CN104810021A - Pre-processing method and device applied to far-field recognition - Google Patents

Abstract

The invention provides a preprocessing method and a preprocessing device applied to far field identification, wherein the preprocessing method applied to the far field identification comprises the steps of carrying out fixed beam forming processing on a sound signal to be processed to obtain a beam signal after the fixed beam forming processing; performing acoustic echo cancellation and optimal beam selection on the beam signals subjected to the fixed beam forming processing; and obtaining a pre-processed signal applied to far-field recognition according to the beam signal after the acoustic echo cancellation and the optimal beam selection. The method can improve the preprocessing effect, and optionally, can reduce the operation amount when the number of the sound signals is large.

Description

Pre-processing method and device applied to far-field recognition

technical field

The invention relates to the technical field of data processing, in particular to a pre-processing method and device applied to far-field recognition.

Background technique

Far-field recognition technology, that is, long-distance recognition technology, is usually used to solve the speech recognition request of the scene where the speaker is 2 meters away from the speech device. In order to obtain relatively stable and reliable far-field recognition performance, the pre-processing (far-field pickup) technology for far-field recognition scenes is particularly urgent and important.

In the prior art, the process of far-field sound pickup includes in series: Acoustic echo cancellation (AECoustic echocancellation, AEC), sound source localization, adaptive beamforming (Adaptive Beamforming, ABF), single microphone enhancement and post-processing.

However, in the prior art, a sound source localization module is required, and the accuracy of the sound source localization module itself is not ideal, and it is connected in series with the subsequent ABF, which will also affect the performance of the ABF, thereby affecting the pre-processing effect. In addition, AEC is performed first, when When the number of sound signals to be processed is large, the amount of computation is also large.

Contents of the invention

The present invention aims to solve one of the technical problems in the related art at least to a certain extent.

Therefore, an object of the present invention is to propose a pre-processing method applied to far-field recognition, which can improve the effect of pre-processing, and optionally, can reduce the amount of computation when the number of sound signals is large.

Another object of the present invention is to propose a pre-processing device applied to far-field recognition.

In order to achieve the above purpose, the pre-processing method applied to far-field recognition proposed by the embodiment of the first aspect of the present invention includes: performing fixed beamforming processing on the sound signal to be processed to obtain beam signals after fixed beamforming processing; Acoustic echo cancellation and optimal beam selection are performed on the beam signal after fixed beamforming processing; and a pre-processed signal applied to far-field recognition is obtained according to the beam signal after acoustic echo cancellation and optimal beam selection.

The pre-processing method applied to far-field recognition proposed by the embodiment of the first aspect of the present invention does not require a sound source localization module, which can avoid the problem of poor pre-processing effect caused by inaccurate sound source localization, thereby improving the pre-processing effect. And, optionally, AEC is performed after FBF is performed first, since the number of beams after FBF is usually smaller than the number of sound signals to be processed, the amount of computation can be reduced.

In order to achieve the above purpose, the pre-processing device applied to far-field recognition proposed by the embodiment of the second aspect of the present invention includes: a fixed beamforming module, which is used to perform fixed beamforming processing on the sound signal to be processed to obtain a fixed beamforming processing The beam signal after processing; the processing module is used to perform acoustic echo cancellation and optimal beam selection on the beam signal processed by the fixed beamforming; the acquisition module is used to perform acoustic echo cancellation and optimal beam selection according to the The beam signal is used to obtain the pre-processed signal applied to the far-field identification.

The pre-processing device applied to far-field recognition proposed by the embodiment of the second aspect of the present invention does not require a sound source localization module, which can avoid the problem of poor pre-processing effect caused by inaccurate sound source localization, thereby improving the pre-processing effect. And, optionally, AEC is performed after FBF is performed first, since the number of beams after FBF is usually smaller than the number of sound signals to be processed, the amount of computation can be reduced.

Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.

Description of drawings

The above and/or additional aspects and advantages of the present invention will become apparent and easy to understand from the following description of the embodiments in conjunction with the accompanying drawings, wherein:

FIG. 1 is a schematic flowchart of a preprocessing method applied to far-field recognition proposed by an embodiment of the present invention;

2 is a schematic flowchart of a preprocessing method applied to far-field recognition proposed by another embodiment of the present invention;

Fig. 3 is a schematic flowchart of a pre-processing method applied to far-field recognition proposed by another embodiment of the present invention;

Fig. 4 is a schematic structural diagram of a pre-processing device applied to far-field recognition proposed by another embodiment of the present invention;

Fig. 5 is a schematic structural diagram of a pre-processing device applied to far-field recognition proposed by another embodiment of the present invention;

Fig. 6 is a schematic structural diagram of a pre-processing device applied to far-field recognition proposed by another embodiment of the present invention.

