ETSI Standard Vocoders
ETSI EFR-GSM
The algorithm implemented is the
ETSI Enhanced Full-Rate (EFR) GSM recommendation, digital mobile telephone
system. The encoder compresses linear-PCM (Pulse Code Modulated), speech input
data, at a sample rate of 8kHz, to 12 200 bps. The encoder input originates
from either a Mobile Station or a converted Public Switched Telephone Network
(PSTN). The decoder input arrives via a channel coding function as defined
by GSM 05.03 [3]. This adds additional flags and bad frame indicators. The
speech decoder converts the data back to 160 speech samples of 16-bit uniform
format.
The EFR-GSM algorithm implements
silence compression techniques to reduce the transmitted bit rate during the
silent intervals of speech. Systems allowing discontinuous transmission (DTX)
are based on Voice Activity Detection (VAD) algorithms and Comfort Noise Generator
(CNG) algorithms that allows the insertion of Silence Insertion Descriptor
(SID) frames during the silence intervals. This also provides the additional
advantage of using lower processing loads and DSP bandwidth resource during
silence frames.
The coding scheme is the Algebraic
Code Excited Linear Prediction Coder (ACELP). An excitation signal at the
input of a short term linear predictor (LP) synthesis filter is constructed
by adding two excitation vectors from adaptive and fixed codebooks. Integrated
IS-641-A and AMR-GSM functionality is also available.
back to menu