JP4476355B2 - Echo and noise cancellation - Google Patents

Description

[Priority claim]
This application claims the benefit of Patent Document 1, which is commonly assigned to this application and is the same as the present application, the entire disclosure of which is incorporated herein by reference. This application claims the benefit of Patent Document 2 that is commonly assigned to this application and is the same as the present application, the entire disclosure of which is incorporated herein. This application also claims the benefit of US Pat. No. 6,057,056, whose assignee is common to this application and is co-pending with this application, the entire disclosure of which is incorporated herein by reference. This application also claims the benefit of US Pat. No. 6,057,056, whose assignee is common to this application and is co-pending with this application, the entire disclosure of which is incorporated herein by reference. This application also claims the benefit of Patent Document 5, which is common to the present application and is assigned at the same time as the present application, the entire disclosure of which is incorporated herein. This application also claims the benefit of US Pat. No. 6,057,097, whose assignee is common to this application and co-pending with this application, the entire disclosure of which is incorporated herein by reference. This application also claims the benefit of Patent Document 7, which is commonly assigned to this application and is pending at the same time as this application, the entire disclosure of which is incorporated herein by reference. This application also claims the benefit of U.S. Pat. No. 6,053,075, whose assignee is common to this application and is co-pending with this application, the entire disclosure of which is incorporated herein by reference. This application also claims the benefit of U.S. Pat. No. 6,053,075, whose assignee is common to this application and is co-pending with this application, the entire disclosure of which is incorporated herein by reference. This application also claims the benefit of US Pat. No. 6,057,056, whose assignee is common to this application and is pending at the same time as this application, the entire disclosure of which is incorporated herein by reference.
US Patent Application No. 11 / 381,728, Shadon Mao, "ECHO AND NOISE CANCELATION", filed May 4, 2006, (Attorney Docket Number SCEA05064US00) US Patent Application No. 11 / 381,729, Shadon Mao, "ULTRA SMALL MICROPHONE ARRAY", filed May 4, 2006, (Attorney Docket Number SCEA05062US00) US Patent Application No. 11 / 381,725, Shadon Mao, "METHODS AND APPARATUS FOR TARGETED SOUND DETECTION", filed May 4, 2006, (Attorney Docket Number SCEA05072US00), US Patent Application No. 11 / 381,727, Shadon Mao, "NOISE REMOVAL FOR ELECTRONIC DEVICE WITH FAR FIELD MICROPHONE ON CONSOLE", filed May 4, 2006, (Attorney Docket Number SCEA05073US00) US Patent Application No. 11 / 381,724, Shadon Mao, "METHODS AND APPARATUS FOR TARGETED SOUND DETECTION AND CHARACTERIZATION", filed May 4, 2006, (Attorney Docket Number SCEA05079US00) US Patent Application No. 11 / 381,721, Shadon Mao, "SELECTIVE SOUND SOURCE LISTENING IN CONJUNCTION WITH COMPUTER INTERACTIVE PROCESSING", filed May 4, 2006, (Attorney Docket Number SCEA04005 JUMBOUS) PCT Application PCT / US06 / 17483, Shadon Mao, "SELECTIVE SOUND SOURCE LISTENING IN CONJUNCTION WITH COMPUTER INTERACTIVE PROCESSING", filed May 4, 2006, (Attorney Docket Number SCEA04005 JUMBOPCT) US Patent Application No. 11 / 418,988, Shadon Mao, "METHODS AND APPARATUSES FOR ADJUSTING A LISTENING AREA FOR CAPTURING SOUNDS", filed May 4, 2006, (Attorney Docket Number SCEA-00300) US Patent Application No. 11 / 418,989, Shadon Mao, "METHODS AND APPARATUSES FOR CAPTURING AN AUDIO SIGNAL BASED ON VISUAL IMAGE", filed May 4, 2006, (Attorney Docket Number SCEA-00400) US Patent Application No. 11 / 429,047, Shadon Mao, "METHODS AND APPARATUSES FOR CAPTURING AN AUDIO SIGNAL BASED ON A LOCATION OF THE SIGNAL", filed May 4, 2006, (Attorney Docket Number SCEA-00500)

[Technical Field of the Invention]
The present invention relates to acoustic signal processing, and more particularly to echo and canceling in acoustic signal processing.

Many portable electronic devices, such as interactive video game controllers, can handle bidirectional acoustic signals. Such devices typically include a microphone that receives a local speech signal s (t) from a user of the device and a speaker that emits a speaker signal x (t) that the user can hear. In order to further reduce the size of the video game controller, it is desirable to install the microphone and the speaker relatively close to each other (for example, within 20 cm). In contrast, the user may be located further away from the microphone (eg, 3 to 5 meters). The microphone generates a signal d (t) that includes both the local speech signal s (t) and the speaker echo signal x ₁ (t). In addition to this, the microphone may experience background noise n (t). Therefore, the entire microphone signal is d (t) = s (t) + x ₁ (t) + n (t). Because it is relatively near the speaker, the microphone signal d (t) may be occupied by the speaker echo signal x ₁ (t).

  Speaker echo is a common phenomenon in telecommunications applications, and echo suppression and echo cancellation are relatively mature techniques. The echo suppressor operates when it detects the presence of a voice signal going in one direction on the line and inserts a large loss in the other direction. Normally, when an echo suppressor at the far end of the line detects voice from the near end of the line, the echo suppressor adds this loss. This added loss can prevent the speaker signal x (t) from being retransmitted into the local speech signal d (t).

  While echo suppression is effective, it often leads to several problems. For example, the local speech signal s (t) and the remote speaker signal x (t) are often generated at least at the same time. This situation is also called double talk. The situation where only the remote speaker signal exists is also called remote single talk. Since each echo suppressor detects voice energy from the far-end of the circuit, this usually results in loss being inserted in both directions at the same time, blocking both-side calls. To prevent this, the echo suppressor can be set to detect only voice activity from the near-end speaker. This ensures that no loss is inserted (or only a smaller loss is inserted) when the near-end speaker and the far-end speaker are speaking at the same time. Unfortunately, this temporarily disappears until the original echo suppressor effect.

  In addition, the echo suppressor alternately inserts and removes losses, so there is often a small delay when a new speaker begins speaking, and the sound at the beginning of the speaker's speech is clipped. . In addition, when the far-end party is noisy, when the far-end speaker is speaking, the near-end speaker can hear the background sound, but when the near-end speaker starts speaking, the echo suppressor Suppress. As a result, the background sound suddenly disappears, so that the user at the near end receives the impression that the line is disconnected.