Detailed ways

Embodiments of the present invention are described in detail below, examples of which are shown in the drawings, wherein the same or similar reference numerals denote the same or similar modules or modules having the same or similar functions throughout. The embodiments described below by referring to the figures are exemplary only for explaining the present invention and should not be construed as limiting the present invention. On the contrary, the embodiments of the present invention include all changes, modifications and equivalents coming within the spirit and scope of the appended claims.

Fig. 1 is a schematic flow chart of a pre-processing method applied to far-field recognition proposed by an embodiment of the present invention, the method comprising:

S11: Perform fixed beamforming processing on the sound signal to be processed to obtain a beam signal after fixed beamforming processing.

Wherein, the sound signal to be processed may refer to a microphone signal, and the microphone signal refers to a signal picked up by the microphone, including a near-end voice signal (voice control command), room reverberation, and various environmental noises.

In far-field recognition, in order to improve recognition performance, a microphone array (directional microphone or omnidirectional microphone) is usually used. Therefore, the sound signal to be processed may specifically refer to a microphone array signal, and the microphone array signal includes multiple microphone signals.

The beamforming technology may include ABF adopted in the prior art, and also includes fixed beamforming (Fixed Beamforming, FBF).

The spatial beam characteristics of ABF are adaptive changes, while the spatial beam characteristics of FBF are fixed. Spatial beam characteristics such as signal gain response in a particular direction.

During FBF processing, optionally, the fixed beam forming process uses multiple fixed beams, each fixed beam covers a part of the space, and all the fixed beams form coverage of the entire space.

Through the full coverage of the space by the beam, it can be ensured that the user's speech can be detected when the user is located in any position in the space, and the restriction on the user's position can be avoided.

When the number of sound signals to be processed (such as microphone array signals) is large, in order to reduce the amount of computation, the number of fixed beams used by FBF may be smaller than the number of sound signals to be processed.

For example, the number of fixed beams is 3, and different fixed beams cover different 120-degree spaces; or, the number of fixed beams is 6, and different fixed beams cover different 60-degree spaces. space.

S12: Perform acoustic echo cancellation and optimal beam selection on the beam signal processed by the fixed beamforming.

Wherein, in order to eliminate interference signals, the voice recognition interactive system usually includes an acoustic echo cancellation (Acoustic echocancellation, AEC) module, and the AEC module is usually called a BargeIn function module.

The interference signal is, for example, music generated by a voice recognition interactive system (hereinafter referred to as the system), a text to speech (TTS) signal, and the like.

Since the AEC module not only has to track and learn the acoustic transfer function (Acoustictransfer function, ATF) from the speaker to the microphone of the system, but also learns the time-varying components produced by various processing modules before it, if these changes are faster than those in AEC Due to the convergence speed of the adaptive filter, there will be a problem that the AEC module has been unable to ideally learn these rapid changes, which leads to the inability to eliminate the interference signals played by the system.

Since the spatial beam characteristics of the ABF are changing, and the change speed of the filter of the ABF is usually much faster than that of the filter of the AEC module, in the prior art, the ABF cannot be placed before the AEC to improve the signal-to-noise ratio . The processing effect of AEC depends on the signal-to-noise ratio, and the higher the signal-to-noise ratio, the better the processing effect. Since the ABF cannot be placed before the AEC to improve the signal-to-noise ratio, the existing technology cannot place the ABF before the AEC for processing, which will affect the AEC effect and further affect the far-field recognition effect.

However, in this embodiment, FBF is used. Since the spatial beam characteristics of FBF are fixed, it is known to the AEC module and does not require the AEC module to perform tracking and learning. Therefore, in this embodiment, the FBF can be placed Before AEC. Since the signal-to-noise ratio will be improved after FBF processing, placing FBF before AEC will improve the processing effect of AEC, thereby improving the far-field recognition effect.

