The USDC Codec
A vector sum excited linear predictive coder (VSELP)is used by U.S. digital cellular system (IS-54). VSELP coder operates at a raw data rate of 7950 bits/s and a total data rate of 13 kbps after channel coding. The VSELP speech coder is a variant of the CELP type Vocoders
Why this codec is designed?
- Highest speech quality,
- Modest computational complexity,
- Robustness to channel errors.
- Reduces the time required for the optimum code word search.
Block diagram of a VSELP encoder
- The 8kbps VSELP codec utilizes three excitation sources.(Figure 15.9)
- One is from the long-term ("pitch") predictor state, or adaptive code book.
- The second and third sources are from the two VSELP excitation code books.
- Each of these VSELP code books contains the equivalent of 128 vectors.
- These three excitation sequences are multiplied by their corresponding gain terms and summed to give the combined excitation sequence.
- After each sub frame the combined excitation sequence is used to update the long-term filter state (adaptive code book).
- The synthesis filter is a direct form 10th order LPC all pole-filter.
- The LPC coefficients are coded once per 20 ms frame and updated in each 5 ms sub frame. The number of samples in a sub frame is 40 at an 8 kHz sampling rate.
Block Diagram of VSELP Decoder (Fig 15.10):