An echo cancellation technique has been developed to address the above problems. Echo cancellation uses an analog or digital filter to remove unwanted noise and echo from the input signal and produces a filtered signal e (t). In echo cancellation, complex algorithm procedures are used to calculate the speech model. The procedure includes inputting a microphone signal d (t) and a portion of a remote signal x (t) into an echo cancellation processor, predicting a speaker echo signal x ₁ (t), and Subtracting from the microphone signal d (t). Echo prediction schemes must be learned by an echo cancellation processor in a process known as adaptation.

The effect of such a technique is measured by an echo suppression ratio (ESR). This is simply the ratio of the true echo energy received by the microphone to the residual echo energy remaining in the filtered signal x ₁ (t) (typically expressed in decibels). According to the standards established by the International Telecommunications Union (ITC), for remote single talk, an attenuation of at least 45 decibels is required for the echo level. During double talk (or during strong background noise), this attenuation level may be as low as 30 dB. However, these recommended criteria were developed in a system where the user generating the local speech signal is closer to the microphone. Thus, the recorded signal-to-noise ratio (ratio of target speech energy to echo noise energy) is often better than 5 decibels. For applications where the user is 3 to 5 meters away, such as a video game controller, and a loudspeaker that is closer than 0.5 meters from the open microphone produces a large echo, these recommended criteria Does not apply. In such an application, the signal-to-noise ratio will be from -15 dB to less than -30 dB. Remote single talk may require an ESR of 60 dB or more, and double talk may require an ESR of 35 dB or more. Existing echo cancellation techniques cannot achieve such high ESR levels.

  Accordingly, there is a need in the art for an echo cancellation system and method that overcomes the aforementioned disadvantages.

[Summary of Invention]
To overcome the aforementioned disadvantages, embodiments of the present invention are aimed at echo cancellation methods and apparatus in a system having a speaker and a microphone. The speaker receives a speaker signal x (t). The microphone receives a microphone signal d (t) that includes a local signal s (t) and an echo signal x ₁ (t). The echo signal x ₁ (t) depends on the speaker signal x (t). The microphone signal d (t) is filtered in parallel by a first adaptive filter and a second adaptive filter having mutually complementary echo cancellation characteristics. The minimum echo output e ₃ (t) is determined as the output e ₁ (t) from the first adaptive filter and the output e ₂ (t) from the second adaptive filter. The energy of the minimum echo output is smaller and the correlation between the minimum echo output and the speaker signal x (t) is smaller. A microphone output is then generated using the minimum echo output e ₃ (t). Optionally, residual echo cancellation and / or noise cancellation may be applied to the minimum echo output.

It is the schematic of the echo cancellation apparatus concerning embodiment of this invention. 1B is a schematic diagram of a voice activity detection adaptive filter that can be used in the echo cancellation apparatus of FIG. 1A. FIG. 1B is a schematic diagram of an adaptive filter with cross-correlation analysis that can be used in the echo cancellation apparatus of FIG. 1A. FIG. It is a flowchart explaining the echo cancellation method concerning embodiment of this invention. It is a flowchart explaining another method for the echo cancellation concerning embodiment of this invention. It is the schematic of the echo cancellation apparatus concerning another embodiment of this invention.

[Description of Specific Embodiment]
The following detailed description includes specific details for purposes of explanation, but it is common in the art that many variations and modifications of the details described below are possible within the scope of the invention. It will be understood by those who have knowledge. Accordingly, the description of the embodiments of the present invention described below does not lose the generality of the invention described in the claims, and the following description is described in the claims. It does not impose any restrictions on the invention.

  According to an embodiment of the invention, a new configuration of an integrated echo and noise canceller with two functionally identical filters is proposed. These filters involve orthogonal control and representations (orthogonal controls and representations). In such a configuration, the two orthogonal filters complement each other so as to increase the robustness of the entire system in noisy hands-free voice communication.

  In particular, the integrated echo noise canceller uses two subsystems that are controlled separately in parallel. Each of these subsystems has a direct control mechanism. The echo noise canceller includes a front echo canceller and a backup echo canceller. The front echo canceller uses double talk detection. To ensure robustness to local speech, the front echo canceller takes a conservative adaptation approach but offers smaller echo suppression and slower adaptation to speech and echo changes . The backup echo canceller uses cross-correlation to measure the similarity between the error signal and the echo signal. The backup echo canceller takes an aggressive strategy so that the filter is updated quickly. Backup echo cancellers are unstable to local speech / noise because they may over-adapt while providing large echo suppression. Integration of the outputs of these two echo cancellers is performed based on a cross-correlation analysis that measures which echo canceller and the difference between the echo signals are large. This integration also checks the filter stability of both echo cancellers. If one filter is over-predicted or under-predicted, it is complemented by the other filter. Such a system is designed to ensure that one filter operates correctly at any time.

  The system may optionally include an echo residual noise predictor that takes a similar approach. The echo residual noise prediction unit uses two independent sub prediction units with direct control in parallel. The first prediction unit is based on an echo-distance-missmatch that relies on a robust double-talk detection unit. The first predictor is relatively accurate but unstable due to double-talk detection errors. The second predictor is based on cross-spectrum-analysis. The prediction of the second predictor is biased but stable, independent of local speech detection and consistent. In integrating these two residual echo predictions, a min / max approach is taken for far-end calls only or double-talk, respectively.

FIG. 1A is a diagram showing an audio system 99 using an echo cancellation apparatus 100 according to an embodiment of the present invention. The operation of the apparatus 100 will be understood by referring to the flowchart of the method 200 shown in FIG. 2A and the method 220 shown in FIG. 2B. Audio system 99 generally includes a speaker 102 and a microphone 104 that receive a remote signal x (t). The local sound source 101 emits a local speech signal s (t). The microphone 104 receives both the local speech signal s (t) and the echo signal x ₁ (t) associated with the speaker signal x (t). The microphone 104 also receives noise n (t) generated from the environment in which the microphone 104 is located. Then, the microphone 104 generates a microphone signal d (t). The microphone signal d (t) will be given by d (t) = s (t) + x ₁ (t) + n (t).

