Aspiration_Mel_Band_Roformer

A Mel-Band-Roformer model that is used to separate aspiration.

How to use

• Clone this repository.

git clone https://github.com/SUC-DriverOld/Aspiration-Mel-Band-Roformer.git

• Install the requirements.

pip install -r requirements.txt

• Download the models.

  • Model: aspiration_mel_band_roformer_sdr_18.9845.ckpt
  • Model (less aggressive): aspiration_mel_band_roformer_less_aggr_sdr_18.1201.ckpt

• Inference.

usage: inference.py [-h] [--config_path CONFIG_PATH] [--model_path MODEL_PATH] [--input_folder INPUT_FOLDER] 
                    [--store_dir STORE_DIR] [--device_ids DEVICE_IDS [DEVICE_IDS ...]]

options:
  -h, --help                    show this help message and exit
  --config_path CONFIG_PATH     path to config yaml file
  --model_path MODEL_PATH       Location of the model
  --input_folder INPUT_FOLDER   folder with songs to process
  --store_dir STORE_DIR         path to store model outputs
  --device_ids DEVICE_IDS       list of gpu ids

Description

Huggingface model repository: https://huggingface.co/Sucial/Aspiration_Mel_Band_Roformer
You can try listening to the performance of this model here.

Instruments: aspiration, other
Dataset: My own datasets(171 songs for training and 17 songs for validation).
Finetuned from: model_mel_band_roformer_ep_3005_sdr_11.4360.ckpt
Configs: config_aspiration_mel_band_roformer.yaml

Model Name	aspiration_mel_band_roformer_sdr_18.9845.ckpt	aspiration_mel_band_roformer_less_aggr_sdr_18.1201.ckpt
Epoch	123	27
Instr SDR aspiration	9.8554	9.0704
Instr SDR other	28.1136	27.1699
SDR Avg	18.9845	18.1201

Thanks to

• Mel-Band-Roformer [Paper, Repository]
• ZFTurbo's training code [Music-Source-Separation-Training]
• KimberleyJensen's simple inference code [Mel-Band-Roformer-Vocal-Model]
• Spacehpc provided the server for training.