Apply to become a certified provider
ReSSInt-EMG (Spanish EMG and Speech Database)
ReSSInt-EMG (Spanish EMG and Speech Database) has been generated in the framework of the ReSSInt project (Voice restoration with silent EMG speech interfaces) and its continuation project DeepRestore (Deep learning approaches for speech restoration from face movement biosignals), coordinated research projects funded by the Spanish Ministry of Science and Innovation, aiming to investigate the use of silent speech interfaces to restore communication in people who have been deprived of the ability to speak. Silent speech interfaces (SSIs) are devices designed to capture non-acoustic biological signals produced during the speech production process and utilize them to predict the intended message. While SSIs have traditionally been explored primarily within the realm of speech recognition, specifically Silent-Speech-to-Text applications, the ReSSint project takes a distinctive approach by concentrating on direct speech synthesis techniques. This involves the direct generation of the speech waveform corresponding to the captured biosignals. The non-acoustic biosignals that are used in this work are EMG signals or, more specifically, surface (i.e., non-invasive) EMG (Electromyography). Electromyography is a technique used to measure and record the electrical activity of muscles. When a muscle is active, it produces an electrical signal, called an action potential that can be detected by an electrode placed on the skin over the muscle. Since this project focuses on speech, muscles in the face and the neck are targeted. The database comprises 22.55 hours of data, recorded by 9 Spanish-native speakers (5 males, 4 females). EMG signals from sensors were located in the speaker’s face and the audio part was recorded in a soundproof recording cubicle with a Neuman microphone. Speech files were recorded in WAV format, 16kHz, 16-bit, Linear PCM (Lo-hi, signed integer), mono (1 channel).
ELRA