Skip to content

Acoustic-to-Articulatory Inversion with Ultrasound Tongue Image target

License

Notifications You must be signed in to change notification settings

BME-SmartLab/speech2uti-ICJ

Repository files navigation

speech2uti-ICJ

Keras implementation of Csapó T.G. & Sepúlveda-Sepúlveda A., ,,Ultrasound Tongue Image Synthesis for Acoustic-to-Articulatory Inversion using Convolutional and Recurrent Neural Networks'', submitted to Infocommunications Journal, 2022.

UltraSound data

private, but can be sent upon request to the first author.

Trained models

Trained models, predicted videos : https://simonyi-my.sharepoint.com/:f:/g/personal/csapszi_sch_bme_hu/EgIc-DOVUmJBjJ-DSTTgsHEBxVroMwU4qNC1bbyfI6TUVQ?e=C42y2Z

Demonstration samples

Ultrasound video samples: http://smartlab.tmit.bme.hu/UTI-inversion

Requirements

  • Python 3.6.9
  • ... see requirements.txt

About

Acoustic-to-Articulatory Inversion with Ultrasound Tongue Image target

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages