CN1165448A - Tone Detector - Google Patents

Abstract

A tone detector for detecting particular frequencies within an incoming signal, such as dial tone on call waiting tone received from exchange equipment. The detector comprises a bandpass filter or analog-to-digital converter, a microprocessor, a buffer and an analyzer. The analyzer periodically examines samples of the imcoming signal stored in the buffer for the presence of the particular frequency by means of a modified Fourier Transform adapted to allow the use of a simple low-cost microprocessor.

Description

Tone Detector

The present invention relates to sound and detect, relate in particular to a kind of microprocessor that allows lower cost and detect characteristic frequency in the input signal for example dialing tone that receives from switching equipment of telephone subscriber's unit or the existence or the non-existent device of call waiting tone.

Several sound detection methods have been known.The step that a kind of method comprises is to allow the signal that contains tested acoustic band pass filter, level shifter by a very high Q value, enters the digital input port of microprocessor then.When having desired frequency in the input signal, microprocessor is found a rectangular wave at the input stitch.This is easy to detect with interruption method or polling method.

But in this method, filter is relatively more expensive.Usually what will be made good use of provides and neededly roll-offs and insert tight tolerance or obtain the centre frequency precision with the element of factory tuned.

Another known method is to insert a phase-locked loop (PLL) in this situation, and the locking signal that input signal enters PLL and PLL output by a low reactance-resistance ratio band pass filter is connected to a digital input port of microprocessor.

This method is more expensive also, and PLL is the parts that cost is higher.Still require the element of less tolerance or factory tuned to obtain the precision of centre frequency.

Another enters a digital input port of microprocessor by a low reactance-resistance ratio band pass filter and level shifter by input signal than the less known method of preceding method cost.Microprocessor will be seen a square input signal of filtering.If each rising of microprocessor (or decline) edge interrupts, just can be to the cycle timing between the edge.If should the cycle corresponding to search rate one of (or several search rates), microprocessor just can point out to exist this signal., generally before positive detection arrives this signal, many match samples must appear.This method is very responsive to input noise.

This method uses less expensive element that a kind of closely device of a plurality of frequencies of detection of bandwidth that uses rationally is provided, and still, along with the difference of microprocessor, this method may be used the useful processing power that reaches suitable vast scale.

The digital signal processor that specialized designs is used for the sound detection also is known, but relatively more expensive.

An object of the present invention is to provide the modified model sound detection device that the lower cost of a kind of usefulness obtains tight bandwidth.

Another object of the present invention provides a kind of modified model sound detection device that obtains tight bandwidth with minimum treat power.

According to a kind of sound detection device provided by the invention, comprise an analog signal input unit and an analog/digital conversion device that is coupled to a digital input port of microprocessor by band pass filter means, wherein microprocessor comprises the sampling apparatus that is used for the signal sampling that is added to described analog signal input unit, the n position buffer storage that is used for the data of storage representation sampled signal, and the analytical equipment that is used for the sample of the described buffer storage of periodic analysis, described analytical equipment then uses a kind of Fourier transform of modification to detect the signal of preset frequency.

For making the present invention be easy to implement, narrate embodiment now in conjunction with the accompanying drawings, in the accompanying drawings:

Fig. 1 illustrates the first embodiment of the present invention.

Fig. 2 illustrates the second embodiment of the present invention.

Fig. 3 performed operational flowchart that is the sampler shown in Fig. 1,2 in the analyzer.

With reference to Fig. 1, device comprises a band pass filter that was used for before microprocessor sampling from the most of noise drawn game of input signal filtering external signal.

One is used for analog signal conversion is the digital signal that can use of microprocessor and the amplitude limiter of filtering low power noise signal.Negative input voltage is moved to a numeral " low " level.The positive noise voltage that is higher than thresholding is moved to a numeral " high level ".Hysteresis in a small amount prevents from because of low level signal false triggering to take place.

Microprocessor uses a kind of algorithm to detect specific frequency, this algorithm separated into two parts.First is stored in to the input signal sampling and with it and is provided with post analysis in the buffer.Sampling must be carried out with quite high frequency, but the instruction count of each sample is but very low.Second portion is that the information of analysis part periodic analysis sampling is to demonstrate specific frequency.This requires each cycle that the more processing time is arranged, but generally can finish in much lower frequency.

Microprocessor moves into a n position buffer to the state sampling of input stitch and with sample regularly.This part is that the time is critical.

Usually, in Digital Signal Processing, sampling frequency need be higher 3 or 4 times than the bandwidth of input signal.But, should with in can take under-frequency sampling.But the designer must guarantee to differentiate between any pseudofrequency that band pass filter can introduce at target frequency with owing to sampling rate.

Give an example, if will search for a 450Hz signal, then the sampling rate of 1500Hz may detect a signal with frequency 1050Hz.Thereby this example requires a center to fix on the band pass filter of 450Hz, with the signal attenuation of any 1050Hz to below the Noise gate limit value.

