Large-scale unsupervised audio pre-training for video-to-speech synthesis

Triantafyllos Kefalas1, Yannis Panagakis2, Maja Pantic1

1Imperial College London

2 Department of Informatics and Telecommunications, University of Athens, Greece

For access to full test set samples, please contact tk15(at)ic(dot)ac(dot)uk

File name lists

Demo - Raw waveform models

GRID (4 speakers, seen)

GRID (33 speakers, seen)

GRID (33 speakers, unseen)

LRW

TCD-TIMIT (3 lipspeakers, seen)

Demo - Acoustic features models

GRID (4 speakers, seen)

GRID (33 speakers, seen)

GRID (33 speakers, unseen)

LRW

TCD-TIMIT (3 lipspeakers, seen)

File name lists

https://drive.google.com/drive/folders/1C7VJrc2vZ7ZAHiQsio_bn1yBmjlVl7QV?usp=drive_link

Demo - Raw waveform models

GRID (4 speakers, seen)

s29_swir8p.mp4

s1_bbal9a.mp4

GRID (33 speakers, seen)

s10_bbii1a.mp4

s20_lbai2s.mp4

GRID (33 speakers, unseen)

s4_bbar7n.mp4

s19_bgie8n.mp4

LRW

CONSERVATIVE_00006.mp4

AMERICA_00010.mp4

DEFICIT_00017.mp4

GREECE_00037.mp4

TCD-TIMIT (3 lipspeakers, seen)

Lipspkr1_si1881.mp4

Lipspkr2_sx385.mp4

Demo - Acoustic features models

GRID (4 speakers, seen)

s29_pbbt1a.mp4

s1_lrbe6n.mp4

GRID (33 speakers, seen)

s31_lwwp4n.mp4

s13_lbwqzn.mp4

GRID (33 speakers, unseen)

s19_bbax7s.mp4

s11_srwbzp.mp4

LRW

HOUSING_00002.mp4

DECISION_00030.mp4

ANNOUNCED_00027.mp4

AFTERNOON_00001.mp4

TCD-TIMIT (3 lipspeakers, seen)

Lipspkr2_si1105.mp4

Lipspkr2_sx334.mp4

Page updated

Google Sites

Report abuse