basic music generation

app.py

@@ -1,7 +1,20 @@
+import torch
 import gradio as gr
 
-def greet(name):
-    return "Hello " + name + "!!"
+from transformers import MusicgenForConditionalGeneration
+model = MusicgenForConditionalGeneration.from_pretrained("facebook/musicgen-small")
+sampling_rate = model.config.audio_encoder.sampling_rate
 
-iface = gr.Interface(fn=greet, inputs="text", outputs="text")
-iface.launch()
+device = "cuda" if torch.cuda.is_available() else "cpu"
+model.to(device)
+
+from transformers import AutoProcessor
+processor = AutoProcessor.from_pretrained("facebook/musicgen-small")
+
+
+def generate_music(desc):
+    inputs = processor(text=[desc], padding=True, return_tensors="pt")
+    audio_values = model.generate(**inputs.to(device), do_sample=True, guidance_scale=3, max_new_tokens=256)
+    return sampling_rate, audio_values[0][0].cpu().numpy()
+    
+gr.Interface(fn=generate_music, inputs="text", outputs="audio").launch()

requirements.txt

@@ -0,0 +1,4 @@
+transformers
+torch
+torchvision
+torchaudio