Dtmf Signals And Speech In Telephony Computer Science Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Dual Tone Multi Frequency (DTMF) is an electronic signalling standard currently used in push button telephone dialling. This method is commonly used in interactive telephony systems such as Telephone banking, Tele marketing and Interactive control services offered by service providers. DTMF that is used in modern telephony for tone dialling is known as Touch-Tone, was first used by AT&T as a registered trademark which was standardized by the Bell Laboratories. DTMF standards were then redefined by the International Telecommunication Union (ITU) Q.23, which introduced the DTMF signalling as an efficient alternative to pulse dialling system in telephony.

The DTMF signal is composed using two superimposed sinusoidal waveforms. Every key press can be represented as a function below, where N is defined as number of samples taken.

dt[n] = sin[ωLn] + sin[ωLn] , 0 ≤ n ≤ N-1

























High Frequencies

Low FrequenciesThese two wave forms are chosen from a set of eight frequencies. The eight frequencies are from two sets of high and low frequency groups. The frequency allocation is as follows.

Figure 1.1 Standard keys and corresponding frequencies

The frequencies used were selected in such a way that no two frequencies are integral multiples of each other. This characteristic leads to accurate detection of key presses by the switching network, even when noise is high in the communication medium. User friendliness, accuracy and efficiency of the DTMF method established its usage in communication systems worldwide.

The table 1.1 states the parameters in the AT&T standard These DTMF standards are crucial to validate and properly decode a DTMF signal even in highly noisy receiving condi­tions. If the parameters are satisfied by the input signal then the Detected DTMF signal is stated as a valid DTMF key press.



Signal frequency

Low group

697,770,852,941 Hz

High group

1209,1336,1477,1633 Hz

Frequency tolerance


=< 1.5%



Power levels per frequency


0 to -25 dBm


Max -55 dBm

Power level difference between frequencies

+4dB to - 8 dB

Signal duration


Min. 40 ms


Max. 23 ms

Table 1.1 DTMF parameters by AT&T standard [4]

2. Aims and Objectives

As the project topic illustrates, discrimination process between DTMF signals and speech signals in telephony is the main purpose of this project. The final outcome of the project should get a data stream consisting of DTMF signals, added speech and noise as the input. The system should do the noise and speech discrimination process from the data stream.

A graphical user interface should display the keys pressed in relevance with the DTMF tone as the output. The system should consider about the accuracy of the DTMF detection process and efficiency of the overall system.

The objectives of this project are as follows

Discover the characteristics of the DTMF pulses and compare, contrast them with the characteristics of the speech signal in telephony.

Investigate the efficient filtering function that can be used to filter out noise and speech signal accurately

Select and use of an efficient DTMF detection algorithm

Discrimination of DTMF tones from normal telephony and Validating the DTMF signal

Implementation DTMF generation, encoding, signal detection and identifying the correct key press

Implementing the GUI

3. Scope

The project borders are defined considering the inputs and outputs and processes of the DTMF detection system. The main input to the system is continues data stream consisting the DTMF signal. The system consists of a DTMF generator also. Main system processes are noise filtering, speech discrimination and DTMF detection. The main output of the system is the digit which was pressed by the user, in this case the generated DTMF signal.

4. Overview of the system

Column FrequencyCalculation

Row Frequency Calculation

DTMF signal generator









Finding the correct key press

Band pass Filter

Frequency detection and validation

Output relevant key press

Noise + speech

DTMF decoding system

A DTMF decoding structure mainly consists of two blocks.

band pass filters which will segregate individual frequency components and reduce noise effect

detectors to identify the frequency components individually

To achieve the goals of the project an efficient algorithm must be selected which is capable of accurately selecting

Specific tones that signify the key presses individually

Relative signal power at each of the frequencies,

The time period of the tones

Check whether a valid DTMF tone is present considering the signal characteristics and industry standards

Figure 1.2 Overview of the system (block diagram)

5. Project Plan - Gantt chart

6. DTMF Detection

DTMF detection process requires the ability to identify and differentiate from the eight DTMF frequencies that are present. Because of the noise introduced to the input signal there is a possibility of misjudging the noise as DTMF tones. To avoid this it is vital to have a technique of identifying and rejecting the false tones produces due to noise. One phenomenon which can be used to estimate the performance of the DTMF detection signal is known as the talk-off error. Talk-off error is defined as the improper detection of a DTMF tone due to speech or other forms of noise in the communication channel. [6] One method for discriminating between DTMF tones and speech is checking for the presence of a second harmonic. If the second harmonic is detected, then the signal can be rejected because true DTMF tones only contain fundamental tones.