  The echo cancellation apparatus 100 generally includes a first adaptive echo cancellation filter EC (1) and a second adaptive echo cancellation filter EC (2). Each adaptive filter receives a microphone signal d (t) and a speaker signal x (t). 2A-2B, filter EC (1) adaptively filters the microphone signal d (t) as shown in step 202, and filter EC (2) is shown in step 204 as The microphone signal d (t) is adaptively filtered in parallel with the first filter EC (1). As used herein, “operating in parallel” by a filter means receiving substantially the same input d (t). Parallel operations are distinguished from serial operations in which the output of one filter is the input of the other filter. Depending on the state of the two filters EC (1), EC (2), one filter serves as the main “front” filter and the other filter serves as the “backup” filter. One filter takes a cautious approach to echo cancellation, while the other filter takes a more aggressive approach.

The state of the filters EC (1), EC (2) will be understood in connection with the following signal model.
y (t) = x (t) ^* h (n)
d (t) = y ₀ (t) + s (t)
e (t) = d (t) -y (t)
Here, y (t) is an echo synthesized by the echo canceller filter.
x (t) is an echo played in the loudspeaker.
h (n) is an adaptive filter function of the echo canceller filter.
d (t) is a noisy signal received by the microphone.
y ₀ (t) is the true echo that appears at the microphone.
s (t) is local voice.
E (t) is an echo-cancelled residual signal generated by the echo canceller filter.

  The two filters EC (1), EC (2) have complementary echo cancellation characteristics. As used herein, “having complementary echo cancellation” means that in two adaptive filters that receive the same input, when one filter is not well adapted to the input, the other filter is The case where it is adapted. In the context of this application, the filter function h (n) is “adapted well” that the filter function h (n) is stable and converges to a true echo-path filter. This is when it is neither an overestimation nor underestimation.

ここで”Ｅ”は、統計的期待値である。
式２に示す演算子は、相互相関演算を表す。
（式２）

離散的な関数ｆ_ｉとｇ_ｉについて、相互相関は式３で定義される。
（式３）

ここで、和は適切な値の整数ｊについてとられており、アスタリスクは、複数共役を表す。連続関数ｆ（ｘ）とｇ（ｘ）について、相互相関は式４で定義される。
（式４）

ここで積分は適切なｔの値についてとられる。 If h (n) has converged to a true echo path filter (y (t) ˜ = y ₀ (t)), that is, if the predicted echo is approximately equal to the true echo, the coherence function α , The state of the echo canceller filters EC (1), EC (2) will be quantified. α is related to the cross-correlation between y (t) and e (t), and Equation 1 is established.
(Formula 1)

Here, “E” is a statistical expectation value.
The operator shown in Equation 2 represents a cross-correlation operation.
(Formula 2)

For the discrete functions f _i and g _i , the cross-correlation is defined by Equation 3.
(Formula 3)

Here, the sum is taken for an integer j of an appropriate value, and the asterisk represents a plurality of conjugates. For continuous functions f (x) and g (x), the cross-correlation is defined by Equation 4.
(Formula 4)

Here, the integral is taken for the appropriate value of t.

  In the coherence function α, the numerator represents the cross-correlation between e (t) and y (t). The denominator represents the autocorrelation of y (t) and serves as a normalization term.

  Ideally, if h (n) converges, α should be close to “0” (since the residual signal e (t) does not include y (t)). If h (n) does not converge, α should be close to “1” (since e (t) contains a strong echo of y (t)). If h (n) behaves strangely or diverges, α should be negative (because of the divergence of the filter, e (t) contains a strong echo whose phase is shifted by 180 degrees. ).

Thus, for example, the value of the coherence function α may be used to define four possible states for the states of the filters EC (1), EC (2). However, it is not limited to this.
(1) If the filter h (n) is stable and converges and is neither overpredicted nor underpredicted, then 0 <= α <= 0.1
(2) When the filter h (n) is stable but underestimated (not yet converged) α> 0.2
(3) When the filter h (n) is overestimated, α <−0.1
(4) When the filter h (n) diverges, α <−0.25.
One skilled in the art will appreciate that for these different states, different α value ranges can be determined.

  If the filter is in good condition (e.g., state (1)), the setting may be saved for recovery when it subsequently diverges. If the filter is diverging, underestimated, or overpredicted, the front and backup echo cancellers exchange their roles. While the front filter serves as a backup, the backup filter serves as a front filter. Since one filter takes a careful adaptive approach and the other takes a positive adaptive approach, this exchange eventually causes both filters to converge faster and more dynamically stable.

  In addition, if the filter is under-predicted or over-predicted, the adaptation step size will increase or decrease with a small delta value to accelerate or decelerate the adaptation speed for faster convergence or better tracking stability. May be. Usually, a larger step size is required for faster convergence. This sacrifices good tracking of details and lowers the echo suppression ratio ESR. Converging more slowly with a small step size is more stable and has the ability to track even small changes, but is not suitable for tracking echo dislocation fast.

  The combination of dynamic step size and front / backup filter exchange improves the overall balance of the system in terms of fast tracking versus detailed tracking and stability versus convergence. These are the twin issues that are really important in adaptive system design.

  If one filter diverges and the other filter is in good condition, the settings of the other filter may be replicated to reinitialize the divergence filter. Alternatively, the diverged filter may be recovered using a previously saved, good state filter setting.

  For example, the echo canceling adaptive filters EC (1) and EC (2) may be based on a frequency domain normalized least squares adaptive filter. However, it is not limited to this. Each filter may be implemented as hardware, software, or a combination of hardware and software.

1B and 1C show examples of suitable complementary adaptive filters. Specifically, FIG. 1B shows an adaptive echo cancellation filter 120 with voice activity detection. The filter 120 can be used as the first adaptive filter EC (1). Filter 120 includes a variable filter 122 having a finite impulse response (FIR) filter characterized by a filter coefficient w _t . The variable filter 122 receives the microphone signal d (t) and performs filtering according to the value of the filter coefficient w _t to generate a filtered signal d ′ (t). The variable filter 124 predicts the desired signal by convolving the input signal with an impulse response determined by the coefficient w _t1 . Each filter coefficient w _t1 is updated at regular intervals of the amount Δw _t according to the update algorithm 124. As an example, the filter coefficient w _t may be selected such that the filter signal d ′ (t) attempts to predict the speaker echo signal x ₁ (t) as the desired signal. The difference unit 126 subtracts the filtered signal d ′ (t) from the microphone signal d (t) to provide the predicted signal e ₁ (t). The prediction signal e ₁ (t) predicts the local speech signal s (t). The error signal e (t) may be generated by subtracting the filtered signal d ′ (t) from the remote signal x (t). The error signal e (t) is used by the update algorithm 124 to adjust the filter coefficient w _t . The adaptive algorithm 124 generates a correction factor based on the remote signal x (t) and the error signal. Examples of coefficient update algorithms include least squares (LMS) and recursive least squares (RLS). In the LMS update algorithm, for example, the filter coefficient is updated based on the expression w _{t1 + 1} = w _t1 + μe (t) x (t). Here, μ is a step size. Initially, w _t1 = 0 for all w _t1 . Note that in this example the quantity μe (t) x (t) is the quantity Δw _t . As described above, the step size μ may be dynamically adjusted according to the state of the adaptive filter. Specifically, when the filter is underestimated, the step size μ may be increased by a small delta amount in order to accelerate the adaptation speed so as to converge quickly. If the filter is overestimated, the adaptation step size μ may instead be reduced by a small delta amount to reduce the adaptation speed so that tracking is better stabilized.

