JP2000209663A - Method for transmitting non-voice information in voice channel - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enhance transmission of data, especially a non-voice signal data through a radio voice channel. SOLUTION: Information is transmitted by bits assigned to an output of one or both code books by setting a gain of a concerned code book to zero. A receiver side VOCODER does not interpret the output of the code book whose gain is set to zero. According to this system, the information can further be transmitted to a VOCODER 10 with a completely transparent method. As an application of this technology where a 'secret' message is transmitted, a parameter to generate a non-voice signal may be transmitted. For example, the information to generate a call waiting tone, a DTMF tone or a TTY/TDD character is imbedded secretly to a compressed bit stream, from which a non-voice tone can be regenerated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to communications, and more particularly, to the transmission of data on a wireless voice channel.

[0002]

2. Description of the Related Art Speech encoders / decoders (vocoders) are used to compress speech signals, reducing the transmission bandwidth on communication channels. By reducing per-call bandwidth, calls on the same channel can be increased. There is a type of vocoder known as a code-excited linear prediction (CELP) vocoder in which speech is modeled by a series of filters. The parameters of these filters can be sent with far fewer bits than the original speech.

Further, it is necessary to send an input (excitation) to these filters to reconstruct the original sound. Because too much bandwidth is required to send the excitation directly, coarser replacement of the excitation with a smaller number of non-zero pulses can be used.
e) Perform approximation. The positions of these pulses can be sent with a very small number of bits, and a crude approximation to this original excitation is appropriate for reproducing high quality speech. This excitation is represented by a fixed codebook contribution and the corresponding gain. Also, the pseudo-periodicity existing in the voice is represented by the gain corresponding to the output of the applicability codebook. A fixed codebook output and its corresponding gain, an adaptive codebook output and its corresponding gain, and filter parameters (also known as linear prediction coder parameters) are sent to represent the encoded audio signal.

[0004] Vocoders were initially designed to model their characteristics and compress speech by sending its parameters with far fewer bits than sending the speech itself. As wireless telephones become more common, people use wireless telephones in the same area of non-voice applications as they are used in traditional terrestrial telephones, such as accessing voicemail and receiving calls. I expect more and more. More recently, the United States Federal Communications Commission (FCC) has defined a text telephone for the deaf (TTY / TDD) using a digital cellular telephone.

[0005]

A problem with non-speech applications is that they do not match the vocoder's speech model. If the non-speech signal is processed by a vocoder, the decoded result is not always satisfactory. This problem is further exacerbated by the fact that wireless telephones operate in error-prone environments. To recover from transmission errors, the vocoder relies on a speech model to recover from random errors. Thus, non-speech signals do not fit this model and reconstruction is not appropriate.

[0006]

SUMMARY OF THE INVENTION The present invention sends information in bits assigned to one or both codebook outputs by setting the gain of the corresponding codebook to zero. By setting the gain to zero,
The output of the codebook is not interpreted by the receiving vocoder. This scheme allows the vocoder to send more information in a completely transparent manner. An application of this technique to send a "secret" message is to send a parameter that produces a non-voice signal. For example, call waiting tone, DTMF tone, or TTY /
The information that generates the TDD characters can be clandestinely embedded in the compressed bitstream to regenerate non-voice tones.

[0007]

FIG. 1 shows a block diagram of a typical vocoder. Vocoder 10 receives digitized audio at input 12. This digitized audio is A / D
An analog audio signal that has passed through a converter and is typically divided into frames of the order of 20 ms. The signal at input 12 is passed to an encoder section 14, which encodes the audio to reduce the amount of bandwidth used for audio transmission. The encoded audio is output 16
It will be available at.

The encoded speech is received by a similar vocoder decoder at the other end of the communication channel. The decoder at the other end of the communication channel is similar or identical to the decoder section of vocoder 10. Each voice is received by the vocoder 10 via an input 18 and passed to a decoder unit 20. The decoder section 20 produces digitized audio at an output 22 using the encoded signal received from the transmitting vocoder.

[0009] Vocoders are well known in the communications arts.
For example, vocoders can use literature, Speechand audio coding f
or wireless and network applications, Bishnu S. At
al, Vladimir Cuperman, edited by Allen Gersho, 1993, Kluw
er Academic Publishers. Vocoders are widely available, including Qualcomm Incorporated and Lucent
Manufactured by companies such as Technologies Inc.

FIG. 2 shows the main functions of the encoder 14 of the vocoder 10. The digitized audio signal is received at input 12 and passed to linear prediction coder 40. The linear prediction coder performs a linear prediction analysis of the input speech once per frame. Linear predictive analysis is well known in the art and creates a linear predictive synthesis model of the vocal tract based on the input speech signal. The linear prediction parameters or coefficients describing this model are sent via output 16 as part of the encoded audio signal.

The coder 40 uses this model to produce a residual speech signal that represents the excitation used by the model to produce the input speech signal. This residual audio signal is made available at output 42. The residual speech from output 42 is provided to input 48 of open loop pitch search unit 50, to the input of applicable codebook unit 72, and to fixed codebook unit 82. The impulse response unit 60 receives the linear prediction parameters from the linear prediction coder and generates an impulse response of the model generated by the coder 40. This impulse response is used in application and fixed codebook units.

