High-Rate Optimized Quantization Structures And Speaker- Dependent Wideband Speech Coding