In another example, if search for the signal of a 2130Hz, then the sampling rate of 1626.6Hz will provide the pseudofrequency of 503Hz, 1123Hz, 2750Hz or the like.This example requires a center to fix on the band pass filter of 2130Hz, and these frequencies are all decayed to below the Noise gate limit value.

The specification n of sample buffer only depends on desired bandwidth and sampling rate according to following formula.

Buffer specification=sampling rate/bandwidth

For example, if the bandwidth of requirement 30Hz when sample with 1500Hz then needs to store 50 samples.

Analysis part periodic analysis sample is to judge whether having characteristic frequency in input signal.This part is not the time criticality.

Algorithm that this part is used is based on the Fourier transform of standard, but done some modifications, so that it is suitable for simple low-cost microprocessor.

The Fourier transform formula is as follows: $F\left(w\right)=\frac{1}{\sqrt{2\mathrm{\π}}}\underset{-\∞}{\overset{\∞}{\∫}}f\left(t\right)e^{\mathrm{\ωt}}\mathrm{dt}$

Theory shows that when radial velocity was ω, the energy among the signal f (t) was by amplitude | F (ω) | provide.In the practice, the complex exponential item is processed into cosine term and imaginary number sine term sum usually.

In order to use on simple microprocessor, the present invention simplifies this formula with following way:

1. what replace analog input signal is the one bit digital signal that input is processed into a sampling.Sample was 0 when input signal was negative, and being input as the timing sample is 1 (in this algorithm, numerical value 0 actual representative-1).

2. what Sin (ω t) and Cos (ω t) were multiply by in replacement is that device of the present invention multiply by two quadrature square waves that detect frequency.

3. replacing multiplication, is that device uses exclusive NOR (or exclusive-OR operation) by turn.

4. replacing in the infinite interval upper integral, is only a last n sample simply to be sued for peace.Be simplified as number counting again to " 1 " position among the result.

5. omit any constant.

6. substitute using Pythagorean theorem that two quadrature components are added to together, is that device takes absolute value simply and two components are added to one.

With reference to Fig. 3, the false code and the flow process that are used for analysis part are as follows:

1. beginning (termly)

2.T _Emp1=sample XOR square wave (T _Emp1 is n position buffer)

T _Emp2=sample XOR quadrature square wave (T _Emp2 is n position buffer)

3.Bits1=T _EmpThe number of " 1 " position in 1

Bits2=T _EmpThe number of " 1 " position in 2

4. the absolute value of numerical value 1=(n/2-Bits1)

The absolute value of numerical value 2=(n/2-Bits2)

5. result=numerical value 1+ numerical value 2

If the result greater than threshold value, then

Frequency exists

Otherwise

Frequency does not exist

With reference to Fig. 2, in another embodiment of the present invention, the digital circuit divided by n places between level shifter and the microprocessor.This modification makes device can detect the existing of frequency (and other modes will just can detect with multiprocessing power more).The software of microprocessor is arranged for seeking the signal of the 1/m frequency with search rate.

This modification also influences the relation between bandwidth samples speed and the buffer specification.

Buffer specification=(sampling rate/bandwidth) m

Because algorithm is carried out a kind of analysis based on a large amount of input samples, so algorithm is more insensitive to the noise ratio of smallest number.

But because the relation of the characteristic of amplitude limiter circuit, if exceed the above noise signal of search rate 3dB by band pass filter, then this algorithm can not detect search rate.

Best, in high-noise environment, band pass filter should be done narrowlyer as much as possible.

Increase the noiseproof feature that a divider also greatly reduces algorithm.Thereby in high-noise environment, preferably without divider.

The present invention is applied to the low-cost microprocessor of any use and detects the occasion that has characteristic frequency in the input signal, for example says can be used on:

1. in telephone set, differentiate different dialing tones.

2. the existence of detection of call wait or CIDCW sound in telephone set.

3. in pay phone, detect the existence of metering pulse.

4. electric weight is supplied non-peak input.

5.FM (frequency modulation) wireless 19KHz carrier wave detects.

This device can be implemented with hardware (for example using integrated circuit) merely, as a dedicated tone detector of above-mentioned all purposes.

Claims (3)

1. sound detection device, comprise an analog signal input unit and a mould/analog-to-digital conversion apparatus that is coupled to the digital input port of microprocessor by band pass filter means, the n position buffer of the data that wherein said microprocessor comprises the sampling apparatus that is used for the signal sampling that is added to described analog signal input unit, be used for the storage representation sampled signal and being used for is analyzed the analytical equipment of the sample of described buffer termly, and described analytical equipment uses a kind of Fourier transform of modification to pick the signal of preset frequency.

2. the device that requires of claim 1, one of them digital means divided by n are placed between the digital input port of described converting means and described microprocessor and work.

3. claim 1 or 2 devices that require are installed in the tele-release user unit.

Images