lecture11

lecture11 - Lecture Outline More on Quantization Reading:...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
Lecture Outline More on Quantization Reading: Textbook sections 6.5.2, 6.6.1 and 6.6.3. Material covered: 1. Uniform quantizer revisited. Input range ] , [ max max x x - , divided into N levels with v N 2 = , 1 max 2 - = v x If we define max x X X = , then we have ] [ 4 3 2 X E SQNR v × × = Or 8 . 4 02 . 6 ] [ log 10 2 10 + + = v X E SQNR dB If X is uniformly distributed in ] , [ max max x x - , we have v SQNR 02 . 6 = dB. Adding an extra bit for quantization for each sample improves performance by 6 dB. 2. Non-uniform PCM If the input to quantizer does not have a uniform distribution, we can reduce SQNR if we use non-uniform quantization, i.e. are not the same for the regions i R . Take speech for example, lower amplitude levels are more likely than higher amplitudes. Therefore, we can use more quantization levels for lower amplitudes. Also, our perception of audio signal is highly nonlinear. To reduce the number of bits needed, we use a compander (i.e. compressor/uniform
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 3

lecture11 - Lecture Outline More on Quantization Reading:...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online