Shypd.aiThe largest open index of AI toolsdeepspeech.pytorch is an open-source speech recognition tool that implements DeepSpeech2 using PyTorch. It supports training, testing, and inference, with optional KenLM language model integration for enhanced decoding.
Trending
About
deepspeech.pytorch is an open-source implementation of the DeepSpeech2 speech recognition model, built with PyTorch and PyTorch Lightning. This tool provides comprehensive functionalities for training, testing, and performing inference with speech-to-text models. It supports various datasets like AN4, TEDLIUM, Voxforge, Common Voice, and LibriSpeech, and allows for custom dataset creation. Key features include multi-GPU and multi-node training capabilities, various augmentation techniques such as SpecAugment, noise injection, and tempo/gain perturbations to improve model robustness. Users can also integrate KenLM language models for more accurate beam search decoding and even build their own custom LMs. A basic inference server is also included for transcribing audio files via POST requests.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
