System and method for improving message delivery in voice systems utilizing microphone and target signal-to-noise ratio

The present invention relates to a system and a method for delivering voice messages, and more specifically, to a system and a method for improving message delivery in voice systems utilizing a microphone and a target Signal-to-Noise Ratio (SNR).

Audio system messages in environments such as an automobile may be affected by both system components and external factors. The system components include, for example, sounds from the auto's radio or noise carried into the auto when the windows are open. The external factors include, for example, the noise caused when a baby is crying in the back seat or a freight train is passing in front of the car. While the system can possibly adjust the system components (such as by turning off the radio or closing the windows), it may be an annoyance to the end user. In addition, the external factors cannot be controlled by the system and may affect the Speech Intelligibility (SI) of the voice system.

Currently, systems attempt to make spoken information clearer by taking actions such as temporarily muting the radio or automatically adjusting the volume of a car radio depending on the level of engine noise. Such actions, however, are typically not sufficient to control external factors. They can also change the state of the system in ways the user may not want. Moreover, conventional techniques intended to make spoken information clearer generally do not take advantage of information provided by microphones typically found in voice systems. In addition, speaker placement is not fixed for some voice systems (such as an automated house) so delivery of the message cannot be guaranteed. For users to adopt voice systems critical information should be delivered with certainty. However, an overall solution has not been developed to solve the above problems.

One aspect of the present invention is a method for delivering a message to a recipient in an environment with ambient noise. The method includes recording the ambient noise in the environment at a certain time interval, analyzing the recorded ambient noise to obtain an average power P noise or RMS amplitude A noise of the ambient noise, providing a predetermined desired SNR desired , calculating an average signal power P signal or RMS amplitude A signal of the message to be delivered based on the P noise or A noise and the desired SNR desired , and adjusting a volume of the message to be delivered according to the P signal or A signal .

Another aspect of the invention also provides a method for delivering a message to a recipient in an environment with ambient noise. The method includes the steps of delivering a message, recording audio at or near the recipient, analyzing the recorded audio to obtain an actual SNR actual , providing a predetermined minimum SNR min , and repeating the message if the actual SNR actual falls below the SNR min .

Yet another aspect of the invention is a system for delivering a message to a recipient in an environment with ambient noise. The system includes a recording unit for recording the ambient noise in the environment at a certain time interval, an analyzing unit for analyzing the recorded ambient noise to obtain an average power P noise or RMS amplitude A noise of the ambient noise, means for providing a predetermined desired Signal-to-Noise Ratio SNR desired , a calculating unit for calculating an average signal power P signal or RMS amplitude A signal of the message to be delivered based on the P noise or A noise and the desired SNR desired , and an adjusting unit for adjusting a volume of the message to be delivered according to the P signal or A signal .

The present invention also provides a system for delivering a message to a recipient in an environment with ambient noise, which includes a delivering unit for delivering the message, a recording unit for recording audio at or near the recipient when the message is delivered, an analyzing unit for analyzing the recorded audio to obtain an actual SNR actual , means for providing a predetermined minimum Signal-to-Noise Ratio SNR min , and means for repeating the message if the actual SNR actual falls below the SNR min .

A further aspect of the present invention is a computer-readable media in which is stored computer-executable instructions for carrying out a method for delivering a message to a recipient in an environment with ambient noise. The method includes the steps of recording the ambient noise in the environment at a certain time interval, analyzing the recorded ambient noise to obtain an average power P noise or RMS amplitude A noise of the ambient noise, providing a predetermined desired Signal-to-Noise Ratio SNR desired , calculating an average signal power P signal or RMS amplitude A signal of the message to be delivered based on the P noise or A noise and the desired SNR desired , and adjusting a volume of the message to be delivered according to the P signal or A signal .

The present invention also provides a computer-readable media in which is stored computer-executable instructions for carrying out a method for delivering a message to a recipient in an environment with ambient noise. The method includes the steps of delivering a message, recording audio at or near the recipient, analyzing the recorded audio to obtain an actual Signal-to-Noise Ratio SNR actual , providing a predetermined minimum Signal-to-Noise Ratio SNR min , and repeating the message if the actual SNR actual falls below the SNR min .

There are shown in the drawings, embodiments which are presently preferred. It is expressly noted, however, that the invention is not limited to the precise arrangements and instrumentalities shown.

