wav2vec2-large-960h-lv60-self-en-atc-uwb-atcc
This model is a fine-tuned version of facebook/wav2vec2-large-960h-lv60-self on the UWB-ATCC corpus.
It achieves the following results on the evaluation set:
- Loss: 0.7287
- Wer: 0.1756
Authors: Juan Zuluaga-Gomez, Amrutha Prasad, Iuliia Nigmatulina, Saeed Sarfjoo, Petr Motlicek, Matthias Kleinert, Hartmut Helmke, Oliver Ohneiser, Qingran Zhan
Abstract: Recent work on self-supervised pre-training focus on leveraging large-scale unlabeled speech data to build robust end-to-end (E2E)acoustic models (AM) that can be later fine-tuned on downstream tasks e.g., automatic speech recognition (ASR). Yet, few works investigated the impact on performance when the data properties substantially differ between the pre-training and fine-tuning phases, termed domain shift. We target this scenario by analyzing the robustness of Wav2Vec 2.0 and XLS-R models on downstream ASR for a completely unseen domain, air traffic control (ATC) communications. We benchmark these two models on several open-source and challenging ATC databases with signal-to-noise ratio between 5 and 20 dB. Relative word error rate (WER) reductions between 20% to 40% are obtained in comparison to hybrid-based ASR baselines by only fine-tuning E2E acoustic models with a smaller fraction of labeled data. We analyze WERs on the low-resource scenario and gender bias carried by one ATC dataset.
Code — GitHub repository: https://github.com/idiap/w2v2-air-traffic
Usage
You can use our Google Colab notebook to run and evaluate our model: https://github.com/idiap/w2v2-air-traffic/blob/master/src/eval_xlsr_atc_model.ipynb
Intended uses & limitations
This model was fine-tuned on air traffic control data. We don't expect that it keeps the same performance on some others datasets, e.g., LibriSpeech or CommonVoice.
Training and evaluation data
See Table 1 (page 3) in our paper: How Does Pre-trained Wav2Vec 2.0 Perform on Domain Shifted ASR? An Extensive Benchmark on Air Traffic Control Communications. We described there the partitions of how to use our model.
- We use the UWB-ATCC corpus to fine-tune this model. You can download the raw data here: https://lindat.mff.cuni.cz/repository/xmlui/handle/11858/00-097C-0000-0001-CCA1-0
- However, do not worry, we have prepared the database in
Datasets format
. Here, UWB-ATCC corpus on HuggingFace. You can scroll and check the train/test partitions, and even listen to some audios. - If you want to prepare a database in HuggingFace format, you can follow the data loader script in: data_loader_atc.py.
Writing your own inference script
If you use language model, you need to install the KenLM bindings with:
conda activate your_environment
pip install https://github.com/kpu/kenlm/archive/master.zip
The snippet of code:
from datasets import load_dataset, load_metric, Audio
import torch
from transformers import AutoModelForCTC, Wav2Vec2Processor, Wav2Vec2ProcessorWithLM
import torchaudio.functional as F
USE_LM = False
DATASET_ID = "Jzuluaga/uwb_atcc"
MODEL_ID = "Jzuluaga/wav2vec2-large-960h-lv60-self-en-atc-uwb-atcc"
# 1. Load the dataset
# we only load the 'test' partition, however, if you want to load the 'train' partition, you can change it accordingly
uwb_atcc_corpus_test = load_dataset(DATASET_ID, "test", split="test")
# 2. Load the model
model = AutoModelForCTC.from_pretrained(MODEL_ID)
# 3. Load the processors, we offer support with LM, which should yield better resutls
if USE_LM:
processor = Wav2Vec2ProcessorWithLM.from_pretrained(MODEL_ID)
else:
processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
