Speech and Channel Coding

Here we will consider two forms of coding techniques used within the GSM system.  Firstly the process used to convert human speech into a digital equivalent and secondly the coding processes for compressing and protecting the data for transmission over the air interface.

1: Speech Coding Techniques

*GSM transmits using digital modulation- speech must be converted to binary   digits.

* Coder and Decoder must work to the same standard

*Simplest coding Scheme is Pulse code Modulation(PCM)

   * Sampling every 123 micro second.

   * Requires data rates of 64 Kbps.

*This is too high for the bandwidth available on the radio channels.

       

There are several schemes available for reducing the bandwidth required for a single voice channel from the 64 kbps PCM requirement.   When the GSM specification was being defined, over 20 different voice coding schemes were initially considered.  This number was rapidly reduced to 6 schemes from 6 different countries before trials started.

2: Advanced Speed Coding

·        We can nor send the 64 Kbps required by PCM

·        We need alternative speed coding techniques.

·        Compare time to speak a word or sentence with time of transmit corresponding text.

                                                            

·        Attempts to encode speech more efficiently.

                 *Speech consists of periodic waveforms-

                  So just send the frequency and amplitude.

                *Model the vocal tract-Phonemes, voiced and unvoiced speech.

·        Vocoder- synthetic speech quality

        

Each of the schemes involved various methods of manipulating the information contained within speech to balance the best quality reproduction with the minimum bandwidth.

Speech obviously contains far more information than the simple text transcription of what is being said. We can identify the person speaking, and be aware of much unspoken information from the tone of voice and so on.

Early vocoders which reduced the voice to just simple waveform information lacked the human qualities which we need to hold a meaningful communication.

Hybrid encoders give greater emphasis to these qualities by using regular pulse excitation which encodes the overall tone of the voice in great detail.

Eventually a hybrid of the German RPE-LPC^[1] scheme and the French MPE-LTP^[2] was chosen.  This became known as Regular Pulse Excitation with Long-Term Prediction (RPE-LTP).

3: GSM Speech Coding

The first stage of speech encoding is to convert human speech, generated by the microphone as an analogue signal, into a digital equivalent.

GSM achieves this by sampling the analogue voice signal every 125mS, or 8000 times per second.  Each sample is quantised into one of 8192 voltage levels.  Each of these levels is represented by a 13-bit (2¹³) binary code.  Therefore, every second 8000x13-bit samples of the analogue signal are produced, resulting in a raw data rate of 104kbps.

This raw bit stream is presented to the RPE-LTP Vocoder where it is chopped into 20mS (2080-bit) blocks.  Each block is then processed separately.

The vocoder categorises the data in each 20mS-block into three parts:

·        Short-term Linear Predictive Coding data (LPC)

·        Long-term prediction data (LTP)

·        Regular Pulse Excitation data (RPE)

The long and short term prediction waveforms are each encoded as frequency and amplitude information in the form of 36-bit blocks, while the RPE is encoded in a 188-bit block primarily to ensure that the characteristic tone of the voice is reproduced well.

The resulting data rate of 13 kbps is suitable for the bandwidth available on the air interface.

Channel coding will continue in next blog…………

J

---

^[1] Regular Pulse Excitation with Linear Predictive Coding

^[2] Multipulse Excitation with Long Term Prediction