On the other hand, when the number of signals included in the microphone array signal is large (such as greater than 6), in the prior art, AEC is performed first, and the number of AEC modules required is the same as the number of microphone signals, which is relatively big. However, in this embodiment, FBF is performed first and then AEC is performed. The number of AEC modules required is the same as the number of FBF beams, and the number of beams of FBF is usually smaller than the number of microphone signals with a large number. For example, the number of beams of FBF is If there are 3 or 6, then the number of required AEC modules can be significantly reduced and the amount of computation can be reduced.

When selecting the optimal beam, it can be selected according to a preset selection criterion. For example, if the preset selection criterion is the maximum signal-to-noise ratio criterion, the beam with the largest signal-to-noise ratio is selected as the optimal beam.

In specific processing, the AEC may be performed first and then the optimal beam selection may be performed, or the optimal beam selection may be performed first and then the AEC is performed.

S13: According to the beam signal after acoustic echo cancellation and optimal beam selection, obtain a pre-processed signal for far-field identification.

After acoustic echo cancellation and optimal beam selection, some post-processing can be performed to further improve the processing effect.

After the pre-processed signal applied to far-field recognition is obtained, the pre-processed signal can be input into a recognizer (far-field recognition engine) for recognition processing.

In this embodiment, there is no need for sound source localization processing, so the overall system performance instability and abnormality caused by sound source localization errors can be effectively avoided; by selecting the optimal beam signal among fixed spatial beam signals, it can effectively break through the traditional method For the constraints and restrictions on the position of the near-end speaker, it can seamlessly adapt to the application scenario where the speaker moves continuously in the room, and significantly improve the overall user experience; using fixed beamforming technology, its spatial beam characteristics do not change with time. This feature can be well learned by the subsequent AEC module, so that the FBF module can be mentioned before the AEC module for processing. In this way, on the one hand, a reference signal with a higher signal-to-noise ratio can be obtained, which can effectively improve the convergence speed and performance of subsequent AEC. On the other hand, since the number of spatial beams using FBF is usually smaller than the number of microphones, it can effectively reduce the number of times the AEC module is used. and reduce the overall computational load.

Fig. 2 is a schematic flowchart of a pre-processing method applied to far-field recognition proposed by another embodiment of the present invention, the method includes:

S21: Perform fixed beamforming processing on the microphone array signal.

In order to improve the AEC performance and reduce the amount of calculation, the entire space can be divided into several spatial beam areas (for example, 3 or 6) by using a microphone array (directional microphone or omnidirectional microphone).

Due to the use of Fixed Beamforming (FBF) technology, the beam characteristics do not change with time, so this characteristic can be well learned by subsequent AEC modules. Therefore, the FBF module can be mentioned before the AEC module for processing. In this way, on the one hand, FBF processing can be used to obtain a reference signal with a higher SNR, thereby effectively improving the convergence speed and performance of subsequent AEC; on the other hand, the number of spatial beams formed by the FBF module is usually smaller than the number of microphones, so it can effectively reduce The number of times the AEC module is used and reduces the overall calculation load.

S22: Using the same number of acoustic echo cancellation modules as the number of beam signals processed by the fixed beamforming process, performing acoustic echo cancellation on each beam signal processed by the fixed beamforming process, to obtain multiple acoustic echo cancellations beam signal.

The FBF module will output beam signals in several directions. These signals are passed to the AEC module to eliminate the interference signals contained in it, such as music played by the system, TTS, etc. The signal after echo removal can significantly improve the far-field recognition performance.

S23: Perform optimal beam selection among the plurality of beam signals after the acoustic echo cancellation, and select the optimal beam signal.

Each space beam signal processed by the above two modules has eliminated various environmental interference to the greatest extent, including background noise, room reverberation, music played by the system, TTS, etc. In this step, this embodiment will select the optimal spatial beam signal from several spatial beam signals according to a certain criterion (such as the maximum signal-to-noise ratio criterion, etc.), as the output signal of this step. In this way, the sound source localization module in the traditional technical solution is removed, which not only effectively reduces the amount of calculation, but also avoids the error transfer effect, that is, the instability and abnormality of the overall system performance caused by the sound source localization error. At the same time, the restriction on the relatively fixed position of the near-end speaker in the traditional technical solution is removed, thereby further improving the user experience; this embodiment replaces the sound source localization module by automatically selecting the optimal signal among several fixed beam signals, thus realizing wireless It can seamlessly adapt to the application scenario where the speaker moves continuously in the room.

