jacktol's picture
Update README.md
8e04c74 verified
|
raw
history blame
4.69 kB
---
license: mit
datasets:
- jacktol/atc-dataset
language:
- en
metrics:
- wer
base_model:
- openai/whisper-medium.en
pipeline_tag: automatic-speech-recognition
tags:
- aviation
- atc
- aircraft
- communication
model-index:
- name: Whisper Medium EN Fine-Tuned for ATC
results:
- task:
type: automatic-speech-recognition
dataset:
name: ATC Dataset
type: jacktol/atc-dataset
metrics:
- name: Word Error Rate (WER)
type: wer
value: 15.08
source:
name: ATC Transcription Evaluation
url: https://jacktol.net/posts/fine-tuning_whisper_for_atc/
---
# Whisper Medium EN Fine-Tuned for Air Traffic Control (ATC)
## Model Overview
This model is a fine-tuned version of OpenAI's Whisper Medium EN model, specifically trained on **Air Traffic Control (ATC)** communication datasets. The fine-tuning process significantly improves transcription accuracy on domain-specific aviation communications, reducing the **Word Error Rate (WER) by 84%**, compared to the original pretrained model. The model is particularly effective at handling accent variations and ambiguous phrasing often encountered in ATC communications.
- **Base Model**: OpenAI Whisper Medium EN
- **Fine-tuned Model WER**: 15.08%
- **Pretrained Model WER**: 94.59%
- **Relative Improvement**: 84.06%
You can access the fine-tuned model on Hugging Face:
- [Whisper Medium EN Fine-Tuned for ATC](https://huggingface.co/jacktol/whisper-medium.en-fine-tuned-for-ATC)
- [Whisper Medium EN Fine-Tuned for ATC (Faster Whisper)](https://huggingface.co/jacktol/whisper-medium.en-fine-tuned-for-ATC-faster-whisper)
## Model Description
Whisper Medium EN fine-tuned for ATC is optimized to handle short, distinct transmissions between pilots and air traffic controllers. It is fine-tuned using data from:
- **[ATCO2 corpus](https://huggingface.co/datasets/Jzuluaga/atco2_corpus_1h)** (1-hour test subset)
- **[UWB-ATCC corpus](https://huggingface.co/datasets/Jzuluaga/uwb_atcc)**
The fine-tuned model demonstrates enhanced performance in interpreting various accents, recognizing non-standard phraseology, and processing noisy or distorted communications. It is highly suitable for aviation-related transcription tasks.
## Intended Use
The fine-tuned Whisper model is designed for:
- **Transcribing aviation communication**: Providing accurate transcriptions for ATC communications, including accents and variations in English phrasing.
- **Air Traffic Control Systems**: Assisting in real-time transcription of pilot-ATC conversations, helping improve situational awareness.
- **Research and training**: Useful for researchers, developers, or aviation professionals studying ATC communication or developing new tools for aviation safety.
You can test the model online using the [ATC Transcription Assistant](https://huggingface.co/spaces/jacktol/ATC-Transcription-Assistant), which lets you upload audio files and generate transcriptions.
## Model Description
Whisper Medium EN fine-tuned for ATC is optimized to handle short, distinct transmissions between pilots and air traffic controllers. It is fine-tuned using data from the **[ATC Dataset](https://huggingface.co/datasets/jacktol/atc-dataset)**, a combined and cleaned dataset sourced from the following:
- **[ATCO2 corpus](https://huggingface.co/datasets/Jzuluaga/atco2_corpus_1h)** (1-hour test subset)
- **[UWB-ATCC corpus](https://huggingface.co/datasets/Jzuluaga/uwb_atcc)**
The **ATC Dataset** merges these two original sources, filtering and refining the data to enhance transcription accuracy for domain-specific ATC communications.
## Training Procedure
- **Hardware**: Fine-tuning was conducted on two A100 GPUs with 80GB memory.
- **Epochs**: 10
- **Learning Rate**: 1e-5
- **Batch Size**: 32 (effective batch size with gradient accumulation)
- **Augmentation**: Dynamic data augmentation techniques (Gaussian noise, pitch shifting, etc.) were applied during training.
- **Evaluation Metric**: Word Error Rate (WER)
## Limitations
While the fine-tuned model performs well in ATC-specific communications, it may not generalize as effectively to other domains of speech. Additionally, like most speech-to-text models, transcription accuracy can be affected by extremely poor-quality audio or heavily accented speech not encountered during training.
## References
- **Blog Post**: [Fine-Tuning Whisper for ATC: 84% Improvement in Transcription Accuracy](https://jacktol.net/posts/fine-tuning_whisper_for_atc/)
- **GitHub Repository**: [Fine-Tuning Whisper on ATC Data](https://github.com/jack-tol/fine-tuning-whisper-on-atc-data/tree/main)