feat: add Stable Audio 3 text-to-audio model

mlx_audio/convert.py

@@ -243,10 +243,11 @@ def get_model_path(path_or_hf_repo: str, revision: Optional[str] = None) -> Path
 
 def load_config(model_path: Path) -> dict:
     """Load model configuration from a path."""
-    config_path = model_path / "config.json"
-    if config_path.exists():
-        with open(config_path, "r", encoding="utf-8") as f:
-            return json.load(f)
+    for name in ("config.json", "model_config.json"):
+        config_path = model_path / name
+        if config_path.exists():
+            with open(config_path, "r", encoding="utf-8") as f:
+                return json.load(f)
     raise FileNotFoundError(f"Config not found at {model_path}")
 
 

mlx_audio/tts/generate.py

@@ -596,6 +596,24 @@ def parse_args():
         action="store_true",
         help="Optional model-specific zero speaker embedding mode (e.g., Ming Omni).",
     )
+    parser.add_argument(
+        "--duration",
+        type=float,
+        default=None,
+        help="Audio duration in seconds (for diffusion models like Stable Audio 3).",
+    )
+    parser.add_argument(
+        "--sampler",
+        type=str,
+        default=None,
+        help="Sampler type (e.g., 'pingpong', 'euler' for Stable Audio 3).",
+    )
+    parser.add_argument(
+        "--seed",
+        type=int,
+        default=None,
+        help="Random seed for reproducible generation.",
+    )
     parser.add_argument("--top_p", type=float, default=0.9, help="Top-p for the model")
     parser.add_argument("--top_k", type=int, default=50, help="Top-k for the model")
     parser.add_argument(

mlx_audio/tts/models/stable_audio_3/__init__.py

@@ -0,0 +1,3 @@
+from .stable_audio_3 import Model, ModelConfig
+
+__all__ = ["Model", "ModelConfig"]