Patent Grant
7272552
This application claims the benefit of the following co-pending application: System and Method for Improved Us of Voice Activity Detection, Ser. No. 10/331,013, filed on Dec. 27, 2002.
This invention relates generally to telecommunications. More particularly, this invention relates to the improved use of voice activity detection/silence suppression technology.
Network provides, like AT&T, are continually working to improve network efficiency. Consider a typical transmission from a calling party to called party in which noise signals, accompanied by noise, are transmitted. The noise component considered in this application is primarily the background noise of the speaking party. During this typical transmission there are periods of time when the calling party is not speaking but the background noise is still present. Even though no voice signal is being transmitted, the network is still committing the same amount of resources, and transmitting solely the noise signal. Conventional technology has made the network more efficient by reducing the amount of resources allocated to such a transmission during periods of “silence,” i.e., when only a noise signal is present, so that the bandwidth can be used for other transmissions.
Conventional technology detects a voice signal in a transmission. This technology, referred to as voice activity detection (VAD) or silence suppression (SS), determines whether an input signal is primarily a voice signal or a noise signal based on one or more parameters. This decision may be based on the current frame as well as a few of the preceding frames, to ensure that there is a significant break in the input voice signal. When the VAD/SS technology determines that no input voice signal is being transmitted, i.e., that just noise is present, instead of using a suitable amount of bandwidth for the particular transmission, the VAD/SS technology informs the receiving end that no signal is going to be transmitted. The VAD/SS technology sends one or more bits, referred to as silence insertion descriptors, that are noise characteristics. The VAD/SS technology, instead of transmitting the noise, or the “lack of voice” over the bandwidth, allocated the bandwidth to another use. At the receiving end, the silence insertion descriptors are converted into a representation of the background noise, also known as comfort noise, representation and transmitted to the called party.
In the event that the silence period continues for some time, and there is no significant change in the background noise, then the comfort noise is continually generated. However, if there is a change in the background noise, new silence insertion descriptors will be transmitted to the receiving end. This process avoids a comfort noise that is constant for the benefit of the listener.
Although there is bandwidth savings with this technology, suppressing silence in a transmission has another associated characteristic. Suppressing the silence “degrades” the quality of the connection. The parties participating in the transmission become familiar with each voice and the associated background noise. During periods of silence when the silence suppression system either transmits no signal or silence insertion descriptors, the parties will hear a difference in background noise, i.e., either no noise at all or a slightly different background noise. Noise pumping refers to the different sounds created when switching between the presence of noise and absence of noise. Noise contrast refers to the different sounds created when switching between the presence of actual noise and the presence of comfort noise. As noted above, the comfort noise results when the silence insertion descriptors are converted into a noise signal, and this noise is generally not identical to the actual noise. Both the noise pumping and the noise contrast yield audible distinctions that are apparent and undesirable to the parties.
Network providers, like AT&T, are interested in using network bandwidth efficiently. This is especially important where bandwidth is limited, such as sub-oceanic fiber cable transmissions given the large cost to place the cable. Network providers are also interested in providing the best quality transmissions possible. Thus, given the foregoing, there is a need in the industry to maintain the bandwidth savings using silence suppression techniques while improving the quality of the signal to the users.
Limitations of the prior art are overcome and a technical advance is made in accordance with the present invention, illustrative embodiments of which are described below.
In general, the present invention is a method and apparatus for improved silence suppression by packetizing voice signals received by an access gateway from a speaking party and transmitting these packetized voice signals via a network to an egress gateway. The egress gateway converts the packetized voice signal into voice signals suitable for output and transmits the output voice signals to a listening party. The invention also includes packetizing actual noise signals, typically background noise, received by an access gateway from a speaking party and transmitting these packetized noise signals via a network to an egress gateway. The egress gateway converts the packetized noise signal into noise signals suitable for output and transmits the output noise signals to a listening party. When the access gateway detects that no voice signal is being received and only a noise signal is being received for a predetermined period of time, the access gateway instructs the egress network to continually transmit output noise signals to the listening party and ceases to transmit packetized noise signals to the egress gateway.
In accordance with aspects of one illustrative embodiment, a silence suppression system is disclosed comprising an access gateway, including an access silence suppressor, a network and an egress gateway, including an egress silence suppressor. The access gateway receives an input voice signal, that includes noise, and also received an input noise signal, that includes no voice. The access gateway packetizes the input voice signal and transmits the packetized voice signal over a network. The access gateway also packetizes the input noise signal and transmits the packetized noise signal over the network. An egress gateway receives the packetized voice signal, converts the packetized voice signal into an output voice signal, and outputs the output voice signal. The egress gateway also receives the packetized noise signal, converts the packetized noise signal into an output noise signal, and outputs the output noise signal. After a predetermined amount of time, if the access gateway continually receives the input noise signal, and not the input voice signal, the access gateway ceases the transmit the packetized noise signal through the network and instructs the egress gateway to continually output the output noise signal.
