Patent Application
20080175407
1. Field of the Invention
The invention relates to array microphones, and more particularly to production line calibration of voice interface devices including array microphones.
2. Description of the Related Art
A single microphone only capable of receive sound from all directions with uniform gain is referred to as an omni-directional microphone. An omni-directional microphone used to receive a target voice from a single direction, simultaneously receives other surrounding noises coming from other directions. Thus, surrounding noise captured with the target voice degrades voice quality.
An array microphone including a plurality of microphones, prevents the described deficiency of an omni-directional microphone by receiving a target sound at different locations. Thus there are small differences between the phases and amplitudes of signals received by the microphones, caused by receiving sound at different locations. Thus, the array microphone can identify the target sound coming from a specific direction according to the phase and amplitude differences, and suppress surrounding noise coming from other directions. Such an array microphone is referred to as a “directional microphone”, because it is capable of capturing sound from a specific direction.
For this reason, the phase and amplitude differences of audio signals received by the microphones in an array microphone are crucial for the extraction of the target sound. The phase and amplitude differences, however, are not always caused by the differences in sound received by the microphones at different locations. The component mismatches between the microphones and the input circuits thereof also induce the phase and amplitude differences of the audio signals. For example, the capacitance difference between diaphragms of different microphones may cause a delay in the audio signals, and the resistance difference of the input circuits of the microphones may cause gain difference in the audio signals. If such phase and amplitude differences are used to extract the target sound coming from a specific direction, the derived target sound may be erroneous. Hence, the phase and amplitude differences induced by component mismatches significantly affect the performance of an array microphone. It is very difficult, however, to fabricate an array microphone with identical microphones. Thus, a method for calibrating phase and gain mismatches during fabrication of an array microphone is desirable.
The invention provides a system for calibrating phase and gain mismatches of an array microphone. The array microphone is installed in a voice interface device and comprises a plurality of microphones. The system comprises a loudspeaker and a computing equipment. The loudspeaker plays a segment of sound to be received by the array microphone. The computing equipment controls the voice interface device which converts the segment of sound to a plurality of audio signals with the microphones of the array microphone, records the audio signals outputted by the voice interface device at bypass mode without any signal processing, calculates delays between the audio signals, and instructs the voice interface device to adjust phase mismatches between the audio signals according to the delays.
The invention also provides a method for calibrating phase and gain mismatches of an array microphone. The array microphone is installed in a voice interface device and comprises a plurality of microphones. First, a segment of sound to be received by the array microphone is played. The voice interface device is then controlled to bypass audio signals converted from the sound by the microphones of the array microphone. The audio signals output by the voice interface device are then recorded. Correlation coefficients based on correlation of the audio signals is then calculated. Delays between the audio signals are then determined according to the correlation coefficients. Finally, the voice interface device is instructed to adjust phase mismatches between the audio signals according to the delays.
A detailed description is given in the following embodiments with reference to the accompanying drawings.
The invention can be more fully understood by reading the subsequent detailed description and examples with references made to the accompanying drawings, wherein:
The following description is of the best-contemplated mode of carrying out the invention. This description is made for the purpose of illustrating the general principles of the invention and should not be taken in a limiting sense. The scope of the invention is best determined by reference to the appended claims.
In addition to the microphone array 110, the voice interface device 100 also includes two microphone input circuits 122 and 132, two analog to digital converters 124 and 134, a digital signal processor 126, a memory 128, a digital I/O interface 142, and a control I/O interface 144. The omni-directional microphones 112 and 114 first respectively convert a received sound to audio signals X1 and Y1. The audio signals X1 and Y1 are then respectively amplified and filtered by the microphone input circuits 122 and 132 to obtain the audio signals X2 and Y2, which are further converted to digital audio signals X3 and Y3 by analog to digital converters 124 and 134.
The digital signal processor 126 can then process the audio signals X3 and Y3 to obtain the audio signals X4 and Y4 according to instructions of the computing equipment 106. The computing equipment 106 is connected to the voice interface device 104 via two interfaces: the digital I/O interface 142 and the control I/O interface 144. The audio signals X4 and Y4 can be transmitted to the computing equipment 106 through the digital I/O interface 142. The computing equipment 106 sends instructions to control the digital signal processor 126 via the control I/O interface 144. Although the array microphone 110 includes only two omni-directional microphones, the system 102 can be used to calibrate a voice interface device 104 including a microphone array containing more than two omni-directional microphones.
