JP2009153053A - Voice estimation method, and mobile terminal using the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To specify a noise signal and an objective sound signal including the noise signal to normally operate an adaptive noise-suppressing filter, and to suppress the noise in the objective sound signal. <P>SOLUTION: A noise and objective-sound component processor 4 estimates the signal with an incoming-call component removed from a sound signal as the noise signal, based on a voice signal obtained by picking up an incoming call through microphones 1a, 1b and receipt sound data. During a following telephone conversation after that, the noise and objective-sound component processor 4 estimates an added voice signal through a predetermined time-difference correction as the objective sound signal including the noise component, in association with the voice signal cosisting of a telephone-call voice picked up through the microphones 1a, 1b and a background noise. An adaptive filter 5 receives these estimated noise signal and the objective sound signal including the noise component as an input, and outputs the objective sound signal (transmitting voice signal) with the noise component suppressed therefrom. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a speech estimation method and a mobile terminal using the same, and more particularly to a speech estimation method using an adaptive noise suppression filter and a mobile terminal using the same.

  In a portable terminal such as a cellular phone, it is necessary to suppress ambient noise such as background sound as much as possible when an incoming call or a call is made. For example, when an incoming call is received, the characteristics of the ambient noise (ambient noise) input from the microphone is analyzed, and the ambient noise parameters distributed and stored in the storage means are analyzed according to the analyzed ambient noise characteristics. It is known that a telephone device that can identify ringtones without being influenced by ambient noise by selecting the optimum ambient noise parameters and using them to control the tone quality of ringtones generated from the sound source. (For example, refer to Patent Document 1).

  In addition, a signal processing is performed on an acoustic signal obtained by using a plurality of microphones arranged at intervals, thereby suppressing noise and emphasizing a target acoustic signal component. An adaptive noise suppression filter is also known (see, for example, Patent Document 2). This adaptive noise suppression filter is an efficient noise suppression system that combines an addition type microphone array for ambient noise and a subtraction type array for directional noise. Therefore, by incorporating this adaptive noise suppression filter into a portable terminal, it is possible to suppress ambient noise during a call as much as possible.

JP 2001-036604 A JP 2003-223198 A

  However, the portable terminal described in Patent Document 1 analyzes the nature of ambient noise in order to identify ringtones, and cannot suppress noise signals during a call using an adaptive noise suppression filter.

  On the other hand, the adaptive noise suppression filter requires a target sound signal including noise and a noise signal as input signals, and can output a target sound signal in which noise components are suppressed for these input signals. However, when an adaptive noise suppression filter is incorporated in a mobile terminal, not only the transmitted speech signal that is the target sound signal but also the noise signal is simultaneously input from the microphone array. Since the target sound signal and the noise signal including the noise component cannot be specified, desired noise suppression cannot be performed.

  The present invention has been made in view of the above points, and it is possible to specify a target sound signal including a noise component and a noise signal, and operate an adaptive noise suppression filter normally to suppress the noise in the target sound signal. An object is to provide an estimation method and a portable terminal using the estimation method.

  In order to achieve the above object, a speech estimation method of the present invention includes a first step of collecting a ringtone at a telephone terminal with a plurality of microphones provided on the telephone terminal, and a plurality of microphones at the first step. And a second step of estimating a voice signal from which a ring tone component is removed from a plurality of voice signals obtained by collecting a ring tone as a noise signal.

  In order to achieve the above object, a mobile terminal of the present invention is a mobile terminal including a plurality of microphones and an adaptive noise suppression filter, and an incoming call that causes a speaker to ring a ringtone based on ringtone data. Based on the sound ringing means, the plurality of sound signals obtained by collecting the ringtones with a plurality of microphones, and the ringtone data, the ringtone component is obtained from the plurality of sound signals obtained by collecting the ringtones. Noise signal calculation means for calculating the removed speech signal to estimate it as a noise signal and outputting it to an adaptive noise suppression filter is provided.

  According to the present invention, it is possible to perform noise suppression with a high degree of freedom by using an adaptive noise suppression filter that is supplied with an estimated noise signal and a target sound signal as input signals.

