You can deploy this model and start earning money today!

MadJ99/
musicgen-melody-as-new-updated

Name: musicgen-melody-as-new-updated
Availability: InStock

safetensors

Views

No views yet

as-new-updated

Model Description

Latest model with more training data

This model was fine-tuned from facebook/musicgen-melody using LoRA adapter technology.

Instance Prompt

Indian classical fusion, Anoushka Shankar

Usage

python

from transformers import AutoProcessor, AutoModelForTextToWaveform
from peft import PeftModel, PeftConfig
import torch

# Load model configuration
config = PeftConfig.from_pretrained("as-new-updated")

# Load base model
model = AutoModelForTextToWaveform.from_pretrained(
    config.base_model_name_or_path,
    torch_dtype=torch.float16
)

# Load fine-tuned weights
model = PeftModel.from_pretrained(model, "as-new-updated")

# Load processor
processor = AutoProcessor.from_pretrained(config.base_model_name_or_path)

# Generate audio
inputs = processor(
    text=["Indian classical fusion with sitar and tabla"],
    padding=True,
    return_tensors="pt"
)

audio_values = model.generate(**inputs, do_sample=True, guidance_scale=3, max_new_tokens=1024)

Sample Prompts

"Indian classical fusion with sitar and tabla, introspective ambient mood"
"Sitar melody with cross-cultural elements and minimalist production"
"Contemplative raga improvisation with electronic textures and organic production"

License

This model adaptation is shared under the same license as the original MusicGen model.

MadJ99/musicgen-melody-as-new-updated