Janvijay Singh

icon indicating a website link https://iamjanvijay.github.io icon indicating an email [email protected]

icon company Georgia Tech icon location Atlanta, GA icon location CS Grad Student at Georgia Tech; ex-Data & Applied Scientist @Flipkart; Natural Language Processing; Speech Recognition and Synthesis;

Results 3 comments of Janvijay Singh

torch. rfft is not the right implementation of DCT

Hi @TianhaoFu! Please have a look at this comment: https://github.com/zh217/torch-dct/issues/15#issuecomment-851439294 I have verified this and it works well for torch==1.10. Thanks.

RNN-Transducer Loss support

@will-rice I have an implementation of RNNT ([link to the repository](https://github.com/iamjanvijay/rnnt)) using TensorFlow ops. We can work on this together, if you are interested?

RNN-Transducer Loss support

@will-rice Yes. That can be a good way to start then we can do iterations over it. What do you suggest?