You are on page 1of 2
554 (1) = €(80 + x90 - D But y(k= 1) = (k= 1) +(e 2) and = on. a ‘Therefore, eq() Hegk- D+ Hk -2 et) or eyo + ee 1+ eq(k-2) + e(k-3) (0) fy original quantized signal x4) by re cant Thus, wo have proved that Ky) =Xk=W) eauffietent for reconstruction of signal with a Te easily generated by 1 simple dolta ‘This equation shows that we can reconstruct the adding the prediction errors at vario transmission of encoded error sign: reasonable accuracy. Such prediction errors ean modulator system. L 41.36. VOICE CODERS (VOCODERS) ‘As a matter of fact, in our day to day life, we liste Fig. 11.44, en to different persons speaking the same English language in dig, styles, And still we can recognize what is being spoken. Thus, to transmit speech, we need not transmit the pris Siveform generated by the speaker. Rather, we can transmit the ‘information’ from which @ waveform can, WRonstructed at the receiver which is ‘similar’ to and not ‘identical’ to the waveform generated by the speaker. 7, Speech coders or vocoders are used for the digital coding of speech. They are used for voice simulation. The advantage using the vocoders is that they operate at very Jow bit rates ranging from 1.2 to 2.4 kbits/sec. 11.36.1. Voice Model (Speech Model) A voice model or speech model shown in figure 11.45 is used to simulate voice. The operation of a voice model js dependent on the fact that speech consists of a sequence of ‘voiced’ and ‘unvoiced’ sounds. ‘Therefore, we can approximate it by passing a sequence of voiced and unvoiced sounds through a filter. The tern ‘voiced sounds’ is defined as the sounds which are Generated by the vibrations of the vocal cords. The ‘unvoiced’ sound: are produced at the time of pronounciation of letters such as ‘s', ‘:p’ or ‘f. These sounds are produced by expelling air through lips and teeth. A speech model has been shown in figure 11.45. The speech model shown in figure 11.45 simulates the voiced sounds by an impulse generator whose Frequency control to frequency is equal to. the accomodate ty pich of sound fundamental frequency of the vocal cords. The noise source simulates the unvoiced sounds. yee ee or | Notced sounds ‘The switch position determines Ae ora whether the sound is voiced or wnt unvoiced. The selected voiced or ‘After anosraet unvoiced sound is then passed 7 representing F—> speech through a filter, which simulates deterring ty wher eee tae I waver the effect of the mouth, throat Tose source sounds is voicadunvoiced and nasal passage of the speaker forumvoced He : on the generated sound. The sounds _| Unvoiced sounds input signal is filtered by this Fig. 14.45, Speech model used in vocoders filter in such a way that the required letter fe prenenzied All the vocoders employ the scheme shown in figure 11.45 for generation ofa syntheszel approximation of speech. A voice coder or vocoder consists of voi¢ der i iscuss thei? prota ta Gt oe ice encoder and voice decoder. Let us dis 11.36.2. Voice Encoder Figure 11.46 shows the block diagram of a voice encoder. Working Operation ‘The working operation of the voice encoder may be described as under: The speech frequency spectrum (200 Hz - 3200 Ha) is divided f i A e into 15 frequen ing 15 bandpass Bach range is of bandwidth 200 Hz (200 to 400, 400 to 600 ete.). Output of each bandpass filter ie rectified and ited to produce a de voltage which is proportional to the amplitude of AC signal at filter output. The input spec? SE™ also applied to a frequency discriminator followed by a 20 Hz low pass filter. The output of this filter is a d¢ "ty, which is proportional to the voice frequency. This frequency represents the pitch of the voice. The de voltae® ng output of the discriminator-filter combination can also indicate whether the speech is voiced or unvoiced: ™ te signals available at the outputs of 16 low pass filters are sampled, multiplexed and A/D converted. Thus, ¥© £° ice digital equivalent of the speech signal at the encoder output. Then, the encoded voice signal transmitted bY !° Yl encoder consists of all the frequency components from 200 to 3200 Hz, information about the ‘pitch’ of the speee ‘and an information about whether the speech is voiced or unvoiced, - wavelom Coding Techniques: PCM, DM, ADM and DPCM ce) [77] 200-400 Hz oe oa ‘voltage ; 400 — 600 Hz er | 0 = | cee = | = a ic =o la vr fee! x = DC voltage proportional 1p converter and multiplex 1 vonage reporional” 4D converter and mutilexer Fig, 11.46, Illustration of voice encoder 1.36.3. Ve Voice decoder first decodes the digital voice signal and then using a speech synthesyzer, produces an approximate voice signal at its output. Figure 11.47 shows the block diagram of a voice decoder. Working Operation ‘The working operation of the voice decoder may be described as under: ‘The received encoded voice signal is demultiplexed and decoded i.e., converted back to analog form by the demultiplexer and D/A converter. Corresponding to each rectifier-ilter combination at the encoder, we have a balanced tnodulator-filter combination at the receiver. The carrier signal to all these balanced modulators comes from either a qoise generator or a pulse generator output depending on the position of switch ‘SW’. The position of ewitch ‘SW depends on the decision made by the decoder of whether the received signals a ‘voiced’ signal or an ‘unvoiced! signal. When the received signal is ‘voiced’, switch ‘SW’ connects the Balanced Bandpass pulse generator output to the modulators ters inputs of all the balanced Tow Samos modulators, and when it is es oe unvoiced, the switch connects the noice generator at the input ofall the balaneed modulators. The Encoded position of switch ‘SW’ is voice dependent on the decision made $9"! by the decoder of whether the received signal is ‘voiced’ or ‘unvoiced’. When the received [7] emis IH 3000 - 3200 Hz }—>} signal is ‘voiced’, the outputs of only certain balanced modulators Voicediumvoiced will be present depending on the ‘itch Take frequencies present in the generator received signal. But, if an kit \Unvoleed sounds ‘unvoiced’ sig ii river nm Wwoiced’ signal is received, ther Decoder I icsaiumvoiced Voiced sounds Pulse ema 400-600 Hz Speech ‘output pmoo> >SORxXCEMO all the balanced modulators will Produce some output. The add n yutput, The adder generator will add all the analog signals as [tensions _ ee input to produce the voice or Pech output. The only disadvan Fig. 11.47. Voice decoder ‘ageof speech trans-mission using Yecoders is that the quality of eech is severally degraded.

You might also like