@sogni-ai/sogni-client 5.58.2 → 5.60.0

package/src/Chat/_hostedToolsManifest.generated.ts

@@ -341,19 +341,19 @@ export const SOGNI_HOSTED_TOOLS_MANIFEST: unknown = {
             },
             "duration": {
               "type": "number",
-              "description": "Duration in seconds. Default: 30. Range: 10-600 (10 seconds to 10 minutes). Short clips: 10-30s. Standard songs: 120-300s.",
+              "description": "Duration in seconds. music3 (the default model): 10-300, default 60, and a ceiling — the song may end earlier at a musical resolution. ACE-Step turbo/sft: 10-600, default 30. Short clips: 10-30s. Standard songs: 120-300s.",
               "minimum": 10,
               "maximum": 600
             },
             "bpm": {
               "type": "number",
-              "description": "Beats per minute / tempo. Default: 120. Range: 30-300. Slow ballad: 60-80. Mid-tempo: 90-120. Upbeat: 120-140. Fast dance: 140-180. Very fast: 180+.",
+              "description": "ACE-Step (turbo, sft) only — music3 does not use it; for music3 put the tempo in the prompt instead. Beats per minute / tempo. Default: 120. Range: 30-300. Slow ballad: 60-80. Mid-tempo: 90-120. Upbeat: 120-140. Fast dance: 140-180. Very fast: 180+.",
               "minimum": 30,
               "maximum": 300
             },
             "keyscale": {
               "type": "string",
-              "description": "Musical key and scale. E.g., \"C major\", \"A minor\", \"F# minor\", \"Bb major\". Default: \"C major\". Only set when the user specifies a key or when a particular mood calls for it (minor keys for sad/dark, major for happy/bright)."
+              "description": "ACE-Step (turbo, sft) only — music3 does not use it; for music3 put the key in the prompt instead. Musical key and scale. E.g., \"C major\", \"A minor\", \"F# minor\", \"Bb major\". Default: \"C major\". Only set when the user specifies a key or when a particular mood calls for it (minor keys for sad/dark, major for happy/bright)."
             },
             "lyrics": {
               "type": "string",
@@ -362,13 +362,13 @@ export const SOGNI_HOSTED_TOOLS_MANIFEST: unknown = {
             "model": {
               "type": "string",
               "enum": [
+                "minimax_music3",
                 "ace_step_1.5_xl_turbo",
                 "ace_step_1.5_xl_sft",
                 "ace_step_1.5_turbo",
-                "ace_step_1.5_sft",
-                "minimax_music3"
+                "ace_step_1.5_sft"
               ],
-              "description": "Canonical music model ID. Default: ace_step_1.5_xl_turbo. Use minimax_music3 (MiniMax Music 3, premium autoregressive composer, ~20x cost, duration is a ceiling) when the user asks for the best quality, realistic vocals, or full songs. Use ace_step_1.5_xl_sft only when the user explicitly requests XL SFT. Use legacy ace_step_1.5_turbo or ace_step_1.5_sft only when the user explicitly requests a legacy model."
+              "description": "Canonical music model ID. Default: minimax_music3 (MiniMax Music 3, premium autoregressive composer with the best vocals, lyric adherence and song structure; 10-300 s, and duration is a ceiling, so the song may end earlier at a musical resolution). Music 3 takes tempo and key in the prompt text; bpm, keyscale and timesig apply to ACE-Step only. Use ace_step_1.5_xl_turbo only when the user asks for a quick, cheap or draft track, names ACE-Step, or wants a track longer than 300 s (with no model named, a duration over 300 s uses ACE-Step XL Turbo). Use ace_step_1.5_xl_sft only when the user explicitly requests XL SFT. Use legacy ace_step_1.5_turbo or ace_step_1.5_sft only when the user explicitly requests a legacy model."
             },
             "timesig": {
               "type": "number",
@@ -378,7 +378,7 @@ export const SOGNI_HOSTED_TOOLS_MANIFEST: unknown = {
                 4,
                 6
               ],
-              "description": "Time signature (beats per measure). 4 = 4/4 time (default, most common). 3 = 3/4 time (waltz). 2 = 2/4 time (march). 6 = 6/8 time (compound). Default: 4."
+              "description": "ACE-Step (turbo, sft) only — music3 does not use it; for music3 put the time signature in the prompt instead. Time signature (beats per measure). 4 = 4/4 time (default, most common). 3 = 3/4 time (waltz). 2 = 2/4 time (march). 6 = 6/8 time (compound). Default: 4."
             },
             "numberOfVariations": {
               "type": "number",

