Shivam Mehta

Hello! Thank you for your interest in our work! We tried something similar to have more control over the synthesised speech in one of our succeeding works which is submitted...

Hello! We have released [OverFlow](https://shivammehta25.github.io/OverFlow/). In [OverFlow](https://shivammehta25.github.io/OverFlow/) not only do we get better naturalness and more accurate pronunciations but we also show speaker adaptation in a low resource setting by...

Hello! I have mainly trained with the datasets containing female voices, different accents but the voices were coincidently female. As much as I have seen, you should not face any...

Hello, I have updated the repository with a [male voice](https://github.com/shivammehta007/Neural-HMM/releases/download/Neural-HMM/Neural-HMM-Male.ckpt) based on [RyanSpeech](https://arxiv.org/abs/2106.08468) corpus . Also, we have noticed it is easier to finetune this checkpoint on a male voice...

Hello, You have not provided files for the validation set. Either set the `validation_files` hyperparameter in `src/hparams.py` or you can turn off validation of your training by adding `limit_val_batches=0` in...

Hello, could you give more details about which line raised this issue? Another reason I could think of was during the plotting but if you have already set the validation_files...

So the solution for this is to copy the training file location and set `limit_val_batches=0`.

Hi! I am assuming this late you might not need this! But I am still writing it for the future in case someone else also encounters this! ```python x_s_sq_r =...

So the available checkpoint in the GitHub repository is also trained up to 300k steps?

I remember someone on [Reddit](https://www.reddit.com/r/MachineLearning/comments/12kjof5/comment/jg4zjqm/) organising a similar list, with the dates which can be useful for this https://docs.google.com/document/d/1sariO32u4a87vfZDzAR-fq2RwuZ_zxBj29vMG8UFH2s/edit#heading=h.y2rv92en8xpt