Shypd.aiThe largest open index of AI toolskospeech is an open-source toolkit for end-to-end Korean automatic speech recognition. It leverages PyTorch and Hydra to develop and train speech recognition models.
Trending
About
kospeech is an open-source toolkit designed for end-to-end Korean automatic speech recognition (ASR), built upon PyTorch and Hydra. It provides a modular and extensible framework for researchers and developers to build and experiment with various ASR models. The toolkit addresses the lack of established preprocessing methods and baseline models for the KsponSpeech corpus, offering guidelines and implementations for several models including Deep Speech 2, LAS, Transformer, Jasper, and Conformer. It supports both KsponSpeech and LibriSpeech datasets, allowing users to train their own voice recognition models and evaluate their performance. The project is archived, recommending OpenSpeech for training and Pororo ASR or Whisper for testing trained Korean models.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