The time domain representation e (t) x (t) is a multiplication. This calculation may be implemented in the frequency domain as follows. Initially, e (t), x (t), and h (n) may be transformed from the time domain to the frequency domain, for example, by Fast Fourier Transform (FFT).
E (k) = fft (e (t))
X (k) = fft (x (t))
H (k) = fft (h (n))

The LMS update algorithm in the actual frequency domain is as follows.
H (k) = H (k) + (μ ^* conj (X (k)). ^* E (k)) / (Δ + X (k) ^* conj (X (k))
Where μ is the filter adaptation step size and is dynamic.
conj (a) represents a complex conjugate of the complex number a.
^{* Indicates} a complex multiplication.
Δ is a regulator that prevents the denominator from becoming quantitatively unstable.

In the above equation, “conj (X (k)). ^* E (k)” performs the “e (t) x (t)” task. In the denominator, “X (k) ^* conj (X (k))” serves to normalize for the purpose of increasing stability.

  The voice activated detection VAD adjusts the update algorithm 124 so that when the remote signal x (t) is present (eg, greater than or equal to a predetermined threshold), the variable filter 122 causes the microphone signal d (t ) May be adaptively filtered. The adaptive filter using the voice activated detection (sometimes referred to as double talk detection) shown in FIG. 1B is a filter that adapts relatively slowly. However, this filter is also very accurate in that it produces few false positives. A complementary adaptive filter for filter 120 may be, for example, a filter that adapts relatively quickly but often tends to produce false positives.

As an example, FIG. 1C shows an adaptive filter 130 that is complementary to the filter 120 of FIG. 1B. The adaptive filter 130 includes a variable filter characterized by a filter coefficient w _t2 and an update algorithm 134 (eg, the LMS update algorithm described above). Filter 132 attempts to predict speaker echo signal x ₁ (t) as the desired signal. The difference unit 136 subtracts the filtered signal d ′ (t) from the microphone signal d (t) to provide a prediction signal e ₂ (t) that predicts the local speech signal s (t). The filtered signal d ′ (t) may be subtracted from the remote signal x (t) to generate the error signal e (t). The error signal e (t) is used by the update algorithm 134 to adjust the filter coefficient w _t2 . In the filter 130, the cross-correlation analysis CCA adjusts the update algorithm 134 so that the variable filter 132 attempts to reduce the cross-correlation between the predicted signal e ₂ (t) and the speaker echo signal x (t). .

When e ₂ (t) and x (t) are very strongly correlated, the filtering process is said to be underestimated and the update algorithm 134 is adjusted to increase Δw _t2 . When the cross-correlation between e ₂ (t) and x (t) is below the threshold, the filtering process is said to be over-predicted and the update algorithm 134 is adjusted to reduce Δw _t2 .

  An adaptive filter using the type of cross-correlation analysis (also referred to as cross-spectrum analysis) shown in FIG. 1C adapts the filter relatively quickly. However, this filter is also unstable in that it often produces false positives. Therefore, the filter 120 and the filter 130 are examples of complementary filters.

（式６）

ここで、式７の演算子
（式７）

は、例えば、上で定義されたように演算子の両側の量について、その間の相互相関をとる演算を表現する。最小エコー出力ｅ_３（ｔ）は、マイクロフォン１０４のフィルタリング処理された出力として用いられてもよい。 Reference is again made to FIG. 1A. The integrator 106 is connected to the first adaptive filter EC (1) and the second adaptive filter EC (2). The integrator 106 is configured to determine the minimum echo output e ₃ (t) from the respective outputs e ₁ (t), e ₂ (t) of the first and second adaptive filters. The minimum echo output e ₃ (t) is either e ₁ (t) or e ₂ (t), and has the smaller energy and the smaller correlation with the speaker signal x (t). The energy of one of e ₁ (t) and e ₂ (t) is smaller, but when the correlation with x (t) is smaller, the smaller correlation is the smallest echo output. Used as e ₃ (t). For example, when one of the filters is overestimated (ie, the energy output is small because the target speech tends to be canceled), the correlation should be small regardless of the energy. The minimum energy may be determined by determining the minimum value of E {e ₁ (t)} and E {e ₂ (t)}. Here, E {} represents an operation for determining the expected value of the quantity in parentheses. Reference is again made to FIGS. 2A-2B. In step _206, cross-correlation to either _e 1 (t) and _e 2 (t) to determine whether the cross-correlation is small between the loudspeaker signal _x 1 _(t), the _e 1 (t) and _e 2 (t) An analysis may be performed. The cross-correlation analysis may include a step of determining a minimum value of Equation 5 and Equation 6 below.
(Formula 5)

(Formula 6)

Here, the operator of Expression 7 (Expression 7)

Represents, for example, an operation that takes the cross-correlation between the quantities on both sides of the operator as defined above. The minimum echo output e ₃ (t) may be used as the filtered output of the microphone 104.