S24: Perform single-mic channel enhancement and post-processing on the beam signal after acoustic echo cancellation and optimal beam selection, and determine the single-mic enhanced and post-processed signal as a pre-processed signal applied to far-field recognition.

Similar to traditional technical solutions, various single-microphone noise cancellation technologies will be used to further eliminate residual noise and special post-processing technologies, such as gain amplification and dynamic range control (DRC), will be connected in series at the back end, so that Better improved far-field recognition performance.

In this embodiment, on the basis of the previous embodiment, AEC can be performed first, and then optimal direction beam selection can be performed. At this time, there is no need to limit that there is no system interference signal, and the applicable scenarios are wider.

Fig. 3 is a schematic flow chart of a pre-processing method applied to far-field identification proposed by another embodiment of the present invention. This method can be applied when there is no system interference signal. The method includes:

S31: Perform fixed beamforming processing on the microphone array signal.

For the content of the fixed beamforming processing, reference may be made to the relevant descriptions in the foregoing embodiments, and details are not repeated here.

S32: Select an optimal beam from beam signals processed by multiple fixed beams, and select an optimal beam signal.

There is a special module in the AEC module to detect the speech state. There will be roughly three states, only the near-end voice signal, dual-talk state (near-end voice and far-end voice) and only the state of the far-end signal. The far-end voice is the system Play music or TTS signal, etc. When the special module in the AEC module detects that there is only a near-end voice signal, it can be determined that there is no system interference signal, so that the optimal beam can be selected first, for example, using the maximum signal-to-noise ratio criterion for selection. For a specific optimal beam selection manner, reference may be made to relevant descriptions in the foregoing embodiments, and details are not repeated here.

S33: Use an acoustic echo cancellation module to perform acoustic echo cancellation on the optimal beam signal.

After the optimal beam is selected, only one signal is output, so only one AEC module can be used for AEC, thereby reducing the amount of computation.

S34: Perform single-microphone enhancement and post-processing on the beam signal after acoustic echo cancellation and optimal beam selection, and determine the single-microphone enhanced and post-processed signal as a pre-processed signal for far-field identification.

Similar to traditional technical solutions, various single-microphone noise cancellation technologies will be used to further eliminate residual noise and special post-processing technologies, such as gain amplification and dynamic range control (DRC), will be connected in series at the back end, so that Better improved far-field recognition performance.

In this embodiment, when there is no system interference signal, the optimal direction beam can be selected first and then the AEC can be performed, so that the number of AEC modules can be reduced and the amount of computation can be reduced.

Fig. 4 is a schematic structural diagram of a pre-processing device applied to far-field recognition proposed by another embodiment of the present invention, the device 40 includes:

The fixed beamforming module 41 is configured to perform fixed beamforming processing on the sound signal to be processed, to obtain beam signals after fixed beamforming processing;

Wherein, the sound signal to be processed may refer to a microphone signal, and the microphone signal refers to a signal picked up by the microphone, including a near-end voice signal (voice control command), room reverberation, and various environmental noises.

In far-field recognition, in order to improve recognition performance, a microphone array (directional microphone or omnidirectional microphone) is usually used. Therefore, the sound signal to be processed may specifically refer to a microphone array signal, and the microphone array signal includes multiple microphone signals.

The beamforming technology may include ABF adopted in the prior art, and also includes fixed beamforming (Fixed Beamforming, FBF).

The spatial beam characteristics of ABF are adaptive changes, while the spatial beam characteristics of FBF are fixed. Spatial beam characteristics such as signal gain response in a particular direction.

During FBF processing, optionally, the fixed beam forming process uses multiple fixed beams, each fixed beam covers a part of the space, and all the fixed beams form coverage of the entire space.

Through the full coverage of the space by the beam, it can be ensured that the user's speech can be detected when the user is located in any position in the space, and the restriction on the user's position can be avoided.

When the number of sound signals to be processed (such as microphone array signals) is large, in order to reduce the amount of computation, the number of fixed beams used by FBF may be smaller than the number of sound signals to be processed.

For example, the number of fixed beams is 3, and different fixed beams cover different 120-degree spaces; or, the number of fixed beams is 6, and different fixed beams cover different 60-degree spaces. space.