Valid input signal

Check row and column frequencies

Check second harmonic

Single raw and column detect


Second harmonic present


False tone detected



Valid DTMF detect


Figure 1.3 Flow chart of the DTMF detection [6]

7. Existing Methodologies for DTMF detection and Decoding

7.1 Filtering method

Theoretically the filtering method is the simplest and most easily implements method for multi frequency detection. The superimposed DTMF signal is usually first filtered with use of a pair of low pass and high pass filters and then the signal is fed in to a set of band pass filters.

The superimposed signal is separated in to low frequency or high frequency group in reference to the DTMF standard frequency. Eight band pass filter blocks are used in the case of DTMF to retrieve the frequencies in each range. Due to very high computational density software based filters are not used. [2]









Low pass

High pass

Signal consisting DTMF

Digit CalculationThe filters are implemented as hardware blocks specified of doing the filtering process. Due to the inefficiency of the filtering method for detection of tones in multiple Pulse Code Modulated channels modern tone detection schemes do not use this method.

Figure 1.4 Frequency filtering method [2]

7.2 Fast Fourier Transform (FFT)

Another methodology to find existence of an individual known frequency in a monitored signal is considering an FFT (Fast Fourier Transform) and analyse it to see whether there are specific frequencies present in the data stream. Due to the reason of ignoring most of the computed signals this method is not efficient. [3]

7.3 Discrete Fourier Transform (DFT)

The discrete Fourier transform (DFT) also produce the same statistical result for a single frequency of interest. This makes the DFT a preferred choice for frequency detection. The Goertzel Algorithm is a DFT in cover up. Goertzel algorithm has numerical methods to eliminate complex number arithmetic which are present in FFT. The DFT method increases the efficiency and ease of implementation. Even though FFT is fast on Data Acquisition Processor boards, the performance in proportional to CPU usage requiring a CPU with high processing power.

8. The Goertzel algorithm

The Goertzel algorithm is principally suitable for detection of particular, pre-defined frequencies consist in a signal by means of minimal number of memory and computation resources. Therefore the Goertzel algorithm has been widely used for detecting DTMF frequencies contaminated by audio signals. This algorithm is commonly implemented in microcontroller devices where both RAM limitations and processing power prevent the use of other algorithms such as the Fast Fourier Transformation.

Blocks of data are processed by the Goertzel algorithm to give in each result. When considering the FFT method, before any processing can begin all the data blocks should be buffered. But in the case of DFT or Goertzel filter, data are processed consecutively and efficiently as they arrive. When the data sequence end the output is processed and ready to be forwarded.[8]

Why use Goertzel Algorithm?

Most efficient approach for smaller sequence lengths

Reduces the data memory required considerably

Requires limited processing power

Best computational load and memory usage efficiency

Real time processing

9. Design, Coding and implementation

To achieve the objectives of the project a system should be implemented to get the signal stream with DTMF signals, noise and speech signals and to discriminate DTMF signals and speech to output the specific key presses according to the DTMF tones. To implement the system there should be a development environment for the system. Considering the requirement of the project MATLAB and Microsoft Visual Studio can be used to code and implement the program.

Considering that this system is more concerned about frequency detection and mathematical functions MATLAB software is chosen to continue with the project system implementation.

MATLAB is most useful for signal processing and computation. It is specially designed to do this task easily and well. Considering the requirements of this project the functions of MATLAB such as plotting of functions and data, implementation of algorithms, creation of user interfaces, and interfacing with programs written in other languages will be useful.[8]

10. Testing and Debugging

Attempts are made to test the system performance in a noisy environment. Test cases should be created comprising of different noise and interference level. The parameters within the system should be varied till an accurate DTMF detection process is implemented.

The system will also be tested in a noise free environment to check whether there are any logical and syntax errors within the implemented system.

11. Resources needed for the project

Development Environment - MATLAB R2010b

12. Conclusion

Considering the facts that Enhances the probability of success by addressing and mitigating factors early on that could affect the project the feasibility study was conducted. Technical and legal feasibility was also considered with the economical feasibility.

In feasibility the following issues are taken into consideration.

Whether the objectives are logical, reachable , time efficient and specific

Whether the required resources are available

Whether the required technology is available or not

Availability of Testers & debuggers

Software and hardware

Considering the facts that leads to the success of this project the work plan will be take in to consideration with the purpose of achieving the objectives.