典型的には、式９が成り立つ。
（式９）

しかしながら、式１０が、ある閾値（例えば約０．２）未満であるときには、ＥＣ（２）が過度にフィルタリング処理することにより、ローカル信号ｓ（ｔ）が除去されている。
（式１０）

このような状況において、インテグレータ１０６は、ｅ_１（ｔ）を最小エコー出力ｅ_３（ｔ）として選択してもよい。適応フィルタリング処理を安定させるために、ステップ２１２において、ＥＣ（２）のフィルタ係数ｗ_ｔ２ が、ＥＣ（１）のフィルタ係数ｗ_ｔ１ として設定されてもよい。そしてステップ２１５において、ＥＣ（２）は、０、または、以前のうまく適応したことが知られている状態に、再初期化されてもよい。例えば、フィルタ係数は、規則的な間隔で（例えば約１０秒から２０秒ごとに）保存されて、ＥＣ（２）が発散したときにこれを再初期化するために用いられてもよい。
In some situations, one of the filters EC (1), EC (2) may over-filter the local signal. In such a situation, the filter is said to be “divergent”. This can happen in particular when EC (2) is a cross-correlation filter of the type shown for example in FIG. 1C. To address this possibility, it is determined in step 208 whether EC (2) diverges. As an example, the integrator 106 may be configured to determine whether the second adaptive echo cancellation filter has removed the local signal s (t) by excessive filtering. This can be done by examining the expected value of the cross-correlation between e ₂ (t) and the speaker echo signal x ₁ (t). That is, it is expressed by Formula 8.
(Formula 8)

Typically, Equation 9 holds.
(Formula 9)

However, when Equation 10 is less than a certain threshold (eg, about 0.2), EC (2) is excessively filtered to remove the local signal s (t).
(Formula 10)

In such a situation, the integrator 106 may select e ₁ (t) as the minimum echo output e ₃ (t). To stabilize the adaptive filtering process, in step 212, the filter coefficients _{w t2} of EC (2) may be set as the filter coefficients _{w t1} of EC (1). Then, in step 215, EC (2) may be re-initialized to 0, or a state known to have been previously well adapted. For example, the filter coefficients may be stored at regular intervals (eg, approximately every 10 to 20 seconds) and used to reinitialize EC (2) when it diverges.

Usually, when a cross-correlation filter does not diverge, it is said to be well adapted. Since EC (2) and EC (1) have complementary filtering characteristics, when EC (2) is well adapted, EC (1) will be underestimated. In order to stabilize the adaptive filtering process, the filter coefficient w _{t1 of} the first adaptive filter EC (1) is exchanged with the filter coefficient w _{t2 of} the second adaptive filter EC (2), as shown in step 214. When the filter is implemented in software, the coefficients w _t1 and w _t2 may be stored at a position specified by the pointer in the memory. The coefficients w _t1 and w _t2 may be exchanged, for example, by switching the pointers to w _t1 and w _t2 .

The minimum echo output e ₃ (t) may include some residual echo xe (t) from the speaker signal x (t). The apparatus 100 is optionally connected to first and second echo residual prediction units ER (1) and ER (2) and echo residual prediction units ER (1) and ER (2) connected to the integrator 106. A connected residual echo cancellation module 108 may be included.

ここで、式１１の値は、ｅ_３（ｔ）が式１２の相互相関の期待値を最小化するときに、真である。
（式１２）

この最小化問題は、本質的に、適応により実現されるであろう。例えば、エコー残差予測部ＥＲ（１）が、初期状態においては単位フィルタ（すべて値”１”）であると仮定されたい。すべてのフレームにおいて、サーチサーフェス（ｓｅａｒｃｈ ｓｕｒｐｈａｃｅ）の接線方向（ｔａｎｇｅｎｔ ｄｉｒｅｃｔｉｏｎ）に向かうにつれて、第１残差エコー予測ＥＲ_１（ｔ）は、増加するかもしれない。これは、ニュートンソルバ（Ｎｅｗｔｏｎ ｓｏｌｖｅｒ）アルゴリズムによって実現されてもよい。第２残差エコー予測部ＥＲ（２）は、最小エコー出力ｅ_３（ｔ）とスピーカ信号ｘ（ｔ）との間のエコー距離ミスマッチ（ｅｃｈｏ−ｄｉｓｔａｎｃｅ ｍｉｓｍａｔｃｈ）を含む第２残差エコー予測ＥＲ_２（ｔ）を決定するように構成されてもよい。図２Ｂのステップ２２４に示されるように、最小エコー出力ｅ_３（ｔ）とスピーカ信号ｘ（ｔ）との間のエコー距離ミスマッチから、例えば、ａｒｇｍｉｎ（Ｅ｛（ｅ_３（ｔ））^２／（ｘ（ｔ））^２｝）を決定することにより、第２残差エコー予測ＥＲ_２（ｔ）が決定されてもよい。ここで、ｅ_３（ｔ）が商（ｅ_３（ｔ））^２／（ｘ（ｔ））^２の期待値を最小化するとき、ａｒｇｍｉｎ（Ｅ｛（ｅ_３（ｔ））^２／（ｘ（ｔ））^２｝）は真である。ここでも再び、最小化は、ニュートンソルバアルゴリズムを用いて実現されてもよい。 The first echo residual prediction unit ER (1) generates a first residual echo prediction ER ₁ (t) including a cross-correlation analysis between the minimum echo output e ₃ (t) and the speaker signal x (t). It may be configured to. As shown in step 222 of FIG. 2B, from the cross-correlation analysis between the minimum echo output e ₃ (t) and the speaker signal x (t), for example, by determining the value of Equation 11, The difference echo prediction ER ₁ (t) may be determined.
(Formula 11)

Here, the value of Equation 11 is true when e ₃ (t) minimizes the expected cross-correlation value of Equation 12.
(Formula 12)

This minimization problem will essentially be realized by adaptation. For example, assume that the echo residual prediction unit ER (1) is a unit filter (all values “1”) in the initial state. In all frames, the first residual echo prediction ER ₁ (t) may increase as it goes toward the tangent direction of the search surface. This may be realized by a Newton solver algorithm. The second residual echo prediction unit ER (2) includes a second residual echo prediction ER including an echo distance mismatch (echo-distance mismatch) between the minimum echo output e ₃ (t) and the speaker signal x (t). ₂ (t) may be determined. As shown in step 224 of FIG. 2B, from the echo distance mismatch between the minimum echo output e ₃ (t) and the speaker signal x (t), for example, argmin (E {(e ₃ (t)) ² / The second residual echo prediction ER ₂ (t) may be determined by determining (x (t)) ² }). Here, when e ₃ (t) minimizes the expected value of the quotient (e ₃ (t)) ² / (x (t)) ² , argmin (E {(e ₃ (t)) ² / (x (T)) ² }) is true. Again, minimization may be achieved using a Newton solver algorithm.

