Upload folder using huggingface_hub

README.md

@@ -13,11 +13,9 @@ base_model: ACE-Step/ACE-Step1.5
 
 4-bit quantized MLX weights for [ACE-Step/ACE-Step1.5](https://huggingface.co/ACE-Step/ACE-Step1.5).
 
-Converted and quantized using [mlx-audio](https://github.com/Blaizzy/mlx-audio).
-
 - Decoder and encoder quantized to 4-bit (group_size=64)
 - VAE, tokenizer, and detokenizer kept in full precision
-- **2.2GB** main model (vs 9.6GB fp32) + 0.7GB VAE + 2.4GB text encoder
+- **2.2GB** main model + 0.7GB VAE + 2.4GB text encoder
 
 ## Usage
 
@@ -34,6 +32,24 @@ for result in model.generate(
     sample_rate = result.sample_rate
 ```
 
+## With Vocals
+
+```python
+for result in model.generate(
+    text="English pop song with clear female vocals, catchy melody",
+    lyrics="""[verse]
+Dance with me tonight
+Under the neon lights
+
+[chorus]
+We're alive, we're on fire
+Dancing higher and higher
+""",
+    duration=60.0,
+    vocal_language="en",
+):
+    ...
+```
+
 The model uses a 5Hz Language Model planner by default (`use_lm=True`) which generates
-a song blueprint before running the diffusion transformer. This is required for the
-turbo model to produce music.
+a song blueprint before running the diffusion transformer.