A processing module 42, configured to perform acoustic echo cancellation and optimal beam selection on the beam signal processed by the fixed beamforming;

Wherein, in order to eliminate interference signals, the voice recognition interactive system usually includes an acoustic echo cancellation (Acoustic echocancellation, AEC) module, and the AEC module is usually called a BargeIn function module.

The interference signal is, for example, music generated by a voice recognition interactive system (hereinafter referred to as the system), a text to speech (TTS) signal, and the like.

Since the AEC module not only has to track and learn the acoustic transfer function (Acoustictransfer function, ATF) from the speaker to the microphone of the system, but also learns the time-varying components produced by various processing modules before it, if these changes are faster than those in AEC Due to the convergence speed of the adaptive filter, there will be a problem that the AEC module has been unable to ideally learn these rapid changes, which leads to the inability to eliminate the interference signals played by the system.

Since the spatial beam characteristics of the ABF are changing, and the change speed of the filter of the ABF is usually much faster than that of the filter of the AEC module, in the prior art, the ABF cannot be placed before the AEC to improve the signal-to-noise ratio . The processing effect of AEC depends on the signal-to-noise ratio, and the higher the signal-to-noise ratio, the better the processing effect. Since the ABF cannot be placed before the AEC to improve the signal-to-noise ratio, the existing technology cannot place the ABF before the AEC for processing, which will affect the AEC effect and further affect the far-field recognition effect.

However, in this embodiment, FBF is used. Since the spatial beam characteristics of FBF are fixed, it is known to the AEC module and does not require the AEC module to perform tracking and learning. Therefore, in this embodiment, the FBF can be placed Before AEC. Since the signal-to-noise ratio will be improved after FBF processing, placing FBF before AEC will improve the processing effect of AEC, thereby improving the far-field recognition effect.

On the other hand, when the number of signals included in the microphone array signal is large (such as greater than 6), in the prior art, AEC is performed first, and the number of AEC modules required is the same as the number of microphone signals, which is relatively big. However, in this embodiment, FBF is performed first and then AEC is performed. The number of AEC modules required is the same as the number of FBF beams, and the number of beams of FBF is usually smaller than the number of microphone signals with a large number. For example, the number of beams of FBF is If there are 3 or 6, then the number of required AEC modules can be significantly reduced and the amount of computation can be reduced.

When selecting the optimal beam, it can be selected according to a preset selection criterion. For example, if the preset selection criterion is the maximum signal-to-noise ratio criterion, the beam with the largest signal-to-noise ratio is selected as the optimal beam.

In specific processing, the AEC may be performed first and then the optimal beam selection may be performed, or the optimal beam selection may be performed first and then the AEC is performed.

For example, referring to FIG. 5, when there are multiple beam signals processed by the fixed beamforming, the processing module 42 includes:

The acoustic echo cancellation module 51 has the same number of beam signals processed by the fixed beamforming, and is connected to the fixed beamforming module, for performing acoustic echo cancellation on each beam signal processed by the fixed beamforming, obtaining a plurality of beam signals after acoustic echo cancellation;

The FBF module will output beam signals in several directions. These signals are passed to the AEC module to eliminate the interference signals contained in it, such as music played by the system, TTS, etc. The signal after echo removal can significantly improve the far-field recognition performance.

The optimal beam selection module 52 is connected with the acoustic echo cancellation module, and is used to select the optimal beam among the plurality of beam signals after the acoustic echo cancellation, and select the optimal beam signal.

Each space beam signal processed by the above two modules has eliminated various environmental interference to the greatest extent, including background noise, room reverberation, music played by the system, TTS, etc. In this step, this embodiment will select the optimal spatial beam signal from several spatial beam signals according to a certain criterion (such as the maximum signal-to-noise ratio criterion, etc.), as the output signal of this step. In this way, the sound source localization module in the traditional technical solution is removed, which not only effectively reduces the amount of calculation, but also avoids the error transfer effect, that is, the instability and abnormality of the overall system performance caused by the sound source localization error. At the same time, the restriction on the relatively fixed position of the near-end speaker in the traditional technical solution is removed, thereby further improving the user experience; this embodiment replaces the sound source localization module by automatically selecting the optimal signal among several fixed beam signals, thus realizing wireless It can seamlessly adapt to the application scenario where the speaker moves continuously in the room.