The residual echo cancellation module 108 determines the minimum residual echo prediction ER ₃ (t) of the two residual echo predictions ER ₁ (t) and ER ₂ (t), and its minimum value ER ₃ (t) The filtered signal e ₃ (t) may be adjusted according to As an example, the minimum residual echo prediction ER ₃ (t) is the _one of ER ₁ (t) and ER ₂ (t) that has the smallest energy and the smallest correlation to x (t). Good. For example, as shown in step 226 of FIG. 2B, the minimum value of ER ₁ (t) and ER ₂ (t) is set, and as shown in step 228, the resulting value of ER ₃ is e Subtracted from ₃ (t), a residual echo cancellation filtered signal e ₃ ′ (t) is generated. If ER ₃ is equal to ER ₁ (t), the residual echo xe (t) is minimally removed when the intensity of the local speech signal s (t) is not zero. If ER ₃ (t) is equal to ER ₂ (t), the residual echo xe (t) is maximally removed when only the far-end echo x (t) is present (period of far-end utterance only). Is done.

As an example, second-order norms N (1) and N (2) may be calculated for the two echo residual prediction units ER (1) and ER (2), respectively.
N (1) = ‖ER (1) ‖
N (2) = ‖ER (2) ‖

Under double-talk situations, an echo residual prediction unit with a smaller norm may be applied to e ₃ (t) to remove echo residual noise. Under a single talk situation, an echo residual predictor with a larger norm may be applied to e ₃ (t) to remove echo residual noise.

In the echo cancellation, noise n (t) may be removed from the filtered signal e ₃ (t) or the residual echo cancellation filtered signal e ₃ ′ (t). However, such noise cancellation may not be desirable. This is because the remote recipient of signal e ₃ (t) or e ₃ ′ (t) may interpret the noise-free condition as an indication that all communication from microphone 104 has been lost. is there. To address this issue, the apparatus 100 may optionally include a noise canceller unit 110. The noise cancellation module 110 may be configured to calculate a predicted noise signal n ′ (t) from the microphone signal d (t), for example, as shown in step 217 of FIGS. 2A-2B. The predicted noise signal n ′ (t) may be attenuated by an attenuation factor α to form a reduced noise signal n ″ (t) = αn ′ (t). The attenuated noise signal n ″ (t) ) Is added to e ₃ (t) as shown in step 218 of FIG. 2A, or added to e ₃ ′ (t) as shown in step 230 of FIG. 2B, It may be incorporated into the microphone output signal s ′ (t).

  In an embodiment of the invention, the apparatus described in connection with FIGS. 1A-1C and the method described in connection with FIGS. 2A-2C are implemented as software on a system having a programmable processor and memory. May be.

  According to an embodiment of the present invention, a signal processing method of the type described in connection with FIGS. 1 and 2A-B and operating as described above is shown in FIG. It may be implemented as a part. The system 300 may include a processor 301 and a memory 302 (eg, RAM, DRAM, ROM, etc.). The signal processing device 300 may further include a plurality of processors 301 when parallel processing is implemented. Memory 302 includes data and code configured as described above. Specifically, the program code 304 and signal data 306 may be stored in the memory 302. Code 304 includes the echo canceling adaptive filters EC (1), ER (2), integrator 106, echo residual filters ER (1), ER (2), residual echo cancellation module 108, and noise canceller 110 described above. May be implemented. The signal data 306 may include a digital representation of the microphone signal d (t) and / or the speaker signal x (t).

  The apparatus 300 may also include well-known support functions 310 such as an input / output (I / O) element 311, a power supply (P / S) 312, a clock (CLK) 313, and a cache memory 314. The device 300 may optionally include a mass storage device 315 such as a disk drive, CD-ROM drive, tape drive for storing programs and / or data. The controller may also optionally include a display unit 316 and a user interface unit 318 to facilitate interaction between the controller 300 and the user. The display unit 316 may be a cathode ray tube type or a flat panel screen. They display text, numbers, graphic symbols, and images. The user interface 318 may include a keyboard, mouse, joystick, light pen, and other devices. Further, the speaker 322 and the microphone 324 may be connected to the processor 301 via the input / output configuration element 311. Processor 301, memory 302, and other components of system 300 may exchange signals (eg, code instructions and data) with each other via system bus 320 as shown in FIG.

  As used herein, the term input / output generally refers to any program, operation, or that transfers data to or from system 300 and to and from peripheral devices. Refers to the device. All data transfers could be considered output from one device and input to another device. Peripheral devices include devices that operate as input and output devices such as keyboards and mice, output only devices such as printers, and overwriteable CD-ROMs. Peripheral devices include external devices such as mice, keyboards, printers, monitors, microphones, game controllers, cameras, external Zip drives, scanners, internal devices such as CD-ROM drives, CD-R drives, internal modems, And other peripheral devices such as a flash memory reader / writer and a hard drive.

  The processor 301 performs digital signal processing on the signal data 306 in response to program code instructions of the program 304 stored and acquired by the signal data 306 and the memory 302 and executed by the processor module 301. Some of the code of program 304 may be one of a variety of different programming languages such as assembly, C ++, Java, or many other languages. The processor module 301 constitutes a general-purpose computer that becomes a special purpose computer when executing a program such as the program code 304. Although the program code 304 has been described here as being implemented as software executed on a general-purpose computer, task management is performed using hardware such as an application specific integrated circuit (ASIC) instead. One skilled in the art will appreciate that the method is implemented. As such, it will be understood that embodiments of the present invention may be implemented in whole or in part by software, hardware, or a combination thereof.

In some embodiments, among other things, the program code 304 may include a set of processor readable instructions for implementing a method having features common to the method 200 of FIG. 2A and the method 220 of FIG. 2B. Program code 304 may generally include instructions such as: That is, the processor 301 causes the microphone signal d (t) to be filtered in parallel by the first and second adaptive filters having complementary echo cancellation characteristics, and outputs e ₁ (t) and e ₂ ( an instruction for generating t), an instruction for determining the minimum echo output e ₃ (t) from e ₁ (t) and e ₂ (t), and an instruction for generating a microphone output using the minimum echo output.

  According to the embodiment of the present invention, more robust yet accurate echo cancellation is possible, which is possible only by cross-correlation analysis or voice activity detection (double talk detection). According to such an improved echo cancellation, it is possible to extract local speech s (t) from the microphone signal d (t) that is mostly occupied by the speaker echo x (t).

    Embodiments of the present invention may be used as presented herein and may be used with other user input mechanisms. A mechanism for tracking and measuring the azimuth direction and volume of sound, and / or a mechanism for actively or passively tracking the position of an object, a mechanism using machine vision, a combination thereof, and the like. The tracked object may include auxiliary controls and buttons that manipulate feedback to the system. Such feedback may include, but is not limited to, emission of light from the light source, sound quality distortion means, other suitable transmitters, modulators, control devices, buttons, pressure pads, etc. It is not a thing. It may affect the transfer and modulation of the same coding state and / or transfer instructions to and from the device being tracked by the system. Such devices are part of, or interact with, or affect the system used in connection with embodiments of the present invention.