To illustrate the calibration process of the system 100, a method 200 for calibrating phase and gain mismatches of array microphones according to the invention is provided in
The recorded audio signals X4 and Y4 are then analyzed by the computing equipment 106 in two different analysis paths. One analysis path 210 is to determine the phase mismatch between the audio signals X4 and Y4, and the other analysis path 220 is to determine the gain mismatch between the audio signals X4 and Y4. With regard to phase mismatching, because the sampling rate of analog to digital converters 124 and 134 may be lower, the computing equipment 106 first interpolates the audio signals in step 210 to increase the sampling rate of the audio signals fitting the requirement for delay calculation with enough precision. The interpolated audio signals are then used to calculate cross-correlation coefficients in step 214. A delay between the samples of the audio signals can then be determined according to the correlation coefficients in step 216. Because the loudspeaker 108 is separated by the same distance from microphones 112 and 114, the sound is delayed by the same amount prior to reception by the microphones, thus, no phase mismatching exists between the audio signals. Thus, the delay between the audio signals is caused completely by component mismatch of the microphones themselves, the input circuits thereof, and the ADCs. A set of predetermined delay values may be stored in the memory 128 in advance, and a delay index can be determined in step 218 to select a delay value nearest the delay calculated in step 216 from the set of delay values. Thus, after the delay index is delivered to the digital signal processor 126, the digital signal processor 126 can then delay the samples of the audio signals X3 or Y3 according to the delay index, and the audio signals X4 and Y4 without phase mismatching.
The gain mismatch is determined in the analysis path 220. The computing equipment 106 first measuring the powers of the audio signals X4 and Y4 in step 222. The measured powers are then smoothed in step 224 to obtain average powers of the audio signals. Because the loudspeaker 108 is separated from the microphones 112 and 114 by the same distance, the sound suffers the same amount of attenuation before being received by the microphones, thus, no amplitude mismatching exists between the audio signals. Thus, the power difference between the audio signals is caused completely by component mismatching of the microphones, the input circuits thereof, and the ADCs. A gain value can then be determined according to the smoothed powers in step 226. After the gain value is delivered to the digital signal processor 126, the digital signal processor 126 can then amplify the samples of the audio signals X3 or Y3 according to the gain value to compensate for the gain mismatch, and the audio signals X4 and Y4 without gain mismatching is obtained.
Moreover, the delay and the gain calculated in steps 218 and 226 can be further used to determine a set of filtering coefficients for compensating the phase and gain mismatches of the audio signals X3 and Y3. The filtering coefficients can be stored in the memory 128, and the digital signal processor 126 then filters the audio signals X3 and Y3 according to the filtering coefficients to obtain the audio signals X4 and Y4 without phase and gain mismatches. In one embodiment, multiple sets of filtering coefficients are stored in the memory 128 in advance, and the computing equipment 106 simply determines a filtering coefficient index which selects an appropriate set of filtering coefficients from the multiple sets of filtering coefficients, and the digital signal processor 126 can then filter the audio signals X3 and Y3 according to the filtering coefficient index to remove the phase and gain mismatches.
The invention provides a method for calibrating phase and gain mismatches of an array microphone. Because the phase and gain mismatches are calibrated when array microphones are fabricated, signals generated by the array microphones will not comprise the delay and attenuation caused by component mismatches of the microphones and the input circuits thereof. Thus, beam-forming can be precisely performed to extract in-band sounds coming from specific directions and suppress the out-of-band noise, and the performance of the voice interface devices including the array microphones is enhanced.
While the invention has been described by way of example and in terms of preferred embodiment, it is to be understood that the invention is not limited thereto. To the contrary, it is intended to cover various modifications and similar arrangements (as would be apparent to those skilled in the art). Therefore, the scope of the appended claims should be accorded the broadest interpretation so as to encompass all such modifications and similar arrangements.
