8

Is it possible to reconstruct the original pure signal? No, that is information-theoretical impossible. Also, that signal doesn't exist, probably, to begin with ;) However, you can definitely increase the the SNR simply by averaging; that becomes pretty obvious when you consider the signal of interest to be correlated within your recording, whereas your ...


6

Comments provided here are in two broad categories: Presentation and Subject matter. The "Presentation" section is the easiest to address. There are some things that could be rephrased in terms of language use but these might be just personal preferences. The "Subject matter" section includes comments in methodology which could take more time to address. ...


3

As Stanley Pawlukiewicz said: even under ideal circumstance, you can gain 3 dB of SNR per doubling of recordings. I.e., to increase SNR by, say, 15 dB, you'd need to average $$ 2^{\frac{15}{3}} = 2^{5} = 32$$ recordings. That alone shows that the whole thing isn't really practical: it just doesn't do much unless you use a crazy-high number of recordings. “...


3

If you are working on chat application (presumably web), then I suggest to take a look at WebRTC. It offers a noise suppressor that works ok for speech. Another option would be to use part of the Speex, which also has a noise suppression module.


2

I don't really find papers that discuss this difference. There are whole books on the subject: Robust Automatic Speech Recognition 1st Edition Do speech intelligibility and speech quality correlate with the accuracy of speech recognition systems? Usually no, usually noise reduction corrupts features in unpredictable way and reduces speech recognition ...


2

I would equalize both channels. The most robust method is to cut both back to a lower bandwidth. Then you only need to equalize the gains. Assuming that your scanning by hand a 1-2 second sweep rate seems reasonable. Unless your imaging, you can narrow the sweep down to smaller distances when a signal needs investigation. A 10-50Hz bandwidth should ...


2

Assuming that the sensors share the same characteristics, have the same timing (acceleration signals are aligned), the model with $y_1= x + n_1$ and $y_2= x + n_2$, $n_1$ and $n_2$ being uncorrelated noises of the same power, averaging them is a way to reduce the noise. The theory that asymptotically, averaging $N$ sensors reduce the variance by a factor of $...


1

I believe you are running into stability issues in an output power control loop design. See below a diagram of similar power control loops that I have implemented, where for stability reasons any filtering in the loop is minimized and only done with the loop filter itself which is designed for stability. The noise you are trying to filter gets filtered by ...


1

This is a beautiful way of noise/interference cancellation technique, please go through the link that I am sharing here. Certainly, you may find a way to implement the concept. https://drive.google.com/file/d/0B2bUtLEhrWp8Wi1JZzdub0U2Wm9JWlZEX290cHByZi1ES3FZ/view?usp=sharing


1

This is a tougher problem than it appears at the first glance. This may be confusing at first, but read carefully, and I hope that it will give a good idea what to do. Here is what is my experience and how I resolved a similar problem: apply a very good band-pass filter to get the frequency range of interest. apply a very tight DFT to get a good resolution ...


1

It's probably better to train your classifier with recordings made in a very large variety of typical noisy situations, rather than to try to modify the input to match that with which the classifier was trained.


1

in the audio biz, we call this the "Two-channel FFT". the cool thing about it is that you can measure the magnitude response of a room or something using music (that is decently broadbanded) as the test signal (and divide the output spectrum by the input spectrum). you don't need to pollute peoples' ears with an impulse train or a maximum-length-sequence ...


1

I've made some progress with plan B: Figure 1. Plan B: both coils equalized. Each Equalizer is a 3-tap FIR filter, which should be sufficient to cancel the apparent two poles of the coil (see Fig. 4 of the question). The impulse response of Equalizer 0 is $[c_2, c_1, c_0]$, and the impulse response of Equalizer 1 is $[d_2, d_1, d_0]$ Equalizer 0 is ...


1

The simplest way I can think of to remove noise from this kind of recording would be to use the noise removal functions in Audacity (https://manual.audacityteam.org/man/noise_reduction.html). This allows you to select a section of noise with none of the target signal and get a noise profile. You then remove spectral content that fits with this profile, ...


1

The "lub dub" sound of a heart beat is primarily between about 30Hz and 40Hz. A steep band pass filter for about 20Hz to 50Hz followed by amplification should bring out the beats themselves. If you need other heart sounds, then it gets difficult. Heart murmurs are a kind of wide band swishing noise - difficult or impossible to isolate using filters. ...


1

I will contribute to this question from my area of experience, which is in speech enhancement. For noise suppression, you can use classical speech enhancement algorithms (spectral subtraction, wiener filtering, etc) but they mostly introduce artefacts in the enhanced speech. Additionally, there are also machine learning approaches which you can try. I am ...


1

First of all the Wiener filter does not remove the noise but reduce it for WSS signals. It does this based on the relationship between the power spectral density of the clean signal $x[n]$ and imposed noise $v[n]$, both assumed as WSS. The frequency response of a noncausal IIR Wiener filter is given as $$ H(\omega) = \frac{ P_x(\omega) }{ P_x(\omega) + ...


1

Plus is the simplest operation : $z = x+y$. Fourier is inherently linear, and good at addressing it. However, most processes and data combination are nonlinear, and they should be dealt with. The second simplest operation is multiplication. Homomorphic filtering deals with $z = x\times y$. It is more complicated, especially because of zeroes. Logarithms ...


1

Here is a rough cut. You could definitely get more elaborate on your peak testing, e.g. comparing it to neighbor peaks. This simply makes a metric for how far the peak "sticks out" and uses a cutoff value on that and the height to select a candidate peak. In the case of two or more isolated peaks in your field of zeroes, it would select all of them. Not ...


1

A typical first approach to solve your problem would probably be to use one of the many existing noise reduction algorithms that target speech signal applications. These algorithms typically consist in noise PSD estimation and noise reduction filtering. For each task, there are a variety of algorithms to choose from. Example noise PSD estimation algorithms ...


1

There are 2 properties which are important to know about your noise: Its Self Correlation What usually called Auto Correlation. This will tell you the relation between 2 noise samples which are apart (On time) from each other (I'm assuming the signal is stationary, namely its properties doesn't change over time). Best thing to hope for is White Noise. Its ...


1

I would recommend checking out "Design of active noise control system with the TMS320 family" by Sen Kuo [[report link]]. TMS320 chips are specifically designed to perform various ANC tasks. In the report, they even provide code for programming LMS and FxLMS. In a lot of the literature on real-time ANC, the control system is implemented using these chip ...


1

I went ahead with the proof of concept. $60 is a minimal investment to make and implementing high-school-level science experiments at home is never a bad idea. What I discovered is that yes, low frequency ANC is possible with a simple analog setup for predictable read-world noise, but no, it's not practical in any useful way in an open-space environment. ...


1

I have found two ways to do it: I can use image calculator and divide image of interest by background image and display the result in a new window as 32-bit (float) result. But the photo will be black and white. Or adjust color threshold for background image in such a way that only the noise will be filtered and than select it. On the image of interest ...


Only top voted, non community-wiki answers of a minimum length are eligible