  While the above is a complete description of the preferred embodiment of the present invention, it is possible to make various other variations, modifications, and equivalents. The scope of the invention should, therefore, be determined not by the above description, but should be determined by the following claims, including their full equivalents. The features described herein may be combined with any of the features described herein, whether or not they are preferred. In the following claims, unless expressly stated otherwise, the quantity of each element is one or more. The claims appended hereto are understood to include means plus function limitations in the given claims, except where explicitly indicated using the phrase “means for”. Must not be done.

Claims (30)

An echo cancellation method in a system having a speaker receiving a speaker signal x (t) and a microphone receiving a microphone signal d (t) including a local signal s (t) and an echo signal x ₁ (t), The echo signal x ₁ (t) depends on the speaker signal x (t),
Filtering the microphone signal d (t) in parallel with a first adaptive filter having a cancellation characteristic complementary to a second adaptive filter and the second adaptive filter;
The filtering characteristics of the first and second adaptive filters are such that when one of the first and second adaptive filters is not well adapted to the input, the other is well adapted to the input. ,
One of the first and second adaptive filters is well adapted because its filter function h (n) is stable, converges to a true echo path filter, and is neither overpredicted nor underpredicted When
The first adaptive filter is a voice activity detection filter;
The second adaptive filter is a cross-correlation analysis filter;
The method further includes
From the output e ₁ (t) from the first adaptive filter and the output e ₂ (t) from the second adaptive filter, the minimum echo output e ₃ () having a smaller correlation with the speaker signal x (t). determining t);
Generating a microphone output using the minimum echo output e ₃ (t);
A method comprising: Filtering the microphone signal d (t) in parallel with the first adaptive filter and the second adaptive filter,
adapting a set of filter coefficients of the first adaptive filter when the intensity of x (t) exceeds a threshold;
The method of claim 1, comprising analyzing a cross-correlation between e ₂ (t) and x (t) with the second adaptive filter. Determining the minimum echo output e ₃ (t) determining whether the second adaptive filter has not excessively removed the local signal s (t) by a filtering process;
The output of the first adaptive filter is used as the minimum echo output when the second adaptive filter excessively removes the local signal s (t) by a filtering process. Method. Determining whether the second adaptive filter has not excessively removed the local signal s (t) by a filtering process;
Taking a cross-correlation between the output e ₂ (t) of the second adaptive filter and the speaker signal x (t);
Determining whether an expected value of cross-correlation between the output e ₂ (t) of the second adaptive filter and the speaker signal x (t) is less than a predetermined threshold;
Determining that the second adaptive filter excessively removes the local signal s (t) by a filtering process when the expected value of the cross-correlation is less than the threshold value. The method described. Replacing the set of filter coefficients of the second adaptive filter with the set of filter coefficients of the first adaptive filter when the second adaptive filter excessively removes the local signal s (t) by a filtering process; The method of claim 3 further comprising: Generating said microphone output using the minimum echo output e _{3 (t),} above, the minimum echo output e ₃ first containing an analysis of the cross-correlation between _(t) and the loudspeaker signal x (t) Determining a residual prediction ER ₁ (t) in parallel, and a second residual prediction ER ₂ (including an echo distance mismatch between the minimum echo output e ₃ (t) and the speaker signal x (t) ( 2. The method of claim 1, comprising determining t). The method of claim 6, wherein the cross-correlation analysis includes calculating an expected value of cross-correlation between e ₃ (t) and x (t). The method of claim 6, wherein the echo distance mismatch includes calculating an expected value of {e ₃ ² (t) / x ² (t)}. Determining the first residual prediction ER ₁ (t) includes calculating a second-order norm N (1) of the first residual prediction unit ER (1);
Determining the second residual prediction ER ₂ (t) includes calculating a second-order norm N (2) of the second residual prediction unit ER (2);
The method further applies the echo residual predictor ER (1) or ER (2) with a smaller corresponding norm N (1) or N (2) to e ₃ (t) during double talk. the method according to steps including claim 6. Determining the first residual prediction ER ₁ (t) includes calculating a second-order norm N (1) of the first residual prediction unit ER (1);
Determining the second residual prediction ER ₂ (t) includes calculating a second-order norm N (2) of the second residual prediction unit ER (2);
The method further applies the echo residual predictor ER (1) or ER (2) with a larger corresponding norm N (1) or N (2) to e ₃ (t) during a single talk. the method according to steps including claim 6. Using the minimum echo output e ₃ (t) to generate a microphone output further includes determining a minimum residual echo prediction ER ₃ (t);
The minimum residual echo prediction ER ₃ (t) is one of ER ₁ (t) and ER ₂ (t) and has a minimum energy and a minimum correlation with x (t). The method described. The method of claim 11, further comprising the step of selectively removing _ER 3 (t) of from the minimum echo output _e 3 (t). The step of selectively removing the ER 3 _(t) from the microphone output, the duration of only far-end speech, according to claim 12 including the step of removing the most ER 3 a _(t) from the microphone output Method. The step of selectively removing ER 3 a _(t) from the microphone output, when the strength of the local signal s (t) is not 0, the step of removing a minimum ER 3 a _(t) from the microphone output The method of claim 12 comprising. Calculating a predicted noise signal n ′ (t) from the microphone signal d (t);
Reducing the level of the predicted noise signal n ′ (t) to form a reduced noise signal n ″ (t);
The method of claim 1, further comprising incorporating the reduced noise signal n "(t) into the microphone output signal. An echo cancellation device used in a system having a speaker and a microphone, the speaker being adapted to receive a speaker signal x (t), wherein the microphone is a local signal s (t) and an echo signal x ₁ (t). The echo signal x ₁ (t) is dependent on the speaker signal x (t), and is adapted to receive a microphone signal d (t) comprising
This device
A first adaptive filter connected to the speaker and the microphone;
A second adaptive filter connected to the speaker and the microphone in parallel with the first adaptive filter;
The second adaptive filter has an echo cancellation characteristic complementary to the first adaptive filter;
The filtering characteristics of the first and second adaptive filters are such that when one of the first and second adaptive filters is not well adapted to its input, the other is well adapted;
One of the first and second adaptive filters is well adapted because its filter function h (n) is stable, converges to a true echo path filter, and is neither overpredicted nor underpredicted When
