Introduction to Seminar Idea on Speech Coding:
The speech coding is by done by converting sound waves of our voices into electric signal. The process is carried out by recording the voice by a microphone. A microphone consist of diaphragm, magnet and a coil of wire when we speak the voice waves vibrate the diaphragm which is connected to magnet this cause diaphragm to move on the coil.
While the magnet is moving a electrical current is created the whole process stored at one place and it comes out as sound. Analog signal is used by microphone. For conversion of analog signal into digital signal DAC is used. The analog signal is first passed over band pass filter. The band pass filter filters the signal between 3000Hz to 3KHZ.
The band pass filter converts analog signal to digital and as well as digital to analog. The analog digital converter measures the analog signal from microphone 8000 times/second. The ADC takes input for every 125 mille seconds. The sample is quantified with 13-block data. The data rate of 104KBPS is too large to handle so it is to be encoded by speech encoder to reduce bit rate.
Linear predictive coding:
In LPC 13-bit coding are saved in memory.160 of the audio samples make a audio of 20ms this contains 2080 bits.LPC analyzes this 20ms and determines 8 coefficients for filtering.LPC identifies the voice of humans.Lpc encode reduces 20ms from 2080 bits of audio data.
The audio block of 20ms is transferred to the block coding. They are splatted for reconstruction based on importance. Class 1 bits are very important and they are further divided into IA and Ib. IA are protected by cyclic code Ia and Ib are combined with 4 additional class bits to tail
In this coding errors are detected and corrected according to limit. Every bit is twice of class 1 entries. The coding uses of 5 consecutive bits to calculate redundancy bits. Class 2 bits are not allowed in convolutional coding.
Download ECE Seminar Idea on Speech Coding.