Speech is a rich biometric signal that contains information about the identity, gender and emotional state of the speaker. In this work, we explore its potential to generate face images of a speaker by conditioning a Generative Adversarial Network (GAN) with raw speech input. We propose a deep neural network that is trained from scratch in an end-to-end fashion, generating a face directly from the raw speech waveform without any additional identity information (e.g reference image or one-hot encoding). Our model is trained in a self-supervised fashion by exploiting the audio and visual signals naturally aligned in videos. With the purpose of training from video data, we present a novel dataset collected for this work, with high-quality videos of ten youtubers with notable expressiveness in both the speech and visual signals.
We used this project as baseline.
- Python 2.7
- PyTorch
This implementation only supports running with GPUs.
Although having initially trained with 10 different identities, we can only publish the dataset in relation to those who have answered our request for working with their images and voice.
If you want to create your own dataset, it works as follows:
-
Run
get_data.py
. It requires the path of a .csv file with the following information: the youtube urlChannel-URL
, the name of the youtuberName
and the genderGender
. The output dataset will consist on a folder for each different youtuber specified in the .csv file as well as two folders (video/audio) in each of them containing the full video/audio downloaded. -
Run
preprocess_audio_frames.py
for audio preprocessing purposes. -
Remove any file whose name does not finish like
preprocessing_wav
in each audio folder. -
Run
face_detector.py
specifying the dataset path, a confidence threshold needed to compute the bounding boxes and the path of the haarcascade pretrained classifier (included in this repo in/assets
) -
The useful audio snippets will be stored in
DATASET-PATH/audio/NAME-OF-THE-DOWNLOADED-VIDEO_frames
. The useful youtuber faces will be stored inDATASET-PATH/video/NAME-OF-THE-DOWNLOADED-VIDEO_cropped_frames
Once the dataset is build, you can store the paths for all the images/audio frames in a pickle file with:
scripts/generate_pickle.py
Include both pickle files (train/test datasets) in:
config.yaml
python runtime.py
Arguments:
lr_D
: The learning rate of the disciminator. default =0.0004
lr_G
: The learning rate of the generator. default =0.0001
vis_screen
: The visdom env name for visualization. default =gan
save_path
: Name of the directory (inside checkpoints) where the parameters of them odel will be stored.l1_coef
: L1 loss coefficient in the generator loss fucntion. default=50
l2_coef
: Feature matching coefficient in the generator loss fucntion. default=100
pre_trained_disc
: Discriminator pre-tranined model path used for intializing training.pre_trained_gen
: Generator pre-tranined model path used for intializing training.batch_size
: Batch size. default=64
num_workers
: Number of dataloader workers used for fetching data. default =8
epochs
: Number of training epochs. default=200
softmax_coef
: Paramete for the scale of the loss of the classifier on top of the embeddingimage_size
: Number of pixels per dimension. They are assumed to be squared. Two possible values:64 | 128
. default =64
inference
: Boolean for choosing whether train or test. default =False
If the code of this repository was useful for your research, please cite our work:
@inproceedings{wav2pix2019icassp,
title={Wav2Pix: Speech-conditioned Face Generation
using Generative Adversarial Networks},
author={Amanda Duarte, Francisco Roldan, Miquel Tubau, Janna Escur,
Santiago Pascual, Amaia Salvador, Eva Mohedano, Kevin McGuinness,
Jordi Torres and Xavier Giro-i-Nieto},
booktitle={2019 IEEE International Conference on Acoustics, Speech
and Signal Processing (ICASSP)},
year={2019},
organization={IEEE}
}