This is a TensorFlow implementation of the WaveNet generative neural network architecture for audio generation.
The WaveNet neural network architecture directly generates a raw audio waveform, showing excellent results in text-to-speech and general audio generation (see the DeepMind blog post and paper for details). The network models the conditional probability to generate the next sample in the audio waveform, given all previous samples and possibly additional parameters.
After an audio preprocessing step, the input waveform is quantized to a fixed integer range.
The integer amplitudes are then one-hot encoded to produce a tensor of shape A convolutional layer that only accesses the current and previous inputs then reduces the channel dimension. The core of the network is constructed as a stack of causal dilated layers, each of which is a dilated convolution (convolution with holes), which only accesses the current and past audio samples. The outputs of all layers are combined and extended back to the original number of channels by a series of dense postprocessing layers, followed by a softmax function to transform the outputs into a categorical distribution. The loss function is the cross-entropy between the output for each timestep and the input at the next timestep. In this repository, the network implementation can be found in wavenet.py. |
![]() |
TensorFlow needs to be installed before running the training script.
TensorFlow 0.10 and the current master
version are supported.
In addition, librosa must be installed for reading and writing audio.
To install required python packages:
sudo pip install -r requirements.txt
The VCTK corpus(around 10.4GB, Alternative host) is currently used.
In order to train the network, you need to download the corpus and unpack it in the same directory as the train.py
script.
Then, execute
python train.py
to train the network.
You can see documentation on the settings by by running
python train.py --help
You can find the hyperparameters of the network in wavenet_params.json
.
You can use the generate.py
script to generate audio using a previously trained model.
Run
python generate.py --samples 16000 model.ckpt-1000
where model.ckpt-1000
needs to be a previously saved model.
You can find these in the logdir
.
The generated waveform can be played back using TensorBoard.
Passing --wav_out_path
yields the output as a wav file.
python generate.py --wav_out_path=generated.wav --samples 16000 model.ckpt-1000
Passing --save_every
in addition to --wav_out_path
will save the in-progress wav file every n samples.
python generate.py --wav_out_path=generated.wav --save_every 2000 --samples 16000 model.ckpt-1000
For fast generation:
python generate.py --samples 16000 model.ckpt-1000 --fast_generation=True
For an explanation of fast generation, see Fast Wavenet.
Currently, there is no conditioning on extra information like the speaker ID.