FIG. 1 is a schematic illustration of one embodiment of a system for delivering a message to a recipient in an environment with ambient noise according to the present invention.

FIG. 2 is a schematic illustration of another embodiment of a system for delivering a message to a recipient in an environment with ambient noise according to the present invention.

FIG. 3 is a diagram showing a defined history of noise selected and analyzed in an example of noise recorded in a car being surrounded by loud noise.

FIG. 4 is a plot showing that non-constant features of audio are discarded.

FIG. 5 is a chart showing a statistical analysis of environmental noise.

FIG. 6 is a schematic diagram of a floor plan of a living room as another example of voice environment.

FIG. 7 is a flow chart of exemplary steps for delivering a message to a recipient in an environment with ambient noise, according to one embodiment of the present invention.

FIG. 8 is a flow chart of exemplary steps for delivering a message to a recipient in an environment with ambient noise, according to another embodiment of the present invention.

The present invention continuously monitors the ambient noise in the environment of a voice system even when a Push-to-Speak button of the voice system is not pressed. This measurement typically will be measured in decibels. In one embodiment, the weighted average of ambient noise would be maintained over a window of a fixed interval. The interval can, for example, be 20 seconds. Other intervals are possible depending on the circumstances. When the system delivers information to the user, the volume can be adjusted to a level which has a satisfactory SNR. This can provide as close as possible 100% certainty that the message has the adequate SI. The system is assumed not to be processing commands until the Push-to-Speak button is pressed. This mode will be referred to as Passive Monitoring Mode (PMM). This adjustment of volume would need to occur after analyzing the average power of the signal to be delivered.

SNR is defined as the ratio of a given transmitted signal to the background noise of the transmission medium. Because many signals have a very wide dynamic range, SNRs are usually expressed in terms of the logarithmic decibel scale. In decibels, the SNR is 20 times the base-10 logarithm of the amplitude ratio, or 10 times the logarithm of the power ratio:

SNR ⁡ ( dB ) = 10 ⁢ ⁢ log 10 ⁡ ( P signal P noise ) = 20 ⁢ ⁢ log 10 ⁡ ( A signal A noise ) ( 1 )

where P is average power and A is RMS amplitude. This equation can be solved for A signal or P signal which are directly related to the RMS amplitude. The known variables in the equation would be P noise or A noise and SNR desired .

The present invention further provides a system and a method which expands upon the above system and method by computing SNR actual . This is achieved through utilizing the microphone at the time the audio message is delivered. Since the noise level in the environment can and will suddenly change, the SNR actual could differ significantly from SNR desired , which is based on the data collection in the frame of 20 previous seconds. In one embodiment of this method, the message could be repeated if SNR actual , falls below certain critical criterion, such as SNR min .

FIG. 1 schematically illustrates a system for delivering a message to a recipient in an environment with ambient noise according to one embodiment of the present invention. As can be seen in FIG. 1 , the system 100 includes a recording unit 101 for recording the ambient noise in the environment at a certain time interval; an analyzing unit 102 for analyzing the recorded ambient noise to obtain an average power P noise or RMS amplitude A noise of the ambient noise; means 103 for providing a predetermined desired Signal-to-Noise Ratio SNR desired ; a calculating unit 104 for calculating an average signal power P signal or RMS amplitude A signal of the message to be delivered based on the P noise or A noise and the desired SNR desired ; and an adjusting unit 105 for adjusting a volume of the message to be delivered according to the P signal or A signal .

FIG. 2 schematically illustrates a system for delivering a message to a recipient in an environment with ambient noise according to another embodiment of the present invention. As can be seen in FIG. 2 , the system 200 includes a delivering unit 201 for delivering a message; a recording unit 202 for recording audio at or near the recipient when the message is delivered; an analyzing unit 203 for analyzing the recorded audio to obtain an actual SNR actual ; means 204 for providing a predetermined minimum Signal-to-Noise Ratio SNR min ; a comparing unit 205 for comparing the actual SNR actual with the SNR min ; and means 206 for repeating the message if the actual SNR actual falls below the SNR min . The means for repeating the message can be the same device as the delivering unit or a different device at a different location.

The system for improving message delivery as described above can be implemented within the voice system (integrated with the voice system) or can be implemented external to the voice system. The latter provides more flexibility, meaning such a system can be used together with a variety of voice systems.

