Karya-MSRI@AmericasNLP

Environment Specifications

Both the translate and asr inference/training require separate environments that can be installed using the following requirement specifications.

For Translate

pip3 install -r translate_requirements.txt

For ASR

For Translate:

pip3 install -r asr_requirements.txt

Scripts

├── src
│   ├── asr.py                          # Fine-tuning ASR models & utilizing KenLM during training 
│   ├── construct_kenlm.py              # Constructing kenlm data + training kenlm models 
│   ├── debug                           # WIP
│   │   └── asr_with_kenlm.py
│   ├── inference.py                    # Inference for ASR Models (KenLM support to be added) 
│   ├── translate.py                    # Training for the translate models 
│   ├── translate_inference.py          # Inference for the translation models 
│   └── translate_preprocess.py         # Preprocessing script to convert free-form data to HuggingFace Dataset form

├── scripts
│   ├── automate.sh                     #Automating the training of the asr models 
│   ├── inference.sh                    #Automating the inference of the asr models
│   ├── train_kenlm.sh                  #Automating the training of the kenlm models
│   ├── translate.sh                    #Automating the training of the translate models
│   └── translate_inference.sh          #Automating the inference of the asr models