k2-fsa/sherpa

Speech-to-text server framework with next-gen Kaldi

What it solves

It provides a streamlined way to deploy pre-trained end-to-end speech-to-text models, removing the need for the training infrastructure when you only need to transcribe speech.

How it works

The framework uses PyTorch to run inference on transducer- and CTC-based models. It offers both C++ and Python APIs to integrate these models into applications.

Who it’s for

Developers who want to use pre-trained speech-to-text models for transcription tasks without needing to train or fine-tune them.

Highlights

  • End-to-end (E2E) model focus
  • Supports both C++ and Python APIs
  • Built on PyTorch
  • Specifically designed for deployment and inference

Related

  • Project
    InterDigitalInc/CompressAI

    A PyTorch library and evaluation platform for end-to-end deep learning-based image and video compression research.

  • Project
    espnet/espnet

    ESPnet is an end-to-end speech processing toolkit built on PyTorch that provides a unified, reproducible framework for tasks like speech recognition, text-to-speech, and speech translation.

  • Project
    nschaetti/EchoTorch

    A PyTorch-based module for implementing and testing Echo State Networks and Reservoir Computing, designed for research purposes and integration into deep architectures.

  • Project
    henry9zhr/QANet-pytorch

    A PyTorch implementation of QANet, a convolutional neural network architecture for question answering tasks.

  • Project
    alibaba/TorchEasyRec

    TorchEasyRec is a PyTorch-based recommendation system framework offering config-driven model definition, a wide range of pre-built models, distributed training, and production deployment capabilities for building large-scale recommendation systems at scale.