# 4. Format the test sample
sample = next(iter(uwb_atcc_corpus_test))
file_sampling_rate = sample['audio']['sampling_rate']
# resample if neccessary
if file_sampling_rate != 16000:
resampled_audio = F.resample(torch.tensor(sample["audio"]["array"]), file_sampling_rate, 16000).numpy()
else:
resampled_audio = torch.tensor(sample["audio"]["array"]).numpy()
input_values = processor(resampled_audio, return_tensors="pt").input_values
# 5. Run the forward pass in the model
with torch.no_grad():
logits = model(input_values).logits
# get the transcription with processor
if USE_LM:
transcription = processor.batch_decode(logits.numpy()).text
else:
pred_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(pred_ids)
# print the output
print(transcription)
Cite us
If you use this code for your research, please cite our paper with:
@article{zuluaga2022how,
title={How Does Pre-trained Wav2Vec2. 0 Perform on Domain Shifted ASR? An Extensive Benchmark on Air Traffic Control Communications},
author={Zuluaga-Gomez, Juan and Prasad, Amrutha and Nigmatulina, Iuliia and Sarfjoo, Saeed and others},
journal={IEEE Spoken Language Technology Workshop (SLT), Doha, Qatar},
year={2022}
}
and,
@article{zuluaga2022bertraffic,
title={BERTraffic: BERT-based Joint Speaker Role and Speaker Change Detection for Air Traffic Control Communications},
author={Zuluaga-Gomez, Juan and Sarfjoo, Seyyed Saeed and Prasad, Amrutha and others},
journal={IEEE Spoken Language Technology Workshop (SLT), Doha, Qatar},
year={2022}
}
and,
@article{zuluaga2022atco2,
title={ATCO2 corpus: A Large-Scale Dataset for Research on Automatic Speech Recognition and Natural Language Understanding of Air Traffic Control Communications},
author={Zuluaga-Gomez, Juan and Vesel{\`y}, Karel and Sz{\"o}ke, Igor and Motlicek, Petr and others},
journal={arXiv preprint arXiv:2211.04054},
year={2022}
}
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 24
- eval_batch_size: 12
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
No log | 1.06 | 500 | 2.9016 | 0.9995 |
2.877 | 2.12 | 1000 | 0.9812 | 0.3485 |
2.877 | 3.18 | 1500 | 0.7842 | 0.2732 |
0.7834 | 4.25 | 2000 | 0.6962 | 0.2192 |
0.7834 | 5.31 | 2500 | 0.6527 | 0.2042 |
0.6084 | 6.37 | 3000 | 0.6220 | 0.1972 |
0.6084 | 7.43 | 3500 | 0.6442 | 0.1934 |
0.5147 | 8.49 | 4000 | 0.6793 | 0.1950 |
0.5147 | 9.55 | 4500 | 0.6432 | 0.1920 |
0.4566 | 10.62 | 5000 | 0.6605 | 0.1853 |
0.4566 | 11.68 | 5500 | 0.6393 | 0.1866 |
0.4155 | 12.74 | 6000 | 0.6918 | 0.1803 |
0.4155 | 13.8 | 6500 | 0.6514 | 0.1791 |
0.372 | 14.86 | 7000 | 0.7010 | 0.1851 |
0.372 | 15.92 | 7500 | 0.6824 | 0.1786 |
0.3368 | 16.99 | 8000 | 0.6895 | 0.1780 |
0.3368 | 18.05 | 8500 | 0.7150 | 0.1759 |
0.3244 | 19.11 | 9000 | 0.7141 | 0.1759 |
0.3244 | 20.17 | 9500 | 0.7225 | 0.1756 |
0.2981 | 21.23 | 10000 | 0.7287 | 0.1756 |
Framework versions
- Transformers 4.24.0
- Pytorch 1.13.0+cu117
- Datasets 2.6.1
- Tokenizers 0.13.2
- Downloads last month
- 10
Model tree for Jzuluaga/wav2vec2-large-960h-lv60-self-en-atc-uwb-atcc
Base model
facebook/wav2vec2-large-960h-lv60-selfDataset used to train Jzuluaga/wav2vec2-large-960h-lv60-self-en-atc-uwb-atcc
Evaluation results
- TEST WER on UWB-ATCC dataset (Air Traffic Control Communications)test set self-reported17.200
- TEST WER (+LM) on UWB-ATCC dataset (Air Traffic Control Communications)test set self-reported13.720
- TEST WER on ATCOSIM corpus (Air Traffic Control Communications)test set self-reported15.310
- TEST WER (+LM) on ATCOSIM corpus (Air Traffic Control Communications)test set self-reported11.880