Error-Resistant Narrowband Voice Encoder.
Abstract
One of the major causes of speech quality degradation in digital voice communication is bit errors introduced by the transmission channel. Until now we did not have an effective way to combat bit errors to improve tactical voice communication. For example, a more robust voice terminal that requires a wider transmission bandwidth would not be helpful to tactical communicators because most of them must rely on narrowband channels. Even if some of them may have access to wideband channels, the limited platform space does not allow them to carry both the narrowband voice terminal based on the Government-standard voice algorithm (that would interoperate with all narrowband users) and an additional more robust terminal. To circumvent this difficulty, we have developed an approach in which voice information is initially encoded at a low data rate (i.e. 800 b/s) and then redundancies for error protection are added to a bit rate that is compatible with transmission over narrowband channels. The necessary software can be integrated into the current narrowband voice terminal so that narrowband communicators have the option of either using the Government-standard 2400-b/s linear predictive coder (LPC) or the more robust voice algorithm. According to our simulation using a high-frequency channel, the new algorithm has a 4 dB advantage over the 2400-b/s system in terms of signal energy to noise density ratio when the bit error rate is 1 to 2%. In essence, the performance of the 800-b/s system is that of the 2400-b/s system with 2.5 times more received signal energy.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 26, 1986
- Accession Number
- ADA177441
Entities
People
- G. S. Kang
- W. M. Jewett
Organizations
- United States Naval Research Laboratory