  Next, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 shows a block diagram of an embodiment of a portable terminal according to the present invention. In the figure, a mobile terminal 1 includes two microphones 2a and 2b for transmission, AD converters 3a and 3b for separately converting analog audio signals output from the microphones 2a and 2b into digital audio signals, A noise / target sound component calculation unit 4 that performs calculation processing for specifying a target sound component and a noise component for the digital audio signals output from the AD converters 3a and 3b, and a noise / target sound component calculation unit 4 An adaptive noise suppression function (AMNOR; Adaptive Microphone) that receives a target sound signal including a noise component and a noise signal output from the signal and outputs a target sound signal in which the noise component is suppressed by performing known noise suppression processing for NOise Reduction) (hereinafter referred to as an adaptive filter) 5 and a ringtone ringing speaker 6.

  Note that the arrangement positions of the microphones 2a and 2b on the portable terminal use the time difference when the transmitted voice is input to the microphones 2a and 2b, so that they are separated from the receiver (speaker different from the speaker 6). Although it is desirable, there are no particular restrictions.

  FIG. 2 is a block diagram of an embodiment of the noise / target sound component calculation unit 4 which is a main part of the present embodiment. As shown in FIG. 2, the noise / target sound component calculation unit 4 includes a random access memory (RAM) 41a for temporarily storing digital audio signals output in parallel from the AD converters 3a and 3b separately. 41b, the RAM 43 for storing the ring tone data, the digital voice signal read from the RAMs 41a and 41b and the ring tone data read from the RAM 43 as inputs, and performing these correlation calculations and matrix addition / subtraction, It comprises a digital signal processor (DSP) 42 that outputs the obtained target sound component and noise component. The RAMs 41a and 41b have a ring buffer configuration because the sound collection times of the microphones 2a and 2b are not always constant. The RAM 43 also has a ring buffer configuration like the RAMs 41a and 41b.

  The adaptive filter 5 in FIG. 1 is well known to those skilled in the art and is not directly related to the present invention, so that the detailed description of the configuration is omitted.

  Next, the operation of this embodiment will be described. In the present embodiment, first, an estimated signal of a target sound signal and a noise signal is prepared from the start of outgoing / incoming calls to the start of a call. This preparation operation will be described with reference to the flowchart of FIG. 3 and FIGS.

  First, when the mobile terminal 1 enters a voice call incoming state (step S1 in FIG. 3), the microphones 2a and 2b are sounded by electro-acoustic conversion by a speaker 6 to which a ring tone signal is supplied by a known method. Picking up the ringtone and ambient noise is started (step S2 in FIG. 3). As a result, the sound signals relating to the ring tone and ambient noise obtained by the acoustic-electric conversion by the microphones 2a and 2b are supplied to the RAMs 41a and 41b in the noise / target sound component calculation unit 4 through the AD converters 3a and 3b. And temporarily stored. On the other hand, the ring tone data supplied to the speaker 6 is also supplied to the RAM 43 and temporarily stored.

  In this state, when the sender presses a call start button (not shown in FIG. 1) (step S3 in FIG. 3), the DSP 42 in the noise / target sound component calculation unit 4 refers to the RAM 43 and receives the ring tone data. Is stored (that is, whether or not the ringing tone is ringing) (step S4 in FIG. 3). When no ring tone data is stored in the RAM 43, it is determined that no ring tone has been played, and the data stored in the RAMs 41a and 41b is estimated to be a noise signal (step S6 in FIG. 3).

  On the other hand, when the ringtone data is stored in the RAM 43, it is determined that the ringtone is ringing, and the DSP 42 determines the time between the data stored in the RAMs 41a and 41b and the ringtone data stored in the RAM 43. The gain of the ringtone data stored in the RAMs 41a and 41b in the same phase as the ringtone data stored in the RAM 43 is calculated (the loopback when output from the speaker 6 and input to the microphones 2a and 2b). The ringtone is removed by subtracting the gain of the ringtone data in the RAM 43 from the gain of the data stored in the RAMs 41a and 41b (step S5 in FIG. 3). The value is estimated (specified) as a noise signal (step S6 in FIG. 3).