For another example, referring to FIG. 6, when there are multiple beam signals processed by the fixed beamforming, and when there is no system interference signal, the processing module 42 includes:

The optimal beam selection module 61 is connected to the fixed beam forming module, and is used to select an optimal beam signal from the beam signals processed by multiple fixed beams, and select an optimal beam signal;

There is a special module in the AEC module to detect the speech state. There will be roughly three states, only the near-end voice signal, dual-talk state (near-end voice and far-end voice) and only the state of the far-end signal. The far-end voice is the system Play music or TTS signal, etc. When it is known through the detection of the special module in the AEC module that there is only the near-end voice signal, it can be determined that there is no system interference signal, so that the optimal beam can be selected first, for example, using the maximum signal-to-noise ratio criterion for selection. For a specific optimal beam selection manner, reference may be made to relevant descriptions in the foregoing embodiments, and details are not repeated here.

An acoustic echo cancellation module 62, connected to the optimal beam selection module, for performing acoustic echo cancellation on the optimal beam signal.

After the optimal beam is selected, only one signal is output, so only one AEC module can be used for AEC, thereby reducing the amount of computation.

The obtaining module 43 is configured to obtain a pre-processed signal applied to far-field identification according to the beam signal after acoustic echo cancellation and optimal beam selection.

After acoustic echo cancellation and optimal beam selection, some post-processing can be performed to further improve the processing effect.

After the pre-processed signal applied to far-field recognition is obtained, the pre-processed signal can be input into a recognizer (far-field recognition engine) for recognition processing.

Optionally, the acquiring module 43 is specifically used for:

Perform single-microphone enhancement and post-processing on the beam signal after acoustic echo cancellation and optimal beam selection, and determine the signal after single-microphone enhancement and post-processing as the pre-processed signal applied to far-field recognition.

Similar to traditional technical solutions, various single-microphone noise cancellation technologies will be used to further eliminate residual noise and special post-processing technologies, such as gain amplification and dynamic range control (DRC), will be connected in series at the back end, so that Better improved far-field recognition performance.

In this embodiment, there is no need for sound source localization processing, so the overall system performance instability and abnormality caused by sound source localization errors can be effectively avoided; by selecting the optimal beam signal among fixed spatial beam signals, it can effectively break through the traditional method For the constraints and restrictions on the position of the near-end speaker, it can seamlessly adapt to the application scenario where the speaker moves continuously in the room, and significantly improve the overall user experience; using fixed beamforming technology, its spatial beam characteristics do not change with time. This feature can be well learned by the subsequent AEC module, so that the FBF module can be mentioned before the AEC module for processing. In this way, on the one hand, a reference signal with a higher signal-to-noise ratio can be obtained, which can effectively improve the convergence speed and performance of subsequent AEC. On the other hand, since the number of spatial beams using FBF is usually smaller than the number of microphones, it can effectively reduce the number of times the AEC module is used. and reduce the overall computational load.

It should be noted that, in the description of the present invention, the terms "first", "second" and so on are only used for description purposes, and should not be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "plurality" means at least two.

Any process or method descriptions in flowcharts or otherwise described herein may be understood to represent modules, segments or portions of code comprising one or more executable instructions for implementing specific logical functions or steps of the process , and the scope of preferred embodiments of the invention includes alternative implementations in which functions may be performed out of the order shown or discussed, including substantially concurrently or in reverse order depending on the functions involved, which shall It is understood by those skilled in the art to which the embodiments of the present invention pertain.

It should be understood that various parts of the present invention can be realized by hardware, software, firmware or their combination. In the embodiments described above, various steps or methods may be implemented by software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented by any one or combination of the following techniques known in the art: Discrete logic circuits, ASICs with suitable combinational logic gates, Programmable Gate Arrays (PGAs), Field Programmable Gate Arrays (FPGAs), etc.

Those of ordinary skill in the art can understand that all or part of the steps carried by the methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. During execution, one or a combination of the steps of the method embodiments is included.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing module, each unit may exist separately physically, or two or more units may be integrated into one module. The above-mentioned integrated modules can be implemented in the form of hardware or in the form of software function modules. If the integrated modules are realized in the form of software function modules and sold or used as independent products, they can also be stored in a computer-readable storage medium.

The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, and the like.

In the description of this specification, descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific examples", or "some examples" mean that specific features described in connection with the embodiment or example , structure, material or characteristic is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention, those skilled in the art can make the above-mentioned The embodiments are subject to changes, modifications, substitutions and variations.

