Examples of Communication Systems/Speech Coding - Revision history

Guenter at 14:06, 20 February 2023

2023-02-20T14:06:33Z

Guenter at 16:28, 11 February 2023

2023-02-11T16:28:50Z

Guenter at 16:27, 11 February 2023

2023-02-11T16:27:26Z

Hwang at 19:31, 6 February 2023

2023-02-06T19:31:25Z

Hwang at 19:21, 6 February 2023

2023-02-06T19:21:26Z

Hwang at 19:20, 6 February 2023

2023-02-06T19:20:56Z

Hwang at 19:19, 6 February 2023

2023-02-06T19:19:04Z

Hwang at 19:17, 6 February 2023

2023-02-06T19:17:19Z

Hwang at 19:15, 6 February 2023

2023-02-06T19:15:00Z

Hwang at 19:02, 6 February 2023

2023-02-06T19:02:22Z

← Older revision		Revision as of 14:06, 20 February 2023
Line 233:		Line 233:


−	[[File:~~EN Bei T 3 3 S8c~~.png\|right\|frame\|Compilation of AMR parameters ~~'''Korrektur''' Hochkomma, enthalten~~]]	+	[[File:EN_Bei_T_3_3_S8c_v2.png\|right\|frame\|Compilation of AMR parameters]]

	The following descriptions mostly refer to the mode with  $12.2 \ \rm kbit/s$:		The following descriptions mostly refer to the mode with  $12.2 \ \rm kbit/s$:

@@ Line 5: / Line 5: @@
 |Nächste Seite=Entire GSM Transmission System
 }}
 ==Various speech coding methods==

@@ Line 10: / Line 10: @@
 <br>
 Each GSM subscriber has a maximum net data rate of&nbsp; $\text{22.8 kbit/s}$&nbsp; available,&nbsp; while the ISDN fixed network operates with a data rate of&nbsp; $\text{64 kbit/s}$&nbsp; $($with &nbsp;$8$&nbsp; bit quantization$)$&nbsp; or&nbsp; $\text{104 kbit/s}$&nbsp; $($with&nbsp;$13$&nbsp;bit quantization$)$&nbsp; respectively.&nbsp;
-*The task of&nbsp; "speech coding"&nbsp; $($"speech coding"$)$&nbsp; '''KORREKTUR''' in GSM is to limit the amount of data for speech signal transmission to&nbsp; $\text{22.8 kbit/s}$&nbsp; and to reproduce the speech signal at the receiver side in the best possible way.
+*The task of&nbsp; "speech encoding"&nbsp;  in GSM is to limit the amount of data for speech signal transmission to&nbsp; $\text{22.8 kbit/s}$&nbsp; and to reproduce the speech signal at the receiver side in the best possible way.
 *The functions of the GSM encoder and the GSM decoder are usually combined in a single functional unit called&nbsp;  "codec".
-Different signal processing methods are used for speech coding '''KORREKTUR: encoding?''' and decoding:
+Different signal processing methods are used for speech encoding  and decoding:
 *The&nbsp; &raquo;'''GSM Full Rate Vocoder'''&laquo;&nbsp; was standardized in 1991 from a combination of three compression methods for the GSM radio channel.&nbsp; It is based on&nbsp; "Linear Predictive Coding"&nbsp; $\rm (LPC)$&nbsp; in conjunction with&nbsp; "Long Term Prediction"&nbsp; $\rm (LTP)$&nbsp; and&nbsp; "Regular Pulse Excitation"&nbsp; $\rm (RPE)$.

← Older revision		Revision as of 19:31, 6 February 2023
Line 284:		Line 284:
	*For example,  in the $12.2 \ \rm kbit/s$  mode,  $40$ bits are derived from this per subblock.		*For example,  in the $12.2 \ \rm kbit/s$  mode,  $40$ bits are derived from this per subblock.

−	* Thus,  in each frame of  $20$  milliseconds:     $160/244 ≈ 65\%$  of the encoding goes back to the block outlined in green in the graph on the last ~~page~~.	+	* Thus,  in each frame of  $20$  milliseconds:     $160/244 ≈ 65\%$  of the encoding goes back to the block outlined in green in the graph in the last section.