In an alternative embodiment, the present invention is a silence suppression system for a telephone call that includes an access network, including an access silence suppressor, a network, and an egress network, including an egress silence suppressor. The access network receives an input voice signal (including noise) and converts the input voice signal into a packetized voice signal. The packetized voice signal is transmitted via a network to the egress network. The egress network receives the packetized voice signal, converts the packetized voice signal into an output voice signal, and outputs the output voice signal. The egress network also extracts and stores noise packets from the received packetized voice signal and converts the packetized noise signal into an output noise signal. When the access network ceases to receive the input voice signal while the call is still ongoing, the access network instructs the egress network to continually output the output noise signal.
The present invention will be more fully understood by reading the following detailed description in conjunction with the drawing, in which:
More specifically, calling party 102 places a telephone call to called party 104 and the connection is established. Calling party 102 speaks a voice signal that will be transmitted through network 110. In addition to the voice signal, any noise, such as background noise, present in the calling party's vicinity is also transmitted through network 110. Likewise, voice signals and noise signals are also transmitted through network 110 in the reverse direction, from called party 104 to calling party 102. A conversation thus begins between calling party 102 and called party 104. At certain times during the conversation, calling party 102 will not be sending a voice signal. The noise signal, however, is continually present and continues to be transmitted.
One embodiment of the present invention is as follows. Access network 106 and egress network 108 includes a VAD/SS device. The VAD/SS device determines when a voice signal is present and when a voice signal is not present as is known in the art. For example, access network 106 monitors the energy of the input signal received. A certain energy pattern (e.g., level and frequency) is associated with a voice signal and a certain energy pattern is associated with a noise only signal. Vendors use proprietary methods for distinguishing between voice and noise only signals.
When the voice signal is not present, the VAD/SS device instructs a packet representing the noise signal to be created. The noise packet contains a sample of the actual noise signal, as opposed to approximate characteristics of the noise. The noise packet is transmitted through network 110 to egress network 108. Egress network 108 is instructed to convert the noise packet into an output noise signal and to continually transmit the output noise signal to called party 104 until a voice signal is present.
In order to prevent clipping of the input voice signal, the VAD/SS method monitors the input signal for voice for a period of time after the voice input ceases, referred to as a hang-over time. In other words, when the voice signal is not present in the input signal, and noise is the only input signal, the VAD/SS device continues to transmit the packetized noise signal through network 110 for an extra period of time. This methodology ensures that there is not voice signal being input, as opposed to a small pause, for example, As a result, this methodology prevents the called party from hearing the calling party's voice cut in an out, so that called party does not miss the beginning portion of voice signals from the calling party.
In an alternate embodiment of the present invention egress network 108 extracts and stores noise packets from the received voice signal. In a typical conversation, calling party 102 is not sending a continuous voice signal to called party 104. There are natural breaks or pauses in the voice signal when only a noise signal is being transmitted. During these portions of time when no voice, and only noise is being transmitted to egress network 108, egress network 108 extracts and stores noise packets. When egress network 108 is informed by access network 106 that there is no voice signal being transmitted through network 110, egress network 108 converts the stored samples of the actual noise to an output noise signal and continually transmits the output noise signal to called party. When an input voice signal is again received at egress network 108, the transmission of the stored noise signal to called party ceases and the input voice signal is transmitted to the called party as described above.
Although the present invention has been described as calling party 102 initiating the input voice signal and called party 104 received the voice signal, one skilled in the art will appreciate that called party 104 also initiates voice signals that are received by calling party 102.
Although embodiments of the present invention are illustrated in the accompanying FIGURES and are described in this Detailed Description, it is understood that the present invention is not limited to these embodiments, but is capable of numerous arrangements, modifications, and substitutions without departing from the spirit and scope of the invention as defined in the claims. Various modifications and adaptations of the present invention will be apparent to persons skilled in the art.
| Number | Name | Date | Kind |
|---|---|---|---|
| 5897613 | Chan | Apr 1999 | A |
| 6243366 | Bradley et al. | Jun 2001 | B1 |
| 6360265 | Falck et al. | Mar 2002 | B1 |
| 6526140 | Marchok et al. | Feb 2003 | B1 |
| 6643617 | Wood et al. | Nov 2003 | B1 |
| 20020006137 | Rabenko et al. | Jan 2002 | A1 |
| 20030055627 | Balan et al. | Mar 2003 | A1 |
| 20060069551 | Chen et al. | Mar 2006 | A1 |