Claims (11)

1. be applied to the pre-treating method that far field identifies, it is characterized in that, comprising:

Wave beam forming process is fixed to voice signal to be processed, is fixed the beam signal after Wave beam forming process;

To the beam signal after described fixed beam formation processing, carry out sound Echo cancellation and optimal beam selection;

Beam signal after selecting according to sound Echo cancellation and optimal beam, obtains the signal be applied to after the pre-treatment of far field identification.

2. method according to claim 1, is characterized in that, the number of the fixed beam that described fixed beam formation processing adopts is multiple, and space, each fixed beam cover part, all fixed beams form the covering to whole space.

3. method according to claim 2, is characterized in that, the number of described fixed beam is 3, and different fixed beams covers the different spaces of 120 degree respectively; Or the number of described fixed beam is 6, different fixed beams covers the different spaces of 60 degree respectively.

4. method according to claim 1, is characterized in that, the number of the fixed beam that described fixed beam formation processing adopts is multiple, and the quantity of described fixed beam is less than the quantity of voice signal to be processed.

5. the method according to any one of claim 1-4, it is characterized in that, when the beam signal after described fixed beam formation processing is multiple, described to the beam signal after described fixed beam formation processing, carry out sound Echo cancellation and optimal beam selection, comprising:

Adopt the sound Echo cancellation module identical with the beam signal number after described fixed beam formation processing, sound Echo cancellation is carried out to the beam signal after each fixed beam formation processing, obtains the beam signal after multiple sound Echo cancellation;

In multiple beam signals after sound Echo cancellation, carry out optimal beam selection, select optimal beam signal.

6. the method according to any one of claim 1-4, it is characterized in that, when the beam signal after described fixed beam formation processing is multiple, and, when there is not system interference signal, described to the signal beam signal after described fixed beam formation processing, carry out sound Echo cancellation and optimal beam selection, comprising:

From the beam signal after multiple fixed beam formation processing, carry out optimal beam selection, select an optimal beam signal;

Adopt a sound Echo cancellation module, sound Echo cancellation is carried out to described optimal beam signal.

7. the method according to any one of claim 1-4, is characterized in that, described according to the beam signal after sound Echo cancellation and optimal beam selection, obtains the signal be applied to after the pre-treatment of far field identification, comprising:

Beam signal after selecting sound Echo cancellation and optimal beam carries out single wheat enhancing and aftertreatment, and is strengthened by single wheat and signal after aftertreatment is defined as the signal that is applied to after the pre-treatment of far field identification.

8. be applied to the pretreating device that far field identifies, it is characterized in that, comprising:

Fixed beam forms module, for being fixed Wave beam forming process to voice signal to be processed, is fixed the beam signal after Wave beam forming process;

Processing module, for the beam signal after described fixed beam formation processing, carries out sound Echo cancellation and optimal beam selection;

Acquisition module, for according to the beam signal after sound Echo cancellation and optimal beam selection, obtains the signal be applied to after the pre-treatment of far field identification.

9. device according to claim 8, is characterized in that, when the beam signal after described fixed beam formation processing is multiple, described processing module comprises:

Sound Echo cancellation module, identical with the beam signal number after described fixed beam formation processing, forming model calling with described fixed beam, for carrying out sound Echo cancellation to the beam signal after each fixed beam formation processing, obtaining the beam signal after multiple sound Echo cancellation;

Optimal beam selects module, with described sound Echo cancellation model calling, in the multiple beam signals after sound Echo cancellation, carries out optimal beam selection, selects optimal beam signal.

10. device according to claim 8, is characterized in that, when the beam signal after described fixed beam formation processing is multiple, and when there is not system interference signal, described processing module comprises:

Optimal beam selects module, forms model calling, for from the beam signal after multiple fixed beam formation processing, carries out optimal beam selection, select an optimal beam signal with described fixed beam;

A sound Echo cancellation module, selects model calling with described optimal beam, for carrying out sound Echo cancellation to described optimal beam signal.

11. devices according to Claim 8 described in-10 any one, is characterized in that, described acquisition module specifically for:

Beam signal after selecting sound Echo cancellation and optimal beam carries out single wheat enhancing and aftertreatment, and is strengthened by single wheat and signal after aftertreatment is defined as the signal that is applied to after the pre-treatment of far field identification.