@@ Line 10: / Line 10: @@
 <br>
 Each GSM subscriber has a maximum net data rate of&nbsp; $\text{22.8 kbit/s}$&nbsp; available,&nbsp; while the ISDN fixed network operates with a data rate of&nbsp; $\text{64 kbit/s}$&nbsp; $($with &nbsp;$8$&nbsp; bit quantization$)$&nbsp; or&nbsp; $\text{104 kbit/s}$&nbsp; $($with&nbsp;$13$&nbsp;bit quantization$)$&nbsp; respectively.&nbsp;
-*The task of&nbsp; "speech coding"&nbsp; $($"speech coding"$)$&nbsp; '''KORREKTUR''' in GSM is to limit the amount of data for speech signal transmission to&nbsp; $\text{22.8 kbit/s}$&nbsp; and to reproduce the speech sibits pergnal at the receiver side in the best possible way.
+*The task of&nbsp; "speech coding"&nbsp; $($"speech coding"$)$&nbsp; '''KORREKTUR''' in GSM is to limit the amount of data for speech signal transmission to&nbsp; $\text{22.8 kbit/s}$&nbsp; and to reproduce the speech signal at the receiver side in the best possible way.
 *The functions of the GSM encoder and the GSM decoder are usually combined in a single functional unit called&nbsp;  "codec".

@@ Line 10: / Line 10: @@
 <br>
 Each GSM subscriber has a maximum net data rate of&nbsp; $\text{22.8 kbit/s}$&nbsp; available,&nbsp; while the ISDN fixed network operates with a data rate of&nbsp; $\text{64 kbit/s}$&nbsp; $($with &nbsp;$8$&nbsp; bit quantization$)$&nbsp; or&nbsp; $\text{104 kbit/s}$&nbsp; $($with&nbsp;$13$&nbsp;bit quantization$)$&nbsp; respectively.&nbsp;
-*The task of&nbsp; "speech coding"&nbsp; $($"speech coding"$)$&nbsp; '''KORREKTUR''' in GSM is to limit the amount of data for speech signal transmission to&nbsp; $\text{22.8 kbit/s}$&nbsp; and to reproduce the speech signal at the receiver side in the best possible way.
+*The task of&nbsp; "speech coding"&nbsp; $($"speech coding"$)$&nbsp; '''KORREKTUR''' in GSM is to limit the amount of data for speech signal transmission to&nbsp; $\text{22.8 kbit/s}$&nbsp; and to reproduce the speech sibits pergnal at the receiver side in the best possible way.
 *The functions of the GSM encoder and the GSM decoder are usually combined in a single functional unit called&nbsp;  "codec".
@@ Line 246: / Line 246: @@
 *&nbsp; $244$&nbsp; bits per&nbsp; $20 \ \rm ms$ &nbsp; &rArr; &nbsp; "mode&nbsp; $12.2 \ \rm kbit/s$",
-*&nbsp; $95$&nbsp; Bit per&nbsp; $20 \ \rm ms$ &nbsp; &rArr; &nbsp; mode&nbsp; "$4.75 \ \rm kbit/s$".
+*&nbsp; $95$&nbsp; bits per&nbsp; $20 \ \rm ms$ &nbsp; &rArr; &nbsp; mode&nbsp; "$4.75 \ \rm kbit/s$".
 <br clear=all>
 == Algebraic Code Excited Linear Prediction==

@@ Line 213: / Line 213: @@
 The GSM codecs described so far always work with a fixed data rate with regard to speech and channel coding,&nbsp; regardless of the channel conditions and the network load.
-In 1997,&nbsp; a new adaptive speech coding method for mobile radio systems was developed and shortly afterwards standardised by the&nbsp; "European Telecommunications Standards Institute"&nbsp; $\rm (ETSI)$&nbsp; according to proposals of the companies Ericsson,&nbsp; Nokia and Siemens.
+In 1997,&nbsp; a new adaptive speech coding method for mobile radio systems was developed and shortly afterwards standardized by the&nbsp; "European Telecommunications Standards Institute"&nbsp; $\rm (ETSI)$&nbsp; according to proposals of the companies Ericsson,&nbsp; Nokia and Siemens.
 :The Chair of Communications Engineering of the Technical University of Munich,&nbsp; which provides this learning tutorial&nbsp; "LNTwww",&nbsp; was decisively involved in the research work on the system proposal of Siemens AG.&nbsp; For more details,&nbsp; see&nbsp; [Hin02]<ref name ='Hin02'>Hindelang, T.:&nbsp; Source-Controlled Channel Decoding and Decoding for Mobile Communications.&nbsp; Dissertation. Chair of Communications Engineering, TU Munich. VDI Fortschritt-Berichte, Series 10, No. 695, 2002.</ref>.

