GSM SPEECH PROCESSING ECE 2526 MOBILE COMMUNICATION Wednesday, 18 - PowerPoint PPT Presentation

GSM SPEECH PROCESSING ECE 2526 – MOBILE COMMUNICATION Wednesday, 18 March 2020 1

BASIC SPEECH PROCESSING FUNDAMENTALS (1) Basic GSM Band 890 + n*0.2 Basic GSM Band 2 890 + n*0.2 +45

GSM SPEECH PROCESSING FUNDAMENTALS -REVISITED 1. A GSM frame is 4.61msec long. 2. Each time slot is therefore 4.61/8 = 577µsec 3

SPEECH PROCESSING IN GSM Data Burst at 12 Kbps Adaptive Filter Removes redundant Parts in each data stream sample 4

QPSK MODULATION USED IN GSM 1. Each byte of data is split into four pairs of bits. 2. Each pair of bits is assigned a particular phase for I(t) and Q(t). 3. The modulated signal is the sum of the two sinusoidal waves, with a phase shift depending on the logic symbol. 5

WHY COMPRESSION IS NECESSARY IN GSM CODEC? 1. If speech were digitised in a linear fashion it would require a high data rate that would occupy a very wide bandwidth. 2. It is necessary to compress the data to send it through the available channel. 3. Speech is captured at a high enough sample rate and resolution to allow clear reproduction of the original sound. 4. It must then be compressed in such a way as to maintain the fidelity of the audio over a limited bit rate, error-prone wireless transmission channel. 5. At the receiver, the compressed data is expanded to regenerate the audio. 6

GSM CODEC RATES Bit rate Codec name (kbps) 1. Full rate 13 2. Enhanced Full Rate (EFR) 12.2 3. Half rate 5.6 4. Adaptive Multirate (AMR ) 12.2 - 4.75 5. Adaptive Multirate Wideband (AMR-WB) – 23.85 - 6.60 baseband is enhanced to the range 50 – 7,000 Hz. 7

SALIENT FEATURES OF ADAPTIVE MULTIRATE (AMR) 1. AMR uses 20ms frames . 2. Discontinuous transmission is employed so that when there is no speech activity the transmission is cut. 3. Voice Activity Detection (VAD) is used to indicate when there is only background noise and no speech. 4. Comfort Noise Generator (CNG) is added at the receiver to provide some background noise, even when no speech data is being transmitted. This provides the feedback for the user that the connection is still present. 8

WHAT IS TRANSMITTED FOR VOICE IN GSM? 20ms of 104 Kbps stream = 2080 bits Best coefficients are selected get the best approximation of the 20mS sample Low order coefficients are coded as 6-bit while Higher order are 9 coded as 3-bit