A novel technique which combines a speech waveform encoding process with multi-level modulation to realise a robust, spectrally efficient land mobile radio channel is described. The speech coding method is based upon a variant of adaptive bit allocation sub-band coding in which samples of the most energetic of eleven sub-bands are transmitted as discrete, multi-level symbols, together with some binary and additional multi-level symbols representing the overheads needed for reconstruction. Using this method, excellent audio quality has been achieved at symbol rates lower than 3 kS/sec under both simulated and real land mobile channel conditions. Description of how the method was used to provide a 4:1 spectrum efficiency over a 25 kHz spaced radio channel using TDMA in conjunction with multi-level (constant envelope) FSK, or more than 10:1 using linear modulation, are presented. At 4:1 with FM, the speech quality under strong signal conditions is very similar to standard FM and the coverage about 3 dB better for the same channel power.
展开▼
