---
license: gpl-3.0
---

Dereverb MelBandRoformer (by @anvuew) OpenVINO Models

This repo stores OpenVINO(TM) models in IR format that are used to perform Reverb Extraction & Removal.

The OpenVINO IRs (.xml, .bin files) stored here have been converted from @anvuew's pytorch model checkpoints / configs from here:  
https://huggingface.co/anvuew/dereverb_mel_band_roformer

They are also uploaded to this repo, under `pytorch` folder.

The OpenVINO IRs are intended to be used with the set of OpenVINO-based AI plugins for Audacity(R), here: https://github.com/intel/openvino-plugins-ai-audacity

To better support a range of OpenVINO-supported devices, the MelBandRoformer model has been sliced / converted to 3 separate OpenVINO IRs:
* mel_band_pre.xml/.bin -> Pre-processing operations (such as STFT) which convert input audio waveforms to frequency domain.
* mel_band_fwd.xml / .bin -> The majority of the layers / ops in the original model.
* mel_band_post.xml / .bin -> Post-processing operations (such as iSTFT) which convert frequency domain outputs from `mel_band_fwd` to output waveforms. 

The OpenVINO IRs in `mono` directory are a conversion of `pytorch/dereverb_mel_band_roformer_mono_anvuew_sdr_20.4029.ckpt`.

## Intel’s Human Rights Disclaimer:
Intel is committed to respecting human rights and avoiding complicity in human rights abuses. See Intel's Global Human Rights Principles. Intel's products and software are intended only to be used in applications that do not cause or contribute to a violation of an internationally recognized human right.