pytorch-kaldi is a project for developing state-of-the-art DNN/RNN hybrid speech recognition systems. The DNN part is managed by pytorch, while feature extraction, label computation, and decoding are performed with the kaldi toolkit.

SincNet is a neural architecture for efficiently processing raw audio samples.

The SpeechBrain project aims to build a novel speech toolkit fully based on PyTorch. With SpeechBrain users can easily create speech processing systems, ranging from speech recognition (both HMM/DNN and end-to-end), speaker recognition, speech enhancement, speech separation, multi-microphone speech processing, and many others.

End-to-End speech recognition implementation base on TensorFlow (CTC, Attention, and MTL training)

The DARPA TIMIT Acoustic-Phonetic Continuous Speech Corpus.

CTC end -to-end ASR for timit and 863 corpus.

MXNet implementation of RNN Transducer (Graves 2012): Sequence Transduction with Recurrent Neural Networks

Tensorflow implementation of "Listen, Attend and Spell" authored by William Chan. This project utilizes input pipeline and estimator API of Tensorflow, which makes the training and evaluation truly end-to-end.

Keras (tensorflow) implementation of SincNet (Mirco Ravanelli, Yoshua Bengio - https://github.com/mravanelli/SincNet)

Python implementation of pre-processing for End-to-End speech recognition

Speech recognition on the TIMIT (or any other) dataset

This code implements a basic MLP for speech recognition. The MLP is trained with pytorch, while feature extraction, alignments, and decoding are performed with Kaldi. The current implementation supports dropout and batch normalization. An example for phoneme recognition using the standard TIMIT dataset is provided.

Pytorch based phoneme recognition (TIMIT phoneme classification)

THEANO-KALDI-RNNs is a project implementing various Recurrent Neural Networks (RNNs) for RNN-HMM speech recognition. The Theano Code is coupled with the Kaldi decoder.

A Simple Automatic Speech Recognition (ASR) Model in Tensorflow, which only needs to focus on Deep Neural Network. It's easy to test popular cells (most are LSTM and its variants) and models (unidirectioanl RNN, bidirectional RNN, ResNet and so on). Moreover, you are welcome to play with self-defined cells or models.

Attention-based end-to-end ASR on TIMIT in PyTorch

Extract mfcc vectors and phones from TIMIT dataset

Sum-Product Networks (SPNs) for Robust Automatic Speaker Identification.

A database of clean and noisy speech for audio research

Python/numpy/pandas convenience wrapper for the TIMIT database.