@@ Line 206: / Line 206: @@
 #By 1994, a new process was developed with the&nbsp; &raquo;'''Half Rate Vocoder'''&laquo;.&nbsp; This has a data rate of $5.6\ \rm  kbit/s$&nbsp; and thus offers the possibility of transmitting speech in half a traffic channel with approximately the same quality.&nbsp; This allows two calls to be handled simultaneously on one time slot.&nbsp; However,&nbsp; the half rate codec was only used by mobile phone operators when a radio cell was congested.&nbsp; In the meantime,&nbsp; the half rate codec no longer plays a role.<br><br>
 #In order to further improve the voice quality, the&nbsp; &raquo;'''Enhanced Full Rate Codec'''&laquo;&nbsp; $\rm(EFR$&nbsp; codec$)$&nbsp; was introduced in 1995.&nbsp; This speech coding method &ndash; originally developed for the US American DCS1900 network &ndash; is a full rate codec with the&nbsp; $($slightly lower$)$&nbsp; data rate&nbsp; $12.2 \ \rm kbit/s$.&nbsp; The use of this codec must of course be supported by the mobile phone.<br><br>
-#Instead of the&nbsp; $\rm RPE &ndash; LPT$ &nbsp; $($"regular pulse excitation - long term prediction"$)$&nbsp;  compression  of the conventional full rate codec,&nbsp; this further development also uses&nbsp; &raquo;'''Algebraic Code Excitation Linear Prediction'''&laquo;,&nbsp;  which offers a significantly better speech quality and also improved error detection and concealment.&nbsp; More information about this can be found on the next page but one.
+#Instead of the&nbsp; $\rm RPE &ndash; LPT$ &nbsp; $($"regular pulse excitation - long term prediction"$)$&nbsp;  compression  of the conventional full rate codec,&nbsp; this further development also uses&nbsp; &raquo;'''Algebraic Code Excitation Linear Prediction'''&laquo;,&nbsp;  which offers a significantly better speech quality and also improved error detection and concealment.&nbsp; More information about this can be found on the page after next.

@@ Line 186: / Line 186: @@
 *From the optimal subsequence for subblock&nbsp; $i$&nbsp; $($with index&nbsp; $M_i)$&nbsp; the&nbsp; amplitude maximum&nbsp; $x_{\rm max,\hspace{0.03cm}i}$&nbsp; is determined.&nbsp; This value is logarithmically quantized with six bits and made available for transmission as&nbsp; $\mathbf{{\it x}_{\rm max}}(i)$.&nbsp; In total,&nbsp; the four RPE block amplitudes require&nbsp; $24$&nbsp; bits.
-*In addition,&nbsp; for each subblock&nbsp; $i$&nbsp; the optimal subsequence is normalised to&nbsp; $x_{{\rm max},\hspace{0.03cm}i}$.&nbsp; The obtained&nbsp; $13$&nbsp; samples are then quantized with three bits each and transmitted encoded as&nbsp; $\mathbf{\it X}_j(i)$.&nbsp; The&nbsp; $4 \cdot 13 \cdot 3 = 156$&nbsp; bits describe the so-called&nbsp; &raquo;''';RPE pulse'''&laquo;.
+*In addition,&nbsp; for each subblock&nbsp; $i$&nbsp; the optimal subsequence is normalised to&nbsp; $x_{{\rm max},\hspace{0.03cm}i}$.&nbsp; The obtained&nbsp; $13$&nbsp; samples are then quantized with three bits each and transmitted encoded as&nbsp; $\mathbf{\it X}_j(i)$.&nbsp; The&nbsp; $4 \cdot 13 \cdot 3 = 156$&nbsp; bits describe the so-called&nbsp; &raquo;'''RPE pulse'''&laquo;.
 *Then these RPE parameters are decoded locally again and fed back as a signal&nbsp; $e_{{\rm RPE},\hspace{0.03cm}i}(l)$&nbsp; to the LTP synthesis filter in the previous subblock,&nbsp; from which,&nbsp; together with the LTP estimation signal&nbsp; $y_i(l)$&nbsp; the signal&nbsp; $e\hspace{0.03cm}'_i(l)$&nbsp; is generated&nbsp; (see&nbsp; [[Examples_of_Communication_Systems/Voice_Coding#Long_Term_Prediction|$\rm LTP graph$]]).

← Older revision		Revision as of 19:02, 6 February 2023
Line 40:		Line 40:
	*is first sampled with  $8 \ \rm kHz$  and		*is first sampled with  $8 \ \rm kHz$  and

−	*then linearly quantized with  $13$  bits   ⇒   »''';A/D conversion'''«,	+	*then linearly quantized with  $13$  bits   ⇒   »'''A/D conversion'''«,