Comfort noise generation for digital communication systems

• US 5,537,509 A
• Filed: 05/28/1992
• Issued: 07/16/1996
• Est. Priority Date: 12/06/1990
• Status: Expired due to Term

• Current Assignee: Hughes Network Systems LLC (Echostar Corporation)
• Original Assignee: Hughes Electronics Corporation (Rtx Corporation)
• Inventors: McCarthy, Brian M., Swaminathan, Kumar
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): SARTORI, MICHAEL A
• Application Number:
  US07/890,747
• Time in Patent Office: 1,510 Days
• Field of Search: 381/47, 381/36, 381/40, 364/717, 395/2.23, 395/2.24, 395/2.28, 395/2.29, 395/2.32, 395/2.35, 395/2.36, 395/2.37, 379/58-59, 455/50.1, 455/51.1, 455/51.2, 455/63, 455/67.3, 455/222, 455/296, 455/297
• US Class Current: 704/228
• CPC Class Codes: G10L 19/012 Comfort noise or silence coding
• CPC Class Codes: H04B 7/26 at least one of which is mobile
• CPC Class Codes: H04B 7/2643 using time-division multiple access [TDMA] H04B7/2615, H04B7/2618 take precedence
• CPC Class Codes: H04J 3/172 Digital speech interpolation, i.e. DSI
• CPC Class Codes: H04W 28/06 Optimizing the usage of the radio link, e.g. header compression, information sizing , discarding information system modifying transmission characteristic according to link quality by modifying frame length H04L1/0007; dynamic adaptation of the packet size for flow control or congestion control H04L47/365
• CPC Class Codes: H04W 72/044 based on the type of the allocated resource
• CPC Class Codes: H04W 72/20 Control channels or signalling for resource management

Abstract

A digital discontinuous cellular communication system has a transmitter that transmits two frames of data following detection of voice inactivity. A receiver includes a comfort noise generator that uses the two frames of data to output noise to the speaker during period of voice inactivity. The comfort noise generator includes synthesis codebook with samples scaled by actual background noise and excitation codebook with samples filtered and scaled by the background noise that are combined to produce comfort noise having attributes and loudness level of the received background noise prior to interruption of transmission. The scaled signals are weighted to vary the loudness level and spectral attributes.

Claims

1. A method of generating background noise during voice inactivity intervals in a communication system having a transmitter with an encoder for encoding and transmitting audio data, and a receiver remote from the transmitter with a decoder for receiving and decoding the transmitted audio data, said method comprising the steps of:

encoding audio data in the transmitter and transmitting the encoded audio data to the receiver;

detecting in the transmitter voice activity and voice inactivity;

continuing transmission of encoded audio data during a predetermined time interval following each detection of voice inactivity;

discontinuing transmission of encoded audio data at expiration of each said predetermined time interval;

resuming transmission of encoded audio data upon detection in the transmitter of voice activity;

decoding in the remote receiver the encoded audio data received from the transmitter;

detecting in the remote receiver voice activity and voice inactivity at the transmitter;

processing in the receiver the decoded audio data including data received during each said predetermined time interval after the detection of voice inactivity in the transmitter, the step of processing includes deriving a first series of output signals corresponding to an average loudness level of received noise, deriving a second series of output signals having spectral shape attributes corresponding to the received noise, combining the first and second series of derived signals to generate the audible analog audio representing background noise, the step of deriving includes weighting each of the first and second series of signals successively to vary the loudness level and spectral shape during periods of voice inactivity, the step of weighting each of the first and second series of signals includes multiplying each of the first series of signals by a first weighting factor and each of the second series of signals by a second weighting factor, the first and second weighting factors being varied to vary the loudness level and spectral shape, the first weighting factor is defined as a first variable α and

the second weighting factor is defined as a second variable 1-α

; and

repeatedly generating audible analog audio representing background noise based upon the audio data processed during each said predetermined time interval until the resumption of transmission of the encoded audio data.

2. The method of claim 1 wherein the weighting of each of the first and second series of signals successively includes repeatedly incrementing the value of the first variable α

in steps from a minimum value to a maximum value and then decrementing the first variable α

from the maximum value to the minimum value.

3. The method of claim 2 wherein the value of the first variable α

is repeatedly incremented in at least ten steps, at a rate of one step per sixty-four signals from zero to one and then decremented from one to zero at said rate.

4. A system for generating comfort noise for a digital communication system during a period of voice inactivity immediately following a period of voice activity based on received data representing background noise during said period of voice activity, comprising:

a synthesis codebook having a first table of values corresponding to long term estimates of background noise;

an excitation codebook having a second table of values corresponding to long term estimates of spectrally flattened background noise;

means including the received data during each said period of voice inactivity and values from the first table of the synthesis codebook for producing a first series of signals having a loudness level averaged over a plurality of frames of data;

means including the received data during each said period of voice inactivity and the second table of values from the excitation codebook for producing a second series of signals having spectral shape attributes corresponding to the received data;

means for combining the first and second series of signals to generate the background noise during said period of voice inactivity;

means for deriving filter coefficients from the received data, wherein the means for producing the second series of signals includes an infinite impulse response filter for filtering values from the excitation codebook second table in accordance with the derived filter coefficients;

means for scaling output signals of the infinite impulse response filter to produce the first series of signals, each having a loudness level corresponding to an average RMS level over a predetermined time period following voice inactivity and to produce the second series of signals each having a spectral shape corresponding to long term spectral shape of background noise with said loudness level; and

means for weighting the first series and second series of signals to vary the loudness level and spectral shape periodically, including means for multiplying each of the first series of signals by a first weighting factor and each of the second series of signals by a second weighting factor, the first and second weighting factors being varied to vary the loudness level and spectral shape, wherein the first weighting factor is defined as a first variable α and

the second weighting factor is defined as a second variable 1-α

.

5. The system of claim 4 wherein the value of the first value α

is repeatedly incremented in steps from a minimum value to a maximum value and then decremented from the maximum value to the minimum value.

6. A system for generating comfort noise for a digital communication system during a period of voice inactivity immediately following a period of voice activity based on received data representing background noise during said period of voice activity, comprising:

a synthesis codebook having a first table of values corresponding to long term estimates of background noise;

an excitation codebook having a second table of values corresponding to long term estimates of spectrally flattened background noise;

means including the received data during each said period of voice inactivity and values from the first table of the synthesis codebook for producing a first series of signals having a loudness level averaged over a plurality of frames of data;

means including the received data during each said period of voice inactivity and the second table of values from the excitation codebook for producing a second series of signals having spectral shape attributes corresponding to the received data;

means for combining the first and second series of signals to generate the background noise during said period of voice inactivity wherein the means for producing the first and second series of signals includes means for weighting each of the signals of the first and second series of signals to vary the spectral shape and loudness level of the background noise, including means for multiplying each of the first series of signals by a first weighting factor and each of the second series of signals by a second weighting factor, the first and second weighting factors being varied to vary the loudness level and spectral shape wherein the first weighting factor is defined as a first variable α and

the second weighting factor is defined as a second variable 1-α

.

7. The system of claim 6 wherein the value of the first variable α

is repeatedly incremented in steps from a minimum value to a maximum value and then decremented from the maximum value to the minimum value.