-
I observed a interesting behaviour after 138K iters where discriminator dominated the training and generator exploded in both train and validation losses. Do you have any idea why and how to prevent …
-
data = pd.read_csv('50_speakers_audio_data/dataset4.csv')
data.head(2513)
valid_data = data[['Name','Full Path','Size','ClasseID','Fold']]
valid_data.shape
valid_data['path'] = valid_dat…
-
Any plans to do this?
-
Branch: Master ([540d811dd5](https://github.com/mozilla/TTS/tree/540d811dd52b5598a7cd21cbbcf197b0bfbeab62))
Hi,
trying to train a multi-speaker model using the current master branch with phonem…
-
## 🐛 Not necessarily a Bug
Not exactly a bug, but I think it makes more sense if the triangular Mel filterbanks are area-normalized, as opposed to the current behavior where they are height-normali…
-
Any plans for integrating it for tensorflow-2.0 ?
Currently here's an issue i faced when using kapre with tf 2.0
```
Python 3.6.8 |Anaconda, Inc.| (default, Feb 21 2019, 18:30:04) [MSC v.1916 6…
-
Hi,
I am trying to convert a pre-trained tensorflow model for running on AWS Inferentia. The neuron compiler successfully converts the model, but gives warnings about failing to fuse subgraphs. The N…
-
Placeholder for issue related to integration of fastspeech with squeezewave
https://github.com/tianrengao/SqueezeWave
seems to quite faster than waveflow.
-
#### Description
As stated in the [documentation](https://librosa.github.io/librosa/generated/librosa.core.stft.html#librosa.core.stft) the default physical length should be 93ms. Using default value…
-
Hi there,
I want to make a mel-spectrogram ops for my custom Onsets-and-Frames models.
I want to make a mel-spectrogram that is can run on Android devices. But the results of mel-spectrogram on tfli…