JPS6165630A - Echo suppression system of digital telephone set - Google Patents

Abstract

PURPOSE:To eliminate exclusive hardware and to avoid adverse effects of a voice analysis section on characteristic quantity extraction processing by discriminating whether a characteristic quantity is based on a echo or not in a digital telephone system using the characteristic quantity extraction system and attenuating and correcting the characteristic quantity by the result. CONSTITUTION:An analog voice signal outputted from a DA converter section 6 is subjected to 4/2-wire conversion by a hybrid transformer 7, fed to a telephone set 8, and a part is inputted again to an AD converter section 10 by reflection as an analog echo signal, converted into a digital echo signal, a characteristic quantity is extracted by an analysis section 11 and outputted to a microprocessor 20. The microprocessor 20 uses the analyzed characteristic quantity and voice characteristic quantity compression information read from a shift register 12, an echo discrimination logic 13 discriminates echo, and when the echo is discriminated, after compression process 14 corrects attenuation of characteristic quantity from the voice analysis section 11, the signal is compressed and transmitted to a transmitter 15 and a low bit rate transmission line 16.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a digital telephone using a vocoder that compresses and transmits the feature amount of voice information and restores the voice from the feature amount of the compressed voice information on the receiving side. k.

More specifically, the present invention relates to an echo suppression system for digital telephones that is suitable for eliminating echoes without using a dedicated circuit.

[Background of the invention]

The conventional echo suppression method requires a dedicated node square, which has the problem of complicating the device. Furthermore, when an echo suppressor is inserted before a voice analysis section that extracts feature quantities of voice information, there is a problem in that it affects the analysis process that extracts the feature quantities.

[Purpose of the invention]

SUMMARY OF THE INVENTION In view of the problems of the prior art described above, an object of the present invention is to provide an echo suppression method for digital telephones that does not require dedicated hardware and does not affect the analysis process for extracting special amounts by a voice analysis unit. Our goal is to provide the following.

[Summary of the invention]

The echo suppression method for a digital telephone of the present invention includes a voice analysis section that extracts voice features, an information compression section that performs information compression on the feature quantities and sends a small amount of audio compressed information to other telephones, and a voice analysis section that extracts voice features. In a digital telephone, which consists of a voice synthesis unit that restores voice from the slightly compressed phonetic information input from the voice analysis unit, the features output from the voice analysis unit and other
Based on the audio micro-compression information input from the 1-speaker device,
Determine whether the feature value is based on Econi,
The echo is suppressed by performing attenuation correction on the feature amount only when it is determined that the feature amount is based on an echo.

[Embodiments of the invention]

The present invention will be described in more detail below with reference to embodiments shown in the accompanying drawings.

The accompanying drawing shows an embodiment of the present invention, in which a receiver 2 receives from a low-bit, rate transmission path 1 audio-related micro-compressed information in which feature quantities of audio information are compressed. The macro processor 20 reads out the compressed audio feature signal from the resino (2), performs a synthesis process 3 to convert the format, and sends it to the synthesis unit 5. The format-converted audio micro-compressed information is input to the shift register 12 having a format.The synthesis section 5 synthesizes a digital audio signal from the input audio-based micro compression information, outputs the synthesized digital audio signal, and sends it to the DA converter section 6. The DA converter section 6 converts the digital audio signal into an analog audio signal.The analog audio signal is converted into 4-wire and 2-wire by a hybrid transformer 7, and the audio can be heard on the telephone 8, but some of it is lost due to one reflection. , which is again input as an analog echo signal to the line of the AD converter section 10 and the analysis section 11).

Here, 9 is a balanced circuit network, and a line between the KAD converter section 10 and the voice analysis section 11 converts the analog voice signal from the telephone 8 into a digital voice signal, extracts its feature quantity, and then outputs the signal to the microprocessor. 20, a transmitter 15, and a low-bit, rate transmission line 16, which is a line for transmitting audible micro-compressed information to the telephone of the other party.

The analog F-signal inputted to the AD converter section 10 is converted into a digital knee signal, and the feature quantity is extracted by the analysis section 11 and outputted to the microprocessor 20. The microprocessor 20 uses the analyzed feature amount and the audio micro-compressed information read from the shift register 12 to perform the engineering process.

Echo determination logic 15 performs echo determination.

That is, the echo determination logic 15 compares the analyzed feature quantity (echo) with the vocal micro-compression information, and determines whether it is an echo or an analog voice signal from the telephone 8. For example, when focusing on sound source information among the feature quantities, the threshold determined by the sound source information output from the shift register 12 and the sound source information output from the speech analysis section 11 are compared, and if the latter is smaller, then 1 microprocessor 20 determines that it is an echo. microprocessor 20
If it is determined to be an echo, the compression processing 14 performs correction (attenuation of the volume) to attenuate the feature quantity (of which the sound source information) from the voice analysis unit 11 is compressed, and the transmitter 15. The low bit rate is transmitted through the transmission line 161C. Through the above processing, echoes can be easily suppressed without using a dedicated echo suppressor. Note that since the process of the echo determination theory 915 described above is executed at a high speed compared to the speed of human conversation (for example, about 10yaIgCK once), the voice information from the telephone 8 is not subject to attenuation correction.

〔Effect of the invention〕

According to the present invention, in a digital telephone system using a feature extraction method, an echo suppression method for a digital telephone that does not require dedicated hardware and does not adversely affect feature extraction processing by a voice analysis unit is provided. can be provided.

[Brief explanation of the drawing]

The accompanying drawings are block diagrams illustrating one embodiment of the invention. DESCRIPTION OF SYMBOLS 1...Low bit, rate transmission path (receiving), 2...Receiver, 5...Combining section, 6...DA converter section,
7...Hybrid, transformer, 8...Telephone, 1
0...AD converter section, 11...analysis section, 12.
...Shift register, 15...Echo judgment logic, 14
...Compression processing, 15...Transmitter, 16...
・Low bit, Renit transmission line.

Claims (1)

[Claims] A voice analysis section that extracts voice features; an information compression section that compresses information on the features and sends the compressed voice feature information to other telephones; In a digital telephone comprising a voice synthesis unit that restores a voice, it is determined whether the characteristic quantity is based on an echo based on the characteristic quantity output from the voice analysis unit and the voice characteristic quantity compression information input from another telephone. A digital telephone characterized by comprising: an echo determining means for making a determination; and a correcting means for attenuating the feature amount only when the echo determining means determines that the feature amount is based on an echo. Echo suppression method.