The open loop pitch search unit 50 models the pitch using the residual speech signal from the coder 40 and provides at output 52 the pitch, ie, what is commonly referred to as the pitch period or pitch delay signal. . The pitch delayed signal from output 52 and the impulse response signal from output 64 of impulse response unit 60 are received at input 70 of applicable codebook unit 72. The application codebook unit 72 produces a pitch gain output and a pitch index output, which becomes part of the encoded audio output 16 of the vocoder 10. Output 74 of applicable codebook unit 72 also provides a pitch gain pitch index signal to input 80 of fixed codebook unit 82. Also, the application codebook unit 72 provides an excitation signal and an application codebook target signal to an input 80.

An application codebook 72 produces its output using the digitized audio signal from input 12 and the residual audio signal produced by linear prediction coder 40. The adaptive codebook 72 uses the digitized audio signal and the residual audio signal of the linear prediction coder 40 to form an adaptive codebook target signal. The applied codebook target signal is used as an input to the fixed codebook unit 82, and is also used as an input to a calculating means for generating a pitch gain, a pitch index and an excitation output of the applied codebook unit 72. Also, the applied codebook target signal, the pitch delay signal from the open loop pitch search unit 50, and the impulse response from the impulse response unit 60 are used to create a pitch index, a pitch gain, and an excitation signal, which are fixed codebook units. It is passed to 82. The manner in which these signals are computed is well known in the vocoder field.

The fixed code book unit 82 has an input 80
Using the input received from, a fixed gain output and a fixed index output are made, which are used at output 16 as part of the encoded speech. The fixed codebook unit attempts to model the stochastic part of the residual speech signal of the linear prediction coder 40. The fixed codebook search target is created by determining the error between the currently applied codebook target signal and the residual audio signal. A fixed codebook search creates a fixed gain and fixed index signal for the excitation pulse, minimizing this error. The manner in which the fixed gain and fixed index signals are calculated using the output from the applied codebook unit 72 is well known in the field of vocoders.

Switches 90 and 92 are used to send data that replaces the bits used to send the fixed codebook output and the applied codebook output, respectively. When the contacts of these switches 90, 92 are in position A, the corresponding codebook output is replaced with data or other information and the corresponding codebook gain is set to zero or nearly zero. As a result, the scaled codebook output or excitation produced at the receiver will be zero or nearly zero, without adversely affecting the filters normally used by the receiving vocoder to model the transmitted speech. I'm done.

FIG. 3 shows a functional block diagram of the decoder section 20 of the vocoder 10. The encoded audio signal is received at input 18 of encoder 20. The encoded audio signal is received by the decoder 100. Decoder 100 produces fixed and applied code vectors corresponding to the fixed index and pitch index signals, respectively. These code vectors along with the pitch gain and fixed gain signals are passed to the excitation building block of unit 110. The pitch gain signal is used to scale the application vector created using the pitch index signal, and the fixed gain signal is used to scale the fixed vector obtained using the fixed index signal.

The decoder 100 passes the linear prediction code parameters to a filter. Pass to the filter or model synthesis portion of unit 110. Subsequently, unit 110 uses the scaled vector to excite the synthesized filter using the linear prediction coefficients produced by the linear prediction coder, producing an output signal representing the digitized speech originally received at input 12. . Optionally, post filter 1
20 can be used to shape the spectrum of the digitized audio signal produced at output 20.

When data is transmitted instead of voice information, the pitch index (applied codebook output) and / or the fixed index (fixed codebook output)
Is used to receive data. The effect of non-data signals on filter synthesis by unit 110 is eliminated. This is because the gain value corresponding to the pitch or chord index is zero.

Such a functional block diagram can be implemented in many forms. Each block may be individually implemented using a microprocessor or a microcomputer, or may be implemented using one microprocessor or a microcomputer. Some or all of the functional blocks can be implemented using programmable digital signal processing devices or special purpose devices available from previous companies and other semiconductor companies.

[Brief description of the drawings]

FIG. 1 is a block diagram of a typical vocoder.

FIG. 2 is a diagram showing main functions of an encoder 14 of the vocoder 10;

FIG. 3 is a functional block diagram of a decoder unit 20 of the vocoder 10.

[Explanation of symbols]

 Reference Signs List 10 vocoder 14 encoder 20 decoder 12 audio signal 40 linear prediction coder 50 open loop pitch search unit 60 impulse response unit 72 applied codebook unit 82 fixed codebook unit 20 decoder unit 100 decoder 110 unit 120 post filter

 ──────────────────────────────────────────────────続 き Continuation of the front page (71) Applicant 596077259 600 Mountain Avenue, Murray Hill, New Jersey 07974-0636 U.S.A. S. A. (72) Inventor Stephen A. Benno United States, 07801 New Jersey, Dover, Princeton Avenue 53 (72) Inventor Michael Charles Richon United States, 07110 New Jersey, Knutsley, Pazaik Avenue 565

Claims (5)

[Claims] 1. A method of transmitting non-voice information over a voice channel, comprising: (A) transmitting non-voice information instead of pitch index information; and (B) providing a pitch gain value of substantially zero value. Transmitting. 2. The method according to claim 1, wherein the non-voice information is DTMF tone information. 3. The method according to claim 1, wherein the non-speech information is information of a text telephone for the hearing impaired. 4. A method for transmitting non-voice information over a voice channel, comprising: (A) transmitting first non-voice information instead of fixed index information; and (B) an index of substantially zero value. Transmitting the gain value. 5. The method according to claim 1, further comprising the steps of: (C) transmitting second non-voice information instead of pitch index information; and (D) transmitting a pitch gain value having a value of substantially zero. Item 5. The method according to Item 4.