The first adaptive filter is a voice activity detection filter;
The second adaptive filter is a cross-correlation analysis filter;
The device further includes
An integrator connected to the first adaptive filter and the second adaptive filter;
The integrator is configured to determine a minimum echo output e ₃ (t) from the output e ₁ (t) from the first adaptive filter and the output e ₂ (t) from the second adaptive filter. And
An apparatus in which the correlation between the minimum echo output e ₃ (t) and the speaker signal x (t) is smaller. The integrator is configured to determine whether the second echo cancellation filter is excessively removing the local signal s (t) by a filtering process, and the second adaptive filter is the local signal s (t). The apparatus according to claim 16, wherein the integrator selects the output e ₁ (t) as the minimum echo output e ₃ (t). A first echo residual prediction unit ER (1) connected to the integrator;
The apparatus according to claim 16, further comprising a second echo residual prediction unit ER (2) connected to the integrator. The first echo residual prediction unit ER (1) performs a first residual prediction ER ₁ (t) including analysis of a cross correlation between the minimum echo output e ₃ (t) and the speaker signal x (t). Configured to generate,
The second echo residual prediction unit ER (2) includes a second residual prediction ER ₂ (t) including an echo distance mismatch between the minimum echo output e ₃ (t) and the speaker signal x (t). The apparatus of claim 18 , wherein the apparatus is configured to determine The apparatus of claim 19 , wherein the cross-correlation analysis includes calculating an expected value of cross-correlation between e ₃ (t) and x (t). The apparatus of claim 19 , wherein the echo distance mismatch includes a calculation of an expected value of {e ₃ ² (t) / x ² (t)}. An apparatus further comprising a residual echo cancellation module connected to the first and second echo residual prediction units,
The residual echo cancellation module is
Calculating a secondary norm N (1) of the first residual prediction unit ER (1) and calculating a secondary norm N (2) of the second residual prediction ER (2);
The residual echo cancellation module performs the echo residual prediction unit ER (1) or ER (2) having a smaller norm N (1) or N (2) corresponding to the e ₃ (t 20. The apparatus of claim 19 , wherein the apparatus is configured to apply to: An apparatus further comprising a residual echo cancellation module connected to the first and second echo residual prediction units,
The residual echo cancellation module is
Calculating a secondary norm N (1) of the first residual prediction unit ER (1) and calculating a secondary norm N (2) of the second residual prediction ER (2);
The residual echo cancellation module performs the echo residual prediction unit ER (1) or ER (2) having a larger corresponding norm N (1) or N (2) during a single talk, e ₃ (t 20. The apparatus of claim 19 , wherein the apparatus is configured to apply to: An apparatus further comprising a residual echo cancellation module connected to the first and second echo residual prediction units,
The residual echo cancellation module is configured to determine a minimum residual echo prediction ER ₃ (t);
The minimum residual echo prediction _ER 3 (t), in one of the _ER 1 (t) and _ER 2 (t), in claim 19 correlation is minimum, and is x (t) the energy is minimum The device described. The apparatus of claim 24 , wherein the residual echo cancellation module is configured to selectively remove ER ₃ (t) from the minimum echo output e ₃ (t). The residual echo cancellation module, a period of only far-end speech, according to claim 25 that is configured to remove maximize ER 3 a _(t) from the microphone output. The residual echo cancellation module, according to the local signal s when the intensity of the (t) is not zero, the claim 25, which is configured to remove minimize the microphone output ER 3 a _(t) Equipment. A noise cancellation module connected to the microphone; and the noise cancellation module includes:
Calculating a predicted noise signal n ′ (t) from the microphone signal d (t);
Reducing the predicted noise signal n ′ (t) level to form a reduced noise signal n ″ (t);
The apparatus of claim 16, wherein the apparatus is configured to incorporate the reduced noise signal n "(t) into the microphone output signal. A microphone,
Speakers,
A processor connected to the microphone and a speaker;
An acoustic signal processing system comprising a memory connected to the processor,
The memory is
A speaker receiving a speaker signal x (t);
For implementing an echo cancellation method in a system having a microphone receiving a microphone signal d (t) including a local signal s (t) and an echo signal x ₁ (t) that depends on the speaker signal x (t) Stores a set of processor-readable instructions;
The processor readable instructions are:
Instructions for filtering the microphone signal d (t) in parallel with a first adaptive filter having a cancellation characteristic complementary to a second adaptive filter and the second adaptive filter;
The filtering characteristics of the first and second adaptive filters are such that when one of the first and second adaptive filters is not well adapted to its input, the other is well adapted;
One of the first and second adaptive filters is well adapted because its filter function h (n) is stable, converges to a true echo path filter, and is neither overpredicted nor underpredicted When
The first adaptive filter is a voice activity detection filter;
The second adaptive filter is a cross-correlation analysis filter;
The processor readable instructions further include:
From the output e ₁ (t) from the first adaptive filter and the output e ₂ (t) from the second adaptive filter, a minimum echo output e ₃ (t) with less correlation with the speaker signal x (t) is obtained. Instructions to decide,
Instructions for generating a microphone output using the minimum echo output e ₃ (t);
An acoustic signal processing system. A processor readable medium comprising a memory connected to a processor comprising:
The memory is
A speaker receiving a speaker signal x (t);
A processor for implementing an echo cancellation method in a system having a microphone that receives a microphone signal d (t) including a local signal s (t) and an echo signal x ₁ (t) that depends on a speaker signal x (t) Stores a set of readable instructions,
The processor readable instructions are:
Instructions for filtering the microphone signal d (t) in parallel with a first adaptive filter having a cancellation characteristic complementary to a second adaptive filter and the second adaptive filter;
The filtering characteristics of the first and second adaptive filters are such that when one of the first and second adaptive filters is not well adapted to its input, the other is well adapted;
One of the first and second adaptive filters is well adapted because its filter function h (n) is stable, converges to a true echo path filter, and is neither overpredicted nor underpredicted When
The first adaptive filter is a voice activity detection filter;
The second adaptive filter is a cross-correlation analysis filter;
The processor readable instructions further include:
From the output e ₁ (t) from the first adaptive filter and the output e ₂ (t) from the second adaptive filter, a minimum echo output e ₃ (t) having a smaller correlation with the speaker signal x (t) is obtained. Instructions to decide,
Instructions for generating a microphone output using the minimum echo output e ₃ (t);
Media containing.

Images