FLAC

Official model weights for FLAC, from the paper "Few-shot Acoustic Synthesis with Multimodal Flow Matching" by Amandine Brunetto (CVPR 2026).

FLAC is a model for generating room impulse responses in novel scenes in a one-shot setting, relying only on a single RIR example and a depth image.

This repository contains the pretrained weights. To run FLAC, please use the official codebase.

Available checkpoints

file	description
FLAC.ckpt	main trained model on the AR dataset
FLAC_EMA.ckpt	EMA weights for finetuning FLAC on the HAA dataset
FLAC_HAA.ckpt	fine-tuned FLAC model on the HAA dataset
VAE.safetensors	VAE model for training FLAC
VAE.ckpt	VAE model for VAE evaluation

Weights can be downloaded with:

huggingface-cli download AmandineBtto/FLAC --local-dir weights/FLAC

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support