fortunedavis/tiny_asr_from_scratch

This repository presents lightweight Automatic Speech Recognition (ASR) systems based on LSTM and Transformer architectures. The models are trained and evaluated on African language speech datasets, with a focus on building efficient and reproducible ASR pipelines. # tiny_asr_from_scratch This repository presents lightweight Automatic Speech Recognition (ASR) systems based on LSTM and Transformer architectures. The models are trained and evaluated on African language speech datasets, with a focus on building efficient and reproducible ASR pipelines.