  After the above noise signal estimation, a call between the mobile terminal 1 of the present embodiment and the counterpart terminal is started. The operation after the start of the call will be described with reference to the flowchart of FIG. 4 and FIGS. Note that the noise / target sound component calculation unit 4 can assume that the sound source of the target sound is in front of the mobile terminal 1 during a call. It is assumed that time difference information of the audio signal is stored.

  First, when ringing is stopped and a voice call is started (step S11 in FIG. 4), the microphones 2a and 2b not only transmit the target voice but also ambient noise (background noise) at the same time. The collected sound is converted into an audio signal which is an electric signal, converted into a digital audio signal by the AD converters 3 a and 3 b, and then supplied to the noise / target sound component calculation unit 4.

  The noise / target sound component calculation unit 4 calculates in advance a first digital audio signal during a call collected by the microphone 2a and a second digital audio signal during a call collected by the microphone 2b. After adding the time difference information in the front direction and correcting the time difference to eliminate the time difference (step S12 in FIG. 4), the first and second digital audio signals are added (step S13 in FIG. 4). Then, the digital audio signal after the addition is estimated (specified) as a target sound component (target sound signal) including ambient noise (step S14 in FIG. 4).

  The adaptive filter 5 receives as input signals the noise signal estimated by the noise / target sound component calculation unit 4 and the target sound signal including ambient noise as described above, and performs a known filtering process based on these input signals. To output a target sound signal in which ambient noise is suppressed. This suppresses, for example, ambient noise during a voice call (for example, a sound of a TV receiver or the like that is ringing behind the sender, or a background sound that is not necessary for a telephone call such as a street speech), and the target sound is suppressed. The voice signal of the transmitted voice can be output clearly.

  As described above, in the present embodiment, the noise signal and the target sound signal including the noise component are estimated in the apparatus, the noise signal and the target sound signal including the noise are specified, and the adaptive filter 5 operates normally. Noise suppression with a high degree of freedom.

  In addition, this invention is not limited to said embodiment, For example, the number of microphones may be three or more, and those microphones can be arrange | positioned freely. However, phase difference information in the front direction is required in advance for target sound estimation. In addition to pressing the call start button, for example, when the mobile terminal 1 is a foldable mobile terminal, the apparatus can recognize the start of the call by detecting that the two housings are opened. Furthermore, the present invention can be applied to fixed telephones other than portable terminals.

It is a block diagram of one Embodiment of the portable terminal of this invention. It is a block diagram of one Embodiment of the noise and the target sound component calculating part in FIG. FIG. 3 is a flowchart (No. 1) for explaining operations in FIGS. 1 and 2; FIG. FIG. 3 is a flowchart (No. 2) for explaining operations in FIGS. 1 and 2; FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Mobile terminal 2a, 2b Microphone 3a, 3b AD converter 4 Noise and target sound component calculation part 5 Adaptive filter (adaptive noise suppression filter)
6 Speakers 41a, 41b, 43 Random access memory (RAM)
42 Digital Signal Processor (DSP)

Claims (11)