package/src/Chat/modelRouting.ts

@@ -165,11 +165,13 @@ export const PREFERRED_MODEL_IDS = {
     animateReplace: 'wan_v2.2-14b-fp8_animate-replace_lightx2v'
   },
   audio: {
+    // Music, in preference order: MiniMax Music 3 is the default music model;
+    // ACE-Step is the explicit draft or legacy choice.
+    minimaxMusic3: 'minimax_music3',
     aceStepXlTurbo: 'ace_step_1.5_xl_turbo',
     aceStepXlSft: 'ace_step_1.5_xl_sft',
     aceStepTurbo: 'ace_step_1.5_turbo',
     aceStepSft: 'ace_step_1.5_sft',
-    minimaxMusic3: 'minimax_music3',
     // Speech, not music: the same audio media type, a different job entirely.
     qwen3TtsCustomVoice: 'qwen3_tts_1.7b_custom_voice_bf16',
     qwen3TtsVoiceClone: 'qwen3_tts_1.7b_voice_clone_bf16',
@@ -177,6 +179,21 @@ export const PREFERRED_MODEL_IDS = {
   }
 } as const;
 
+/**
+ * Music models in preference order for a model-less generate_music call:
+ * MiniMax Music 3 first, then ACE-Step. Speech models are deliberately absent.
+ */
+export const MUSIC_MODEL_IDS: string[] = [
+  PREFERRED_MODEL_IDS.audio.minimaxMusic3,
+  PREFERRED_MODEL_IDS.audio.aceStepXlTurbo,
+  PREFERRED_MODEL_IDS.audio.aceStepXlSft,
+  PREFERRED_MODEL_IDS.audio.aceStepTurbo,
+  PREFERRED_MODEL_IDS.audio.aceStepSft
+];
+
+/** MiniMax Music 3's longest track, in seconds (ACE-Step renders up to 600). */
+export const MINIMAX_MUSIC3_MAX_DURATION_SECONDS = 300;
+
 const GPT_IMAGE_MODEL_ALIASES = [
   'chatgpt',
   'chatgpt-image',

package/src/Projects/createJobRequestMessage.ts

@@ -1187,6 +1187,15 @@ function getControlNet(params: ControlNetParams): ControlNetParamsRaw[] {
       propertyName: 'guidanceEnd'
     });
   }
+  if (params.preprocess !== undefined) {
+    if (typeof params.preprocess !== 'boolean') {
+      throw new Error('controlNet.preprocess must be a boolean');
+    }
+    // Sent only when true, so requests without it stay identical to before.
+    if (params.preprocess) {
+      cn.preprocess = true;
+    }
+  }
   return [cn];
 }
 

package/src/Projects/types/ControlNetParams.ts

@@ -42,6 +42,11 @@ export interface ControlNetParamsRaw {
   controlMode?: 0 | 1 | 2;
   controlGuidanceStart?: number;
   controlGuidanceEnd?: number;
+  /**
+   * Sent only as `true`: the worker makes the control map from the uploaded image.
+   * Omitted: the uploaded image is the control map.
+   */
+  preprocess?: true;
 }
 
 /**
@@ -88,6 +93,19 @@ export interface ControlNetParams {
    * Must be greater than guidanceStart
    */
   guidanceEnd?: number;
+  /**
+   * Make the control map from `image` on the worker. Default `false`.
+   *
+   * - `true`: `image` is an ordinary photo. The worker generates the control map that matches `name`
+   *   from it (edges for `canny`, a depth map for `depth`, a pose skeleton for `openpose`, and so on).
+   *   Use this for ordinary photos.
+   * - `false` or omitted: `image` is used as the control map as-is, exactly as before this option existed.
+   *   Use this for ready-made depth maps, pose skeletons, edge maps and scribbles.
+   *
+   * Only affects the Stable Diffusion ControlNet types. InstantID always uses the face photo as given,
+   * and `inpaint` / `instrp2p` always use the image as given.
+   */
+  preprocess?: boolean;
 }
 
 /**

package/src/version.ts

@@ -1,3 +1,3 @@
 // AUTO-GENERATED by scripts/generate-version.mjs — do not edit.
 // Source of truth: package.json version.
-export const LIB_VERSION = '5.58.2';
+export const LIB_VERSION = '5.60.0';