Skip to main content
EURASIP Journal on Audio, Speech, and Music Processing
Figure 3 | EURASIP Journal on Audio, Speech, and Music Processing

Figure 3

From: Adaptive Long-Term Coding of LSF Parameters Trajectories for Large-Delay/Very- to Ultra-Low Bit-Rate Speech Coding

Figure 3

Sentence " Elderly people are often excluded " from the TIMIT database, pronounced by a female speaker. (a) The speech signal; the n th voiced/unvoiced section is denoted V/U n; the total number of voiced (resp., unvoiced) LSF vectors is 79 (resp., 29). The vertical lines define the V/U boundaries given by the TIMIT label files. (b) LSF trajectories; solid line: original LSF vectors; dotted line: LT-coded LSF vectors with = 2.1 dB for the voiced sections (r = 22 bits/vectors) and = 1.9 dB for the unvoiced sections (r = 20 bits/vectors) (see the text). The vertical lines define the V/U boundaries between analysis frames, that is, the limits between LT-coded sections (the analysis frame is 30 ms long with a 20 ms hop size).

Back to article page