A first step of collecting a ringtone at a telephone terminal with a plurality of microphones provided on the telephone terminal;
A second step of estimating, as a noise signal, a voice signal from which a ring tone component has been removed from a plurality of voice signals obtained by collecting the ringtone with the plurality of microphones in the first step. A featured speech estimation method.   The noise signal estimated in the second step is supplied to an adaptive noise suppression filter, and a transmitted voice signal obtained by collecting with the plurality of microphones during a subsequent call is supplied to the adaptive noise suppression filter. The speech estimation method according to claim 1, further comprising a third step of performing filtering processing of the adaptive noise suppression filter. A first step of collecting a ringtone at a telephone terminal with a plurality of microphones provided on the telephone terminal;
A second step of estimating, as a noise signal, a voice signal obtained by removing a ringtone component from a plurality of voice signals obtained by collecting the ringtone with the plurality of microphones in the first step;
A time difference correction is performed on a plurality of sound signals obtained by collecting transmission voices after the start of a call with the plurality of microphones based on sound signal time difference information from the front direction of each of the plurality of microphones. 3 steps,
A speech estimation method comprising: a fourth step of estimating a signal obtained by adding the plurality of speech signals corrected in time difference in the third step as a target sound signal including a noise component.   In the second step, the time correlation between the ring tone data that causes the speaker to ring the ring tone and the plurality of audio signals obtained by collecting the ring tone with the plurality of microphones in the first step is calculated. Then, by subtracting the gain of the ringtone data from the gain of the plurality of voice signals, the gain of the ringtone signal in the plurality of voice signals in phase with the ringtone data is subtracted from the gain of the plurality of voice signals. The speech estimation method according to claim 1 or 3, wherein the speech signal from which the sound component is removed is calculated and estimated as a noise signal.   The noise signal estimated in the second step and the target sound signal including the noise component estimated in the fourth step are supplied to an adaptive noise suppression filter, and the noise component is included by filtering processing. The speech estimation method according to claim 3, further comprising a fifth step of outputting a target sound signal in which a noise component is suppressed from the sound signal. A mobile terminal comprising a plurality of microphones and an adaptive noise suppression filter,
A ringing tone ringing means for ringing a ringtone on a speaker based on ringtone data;
Based on the plurality of sound signals obtained by collecting the ringtones with the plurality of microphones and the ringtone data, the ringtone components are removed from the plurality of sound signals obtained by collecting the ringtones. A mobile terminal comprising: a noise signal calculating means that calculates a noise signal, calculates a noise signal, and outputs the estimated noise signal to the adaptive noise suppression filter.   A transmission voice signal obtained by collecting sounds by the plurality of microphones during a call is supplied to the adaptive noise suppression filter, and filtering processing of the adaptive noise suppression filter based on the noise signal and the transmission voice signal 7. The portable terminal according to claim 6, further comprising transmission voice signal supply means that suppresses a noise component contained in the transmission voice signal by performing A mobile terminal comprising a plurality of microphones and an adaptive noise suppression filter,
A ringing tone ringing means for ringing a ringtone on a speaker based on ringtone data;
Based on the plurality of sound signals obtained by collecting the ringtones with the plurality of microphones and the ringtone data, the ringtone components are removed from the plurality of sound signals obtained by collecting the ringtones. A noise signal calculation means for calculating the noise signal and estimating it as a noise signal and outputting it to the adaptive noise suppression filter;
Correction for performing time difference correction based on sound signal time difference information from the front direction of each of the plurality of microphones for a plurality of sound signals obtained by collecting transmission voices after the start of a call with the plurality of microphones Means,
And a target sound signal calculation unit that estimates a signal obtained by adding the plurality of audio signals corrected for time difference as a target sound signal including a noise component and outputs the target sound signal to the adaptive noise suppression filter. . The noise signal calculation means is
First storage means for storing ringtone data for causing the ringtone to ring on a speaker;
Second storage means for storing a plurality of audio signals obtained by collecting the ringtone with the plurality of microphones;
Taking a time correlation between the ring tone data and a plurality of voice signals obtained by collecting the ring tone, the gain of the ring tone signal in the plurality of voice signals in the same phase as the ring tone data is set to the incoming call signal. The calculation means for calculating, as the noise signal, a voice signal from which the ring tone component has been removed by subtracting a sum of gains of sound data from gains of the plurality of voice signals. The mobile terminal according to 6 or 8.   The adaptive noise suppression filter receives as input signals the noise signal estimated by the calculation by the noise signal calculation means and the target sound signal including the noise component estimated by the calculation by the target sound signal calculation means. 9. The portable terminal according to claim 8, wherein a target sound signal in which the noise component is suppressed from the target sound signal including the noise component is output by filtering processing.   The portable terminal according to claim 9, wherein each of the first storage unit and the second storage unit has a ring buffer configuration.

Images