 Welcome to the session of introduction to the audio compression. These are the learning outcome of the session. I would like to ask the question, what is the sound? You may pause this video, think about this question and write down your answer in your notebook. Return to the video, then resume the video and see the answer. Let me answer the question. The sound is a physical phenomenon produced by the vibration of the matter such as violent string or the block of the wood. As the matter vibrates, pressure and variations are created in the air which is surrounding to it. The high and low pressure alternation is propagated through the air like the wave motion. Now when this wave reaches to the ear of the human, the sound is heard. The basics of the sound. A sound is produced by the vibration of the matter. During the vibration, the pressure variations are created in the air surrounding to it. The pattern of the oscillation is called as a waveform. The waveform repeats itself after the particular time of the interval which is called as a one period. Let us move to the frequency. The frequency of the sound is reciprocal value of the period. It represents the number of periods in second and it is measured in hertz or cycle per second. Kilohertz is used in order to indicate thousands of oscillations per second. That means 1 kilohertz will be equal to 1000 hertz. The frequency ranges are divided into infrared sound. It varies from 0 to 20 hertz. The human hearing frequency range is from 20 hertz to 20 kilohertz. The ultrasound ranges from 20 kilohertz to 1 gigahertz and the hypersound ranges from 1 gigahertz to 10 terahertz. A speech is an audio signal which is produced by the human or the music signals which have the frequency ranges in between 20 hertz to 20 kilohertz. The amplitude of the sound. The amplitude of the sound is also called as a loudness. The amplitude of the sound is major of the displacement of the air pressure waveform from its mean value. The digital sound. A digital audio data is actual representation of the sound which is stored in the form of samples. Samples represent the amplitude of the sound at discrete points in the time. The quality of the digital recording depends on the sampling rate. That is the number of samples taken per second. The PCM technique which is used for the audio compression. In pulse code modulation technique, the analog input signal information is converted into the binary coded sequence. That is once and zeros. The output of the PCM will be the binary coded sequence as shown in the figure. There are three main steps in the PCM technique. The first is the sampling, then quantizing and then encoding. Sampling is nothing but the conversion of the analog signal into the discrete samples. The quantization is nothing but the rounding of the samples to the nearest values. And then encoding is nothing but the each quantizer samples are coded into the binary data. So, let us move to the sampling. The sampling is a periodic measure of the analog signal. And it changed the continuous time signal into the discrete time signal. After the sampling, the signal value is known only at a discrete point in the time. These points are called as a sampling instant. Therefore, continuous time signal curve can be described by the sample values. The sampled values. The value of the function at sampling point is nothing but the sampled values. The sampling interval. The time that separate the two sampling points is nothing but the sampling interval. As per the sampling theorem, the band limited signal can be reconstructed exactly if the sample it is sampled at the rate at least twice the maximum frequency component which is present in it. That means fs should be greater than 2b that the b defined the band limited signal bandwidth. So, we know that the ts that is time period is reciprocal of the frequency. Hence, sampling interval will be 1 upon 2b. Let me ask you the question. The narrow band speech signal is limited to 0 to 4 kilohertz. What should be the sampling rate and the sampling interval? You may pause the video, think about this question and write your answer in a notebook. Let me answer the question. The sampling rate is 8 kilohertz because as per the sampling theorem, the sampling rate should be twice than that of the bandwidth of the baseband signal. So, if the 8 kilohertz is a sampling rate, then the difference between the two consecutive sample will be 0.125 millisecond. Quantization. Each accurate sample value is rounded off to the closest numerical value in the given numerical set. In quantizing process, the information in accurate signal value is lost because of the rounding of the and original signal cannot be reproduced exactly anymore. The more quantization steps can be used for the better performance. For the binary coded, the number of quantum levels are defined as q is equal to 2 raised to n, where the q denotes the number of quantum levels and n is a length in the beats of the binary words that describe the sample values. The difference between the quantization amplitude and the actual amplitude is called as a quantization error. So, more the quantization step, there will be the lower quantization error. Coding. The coding process will convert the discrete amplitude signal into the series of binary beats for the transmission and the storage. As shown in figure, the