Until recently, high quality speech recordings required that either a recording team visits the speakers, or that the speakers come to a recording studio. With speech recordings via the Internet, recordings can be performed in parallel, in high signal quality, and without any time-consuming and expensive travel.
The VOYS (Voices of Young Scots) project aims to create a speech database of 300 pupils from several dialectal regions of Scotland.
This database will be used to develop spoken language technology and to document regional variation in adolescent speech. It provides a large empirical basis for phonetic and linguistic studies. Due to the demographic data collected from each speaker it is also well-suited for sociolinguistic analyses.
The speech material consists of read speech, e.g. phonetically rich sentences, date and time expressions, numbers and digit strings, and spontaneous responses to questions, e.g. "What is your favourite computer game? What is it about?".
The recordings will be performed via the Internet in secondary schools. To achieve a consistent signal quality (22.05 kHz @ 16 bit) all recordings will be made with a standardized recording kit.
During a recording session, the client software displays prompts to the speaker, records the speech signal and uploads the recorded files to the server. The data is then immediately available for further processing.
VOYS is a successor project to the Ph@ttSessionz data collection in Germany with more than 1000 recorded speakers.