Digital Speech Processing
Generation of Test Material for Voice Communication Systems
ARCON has developed and published procedures for the recording of source materials in acoustic noise for the purpose of testing and evaluating voice communication systems . The Digital Speech processing lab maintains 4 acoustic isolation rooms that allow audio recordings in environments ranging from an extreme quiet below 20dBa SPL to well above the 110 db SPLc encountered in some of the harshest military environments.
Synchronous multi-channel digital recordings are produced direct to hard disk at sampling rates up to 96kHz and resolutions up to 24 bits. Editing also is performed digitally as well as any other desired post processing functions such as down sampling, windowing, and gain equalization.
ARCON maintains contact with a large population of both experienced and naive talker/communicators across the adult age range in order to build balanced corpora that can reveal any talker dependencies within a communication system.
Custom laptop-based software presents talkers with speech material for any given Subjective Test Method in a precisely timed format in order to maintain consistency.
More details on Corpus Collection can be found at this link.