amplitude space is evenly divided into 6 different quantization step. Thus, 3-bit binary codes can be used in order to convert this analog signal into the discrete signal. So, for the first sample, the quantized amplitude is 0 and it is coded beats into the 1 0 0. For the second sample, the quantized amplitude is 2 and then it is coded into 0 1 0. The audio formats, they are basically two audio formats. The voice quality audio format and the CD quality audio format. The voice quality audio format. In this audio format, the 8-bit encoded quantization level is used and the sampling rate is 8000 hertz. This is considered as a fast and accurate enough for the telephone quality speech input, CD quality format. In this, the 16-bit linear pulse coded modulation technique is used for the quantization and the sampling rate is 44100 hertz. This is a block diagram of the PCM encoder and decoder. So, these are the steps. The low pass filter, this filter eliminates the higher frequency component which is present at the input analog signal. The sampler, this is a technique which helps to collect the sample data at the instantaneous value of the message signal. So, in order to reconstruct the original signal, the sampling rate must be greater than twice the highest frequency component of the message signal in accordance with the sampling theorem. Quantizer, the quantizing process is used in order to reduce the excessive beats and confining the data. The sampler output when given to the quantizer reduces the redundant beats and compresses the value. Encoder, the digitization of analog signal is done by the encoder. It designates the each quantized value by the binary code. Regenerative repeater, this section increases the signal strength. The output of the channel also has a one regenerative repeater circuit to compensate the signal loss and reconstruct the signal and also to increase its strength. Decoder, the decoder circuit decodes the pulse coded waveform to reproduce the original signal. This circuit act as a demodulator. After the digital to analog conversion, which is done by the regenerative circuit and the decoder, the low pass filter is employed which is called as a reconstruction filter. Pulse code modulator circuit digitizes the given analog signal, cores it and samples it and then transmit it in the analog form. This whole process is repeated in a reverse pattern to obtain the original signal. DPCM technique, a differential pulse code modulation technique is a derivative standard of the pulse coded modulation technique. This technique samples the analog signal, then quantizes the difference between the sampled value and its predicted value, then encode the signal to form the digital value. For the most of the audio signal, the range of differences in amplitude between the successive sample of the audio signal is less than the range of actual sample amplitude. The sample of the signal are highly correlated with each other. The signal's value from present sample to the next sample does not differ by large amount. The adjacent samples of the signal carry the same information with the small differences. Hence, if the difference between the value is digitized and encoded, then the fewer bits are required than the pulse coded modulator signal at the same sampling rate. So, principle of the differential pulse code modulation is nothing but if the redundancy is reduced, then overall bit rate will be decreases and the number of bits required to transmit the sample will be also reduced. This type of the digital pulse modulation technique called as a differential pulse code modulation technique. A DPCM works on the principle of the prediction. The value of the present sample is predicted from the previous sample. The prediction may not be exact, but it is very close to the actual sample value. A DPCM encoder is shown in a figure. The PCM coded mesh signal is given to the substractor. Basically, a predicted value is stored into the register. So, the PCM sample and the predicted samples are subtracted and then converted by the parallel to serial conversion for the transmission. This subtracted value is also added with the previous value of the register in order to update the value of the register. At the decoder side, the data will be arranged to serial to parallel converter. So, this will convert that serial data into the parallel and then it is added with the predicted value of the register and then this added value is decoded by the PCM decoded decoder and we will get the audio mesh signal. So, what is the difference between the PCM and DPCM audio compression? The number of bits which is required for the PCM uses 4, 8 or 16 bit per sample, where in case of the differential pulse code modulation less than that of the PCM bits are required. Step size. The fixed step size is used in this PCM modulation technique. Here, the fixed number of the levels are used. Redundant bits are present. It can be removed permanently in case of the DPCM, quantization error and the distortion. It depends on the number of levels which are used and here in case of the DPCM, slope overloaded distortion and quantization noises are present, but very less as compared to the PCM. The bandwidth, the higher bandwidth is needed for the PCM and for the DPCM a lower bandwidth is needed. So, these are the references which are used for the session. Thank you.