On quantizer dimensions in joint speech/channel coding

In mobile speech communication usually vector quantization (VQ) is employed to ensure high coding efficiency. Convolutional coding and softbit speech decoding can add a considerable amount of robustness. VQ as well as softbit decoding however can be a quite complex task. Under the constraint of a constant gross bit rate and clean channel quality we propose the use of lower dimensional VQ or even scalar quantization (SQ) with a higher bit rate which leaves then fewer redundancy to be added by channel coding. This concept of joint speech/channel coding with its suboptimal speech coder and weaker channel coder can efficiently employ softbit speech decoding yielding a low overall complexity transmission scheme. Cases are shown where its performance is even better as compared to a high dimensional VQ with softbit decoding.