Intelligibility of the Adaptive Multi-Rate Speech Coder in Emergency-Response Environments.

摘要

Recent U.S. legislation has paved the way for the creation of the First Responder Network Authority, or FirstNet, to oversee the creation of a new nationwide interoperable broadband network supporting the work of public safety officials. It is anticipated that this broadband public safety network will employ long term evolution (LTE) radio technology, and that traditional telephony voice communications could be accomplished through voice over LTE (VoLTE). A popular choice for digital speech coding in VoLTE systems is the Adaptive Multi-Rate (AMR) speech coder. This speech coder delivers very good speech quality and very high speech intelligibility in traditional wireless telephony applications. A previous investigation confirmed field reports that some of the unique environmental conditions associated with firefighting can be problematic for digital speech coding algorithms. When high noise levels are present (e.g., from pumps, nozzles, saws, or alarms) the signal presented to a digital speech coder can deviate significantly from what the speech coder anticipates, and as a result output speech can become distorted. In extreme cases, the intelligibility of the output speech may suffer. This problem can be compounded when the mask associated with a self-contained breathing apparatus (SCBA) is used because a mask can further alter the speech signal. This may cause additional difficulties for the speech coder and associated distortion and reduction of intelligibility.