Rambo Lan

@LiYijin It's English.

@zzw922cn regarding this, I also have a question. The num_classes for 'cha' is 29. But why it's 29 not 28? According to my understanding, we only have 26 letters +...

@euler16 According to the author's blog, the original paper is [Deep Recurrent Convolutional Neural Network: Improving Performance For Speech Recognition](https://arxiv.org/pdf/1611.07174.pdf)

@duxiansheng I use sndfile converter to convert NIST to WAV. Please refer to [this repo](https://github.com/erikd/libsndfile) and their [home page](http://www.mega-nerd.com/libsndfile/). The TIMIT files you downloaded maybe ended with .wav, but they...

I met exactly the same problem.

> Thanks for your great work！ > > When I try to train maptrv2 with vovnet99 as backbone,the grad_norm is always nan. Reducing the learning rate and applying gradient clipping...

> @ChenccLOL @brianlan Did you try using pretrained weights from https://github.com/TRI-ML/dd3d ? > > GeMap provides config file for using V2-99：https://github.com/cnzzx/GeMap Oh, that's good to know. Thanks for the information,...

> I think you're encountering a common issue that isn't a flaw exactly. The current chunking approach in this repo splits Text into 1000 characters with 0 overlap by splitting...