1/video-to-music",
       { method: "POST", body: form, signal: params.signal },
@@ -452,6 +455,9 @@ var VideoToMusic = class {
     if (params.variantsNum !== void 0) {
       form.set("variants_num", String(params.variantsNum));
     }
+    if (params.promptInfluence !== void 0) {
+      form.set("prompt_influence", String(params.promptInfluence));
+    }
     const res = await this.client.request("/v1/video-to-music", {
       method: "POST",
       body: form
@@ -548,6 +554,9 @@ var VideoToVideoMusic = class {
     if (params.variantsNum !== void 0) {
       form.set("variants_num", String(params.variantsNum));
     }
+    if (params.promptInfluence !== void 0) {
+      form.set("prompt_influence", String(params.promptInfluence));
+    }
     const res = await this.client.request("/v1/video-to-video-music", {
       method: "POST",
       body: form
@@ -661,6 +670,43 @@ var VideoToVideoSound = class {
   }
 };
 
+// src/resources/audioDucking.ts
+var AudioDucking = class {
+  constructor(client) {
+    this.client = client;
+  }
+  async submit(params) {
+    if (params.voice === void 0 === (params.voiceUrl === void 0)) {
+      throw new SoniloError("Provide exactly one of voice or voiceUrl");
+    }
+    if (params.music === void 0 === (params.musicUrl === void 0)) {
+      throw new SoniloError("Provide exactly one of music or musicUrl");
+    }
+    const form = new FormData();
+    if (params.voice !== void 0) {
+      const { blob, filename } = await toUploadBlob(params.voice);
+      form.set("voice_file", blob, filename);
+    } else {
+      form.set("voice_url", params.voiceUrl);
+    }
+    if (params.music !== void 0) {
+      const { blob, filename } = await toUploadBlob(params.music);
+      form.set("music_file", blob, filename);
+    } else {
+      form.set("music_url", params.musicUrl);
+    }
+    const res = await this.client.request("/v1/audio-ducking", {
+      method: "POST",
+      body: form
+    });
+    return await res.json();
+  }
+  async generate(params, opts) {
+    const task = await this.submit(params);
+    return this.client.tasks.wait(task.task_id, opts);
+  }
+};
+
 // src/resources/dubbing.ts
 async function buildDubbingForm(params) {
   if (params.video === void 0 === (params.videoUrl === void 0)) {
@@ -705,7 +751,7 @@ var Dubbing = class {
 };
 
 // src/version.ts
-var VERSION = "0.12.0";
+var VERSION = "0.14.0";
 
 // src/client.ts
 var DEFAULT_BASE_URL = "https://api.sonilo.com";
@@ -736,6 +782,7 @@ var SoniloClient = class {
     this.videoToVideoSfx = new VideoToVideoSfx(this);
     this.videoToSound = new VideoToSound(this);
     this.videoToVideoSound = new VideoToVideoSound(this);
+    this.audioDucking = new AudioDucking(this);
     this.dubbing = new Dubbing(this);
   }
   /**