FIG. 3 shows, as an example, a defined history of noise selected and analyzed in an extreme example of noise recorded in a car being surrounded by loud noise. The noise levels in the car will be monitored and computed in a time interval of about 10-30 seconds, preferably 20 seconds. When a message is to be delivered, the defined window of background data could be analyzed by known methods. First, the last 20 seconds of data would be considered. In one embodiment, all the data would be analyzed for RMS noise . In an alternate embodiment, the data would eliminate the extremes to discard singular spikes (such as the door slamming as a passenger gets in). This could be accomplished by discarding the most extreme 5% of the data (see FIG. 4 ). In either case, known methods would be applied to compute RMS noise .

Equation (1) would subsequently be solved for A signal and an amplification of the delivered message would occur through known methods in order to achieve the SNR min . At the time of delivery, record the delivery of the message to compute SNR actual . If this value falls below SNR min then the message is repeated (if necessary, indicating it is a repetition by prefixing the message with a keyword such as “Again . . . ”). Microphone placement should be at or near the location of the intended recipient.

FIG. 5 shows a statistical analysis of environmental noise. An average power P noise or RMS amplitude A noise of the noise can be obtained from this analysis.

FIG. 6 depicts a floor plan of a living room, another type of voice environment. Possible sources of noise which could be controlled by the system are the fan, radio, and television. Possible sources outside control of the system are the piano, people in the room, or a vacuum cleaner being operated within the room. Speaker placement may be variable so the microphone at or near the center of the room could be used to calculate both SNR desired and SNR actual .

FIG. 7 is a flow chart of exemplary steps for delivering a message to a recipient in an environment with ambient noise, according to one embodiment of the present invention. As shown in FIG. 7 , first, at step 702 , the ambient noise in the environment is recorded at a certain time interval. The recorded ambient noise is then analyzed, at step 704 , to obtain an average power P noise or RMS amplitude A noise of the ambient noise. Subsequently, at step 706 , an average signal power P signal or RMS amplitude A signal of the message to be delivered is calculated based on the P noise or A noise and a predetermined desired SNR desired . Finally, at step 708 , a volume of the message to be delivered is adjusted according to the P signal or A signal .

FIG. 8 is a flow chart of exemplary steps for delivering a message to a recipient in an environment with ambient noise according to another embodiment of the present invention. More specifically, FIG. 8 shows the process of determining if message needs to be redelivered. FIG. 8 illustrates the possible iterative nature of determining if a message has been properly delivered to the recipient. Due to the dynamic nature of a speech system's environment, it may be desirable to say the message a few times until it is certain that it is delivered.

As shown in FIG. 8 , first, at step 801 , a voice message is delivered. Then, at step 803 , the audio at or near the recipient is recorded and, at step 805 , the SNR actual calculated. If the SNR actual is greater than the SNR min , the system, at step 807 , will wait to deliver the next message. If, however, the SNR actual is smaller than the SNR min , the system will, at step 809 , repeat the message, preferably with a keyword before it.

In another embodiment of the method, the system can calculate the SNR and adjust the volume of TTS in real-time based on a sliding window of the last x seconds of audio. The benefit of this approach is that the message would not have to be repeated, but would require more calculations.

By using the systems and methods of the present invention, the message will be delivered to the user with certainty and with adequate SI without any discomfort of the user. Further advantages of the invention can be seen from the above description and the associated drawings.

The invention can be realized in hardware, software, or a combination of hardware and software. The invention can be realized in a centralized fashion in one computer system, or in a distributed fashion where different elements are spread across several interconnected computer systems. Any kind of computer system or other apparatus adapted for carrying out the methods described herein is suited. A typical combination of hardware and software can be a general purpose computer system with a computer program that, when being loaded and executed, controls the computer system such that it carries out the methods described herein.

The invention can be embedded in a computer program product, which comprises all the features enabling the implementation of the methods described herein, and which when loaded in a computer system is able to carry out these methods. Computer program in the present context means any expression, in any language, code or notation, of a set of instructions intended to cause a system having an information processing capability to perform a particular function either directly or after either or both of the following: a) conversion to another language, code or notation; b) reproduction in a different material form.

The foregoing description of preferred embodiments of the invention has been presented for the purposes of illustration. The description is not intended to limit the invention to the precise forms disclosed. Indeed, modifications and variations will be readily apparent from the foregoing description. Accordingly, it is intended that the scope of the invention not be limited by the detailed description provided herein.