DSPNR and Notch2- Northern Utah WebSDR

Northern Utah WebSDR

The "DSP Noise reduction" "Notch2", "High Boost",
"Vari-Notch", "CW Peak" and Synchronous AM features

"Autonotch" is now "Notch1"

The filter that had been labeled "Autonotch" is now "Notch1".

Unlike the new "Notch2" filter, this "Notch1" filter operates on the server side of the signal path and its operation can reduce the strength of strong CW notes (carriers) and reduce the S-meter deflection and prevent AGC action on that signal.

Unfortunately the original "Autonotch" notch filter (a.k.a. "Notch1") is susceptible to other strong signals meaning that it can be prone to "hunting" in the presence of strong modulation - and with weaker signals, it may not reliably "lock" on CW notes at all.

"DSP Noise Reduction", "Notch2" and "High Boost"

Operational notes:

Because these these new features can take significant resources on the user's device, they are not available on the "mobile" version of the Northern Utah WebSDR pages. This is done in order to allow their use on the widest variety of mobile/portable devices (e.g. use less processing power for units with less processing power) and also reduce battery consumption.

Recordings made on the Northern Utah WebSDR WILL reflect the settings of these features (e.g. the recordings will contain noise reduction, notch filtering, etc.)

DSP Noise reduction:

Also added is a noise reduction filter that can take out some of the background "hiss" and improve the signal-noise ratio. This is the same type of filter may be found on nearly all HF transceivers these days:

It can reduce the background hiss on medium to strong signals: Reducing this hiss can reduce fatigue.

Weaker SSB signals may benefit from noise reduction to improve intelligibility.

The readability of CW signals may be improved by the nature of the filter to converge on coherent signals (like CW notes) and reject noise.

Like any filters of this type, it probably will not help with signals that are very weak and "in the noise".

In the presence of only noise, this filter will cause a "hollow" barrel-like sound: This is normal and is the nature of the filter.

"I reloaded the page, but DSP noise reduction (or related option) suddenly isn't working"

There's a quirk related to the "tick" boxes and drop-down settings on the WebSDR - and other similar boxes on other web pages and that is that if you reload or refresh the web page, the boxes or selections may look as though they are active, but they will not be.

For example: If you were to select a DSP setting of Low, the DSP noise reduction would start working - but if you refreshed the page, it may still show that selection, but the DSP noise reduction would not be active.

The work-around is to select Off (or any other setting than the current one) and then select Low again to reinstate the filter.

This same quirk applies to practically every option on the WebSDR page from the frequency to the waterfall setting to the volume setting - and everything in between. This is not a bug per-se, but just the way the HTML/Javascript that is being used works.

There are currently four settings for the DSP Noise reduction filter:

Low. The noise-reduction effects are fairly light and is probably the best choice for casual listening.

Medium. This has stronger noise-reduction effects, but more apparent artifacts, particularly on background (white) noise when no signals are present in the form of a "swishing" sound.

High. This has even strong noise-reduction effects: It may or may not help with intelligibility

Strong. This filter is designed to have even stronger noise-reduction: Its adaption rate is quite slow compared to the others, but it can be fairly effective in some cases. This setting often tends to make audio sound a bit "quieter" and cut the "highs" a bit. The "High Boost" feature was added to help mitigate this effect.

Extreme. This noise reduction setting may be a bit "over the top" in its strength, but it may be useful to someone!

A few comments about DSP noise reduction:

DSP noise reduction works by quickly "locking on" to the voiced elements of human speech and filtering out the rest - and this works because voiced human speech consists of tones. For signals that range from good to "moderately" noisy, this works quite well - but when signals are very weak, it may actually reduce intelligibility in some cases because the "tones" of the speed are increasingly difficult to discern from the background noise. When only noise is present or signals are very weak, the noise reduction tries to lock on to the random elements of noise causing the "swishing" (or "voices in a barrel") effect.

Because of its nature, this filter will also "lock on" to tones and it may be useful for CW use in some cases. Note that this filter is adaptive and it takes time for it to "lock" onto a signal - and the higher settings take even more time to do this, so it may not be the best choice for rapid-fire contesting.

For casual listening, the Low setting is recommended as its effects are moderate, offering reasonable noise reduction without being too strong or causing deleterious effects on weak signals.

Use of DSP noise reduction with digital modes:

It is recommended that DSP noise reduction be turned OFF if you are trying to use any digital mode - and this is true not only for the WebSDR, but for ANY radio with DSP filtering.

The "Notch2" filter:

This filter uses digital signal processing techniques to detect a coherent signal - like pure tone like a CW note (carrier) - and block it out. Because it is in the audio path its action will have no effect on the S-meter, but is may be more effective in removing weaker carriers that may still be annoying, but too weak to be reliably locked onto by "Notch1". For very strong interfering signals you may wish to use both Notch1 and Notch2.

Of course, one should not try to use any notch filter with CW or digital modes as this filter's job is to remove the very signals comprising such signals!

The "Vari-Notch" filter:

This is a simple, manually-tunable notch filter that may be used in addition to the notch filters above. It is most useful for a single, unchanging tone (e.g. carrier from a broadcast station) - but like "Notch2", it has no effect on the S-meter reading and will not prevent receiver desense from strong signals.

The "CW Peak" filter:

This is a simple, moderate-Q band-pass filter that is intended to reduce "ear fatigue" for CW operation by attenuating - but not eliminating - off-frequency signals. Related to this is the "Ref Tone" button that, when activated, produces a tone that is exactly that of the bandpass filter to allow quick and precise adjustment of the bandpass filter.

The "High Boost" setting:

This setting boosts the high audio frequencies by 6 dB with the "knee" of the upward curve centered on 1500 Hz. The purpose of this setting is to boost the "highs" which can boost the intelligibility of speech.

This feature may be helpful if you are using a higher setting of the DSP noise reduction, or you may have age-related hearing loss.

Synchronous AM:

While it may not seem like it's a DSP feature, doing Synchronous AM reception on the PA3FWM WebSDR really is. Because the WebSDR server code itself isn't set up do this, frequency shifting and carrier recovery is done at the user end of the pipeline. Because of the way that it must be implemented, there are a few restrictions:

Only UPPER or LOWER sideband AM may be received: The bandwidth (at least with the server configured for an 8 kHz channel) is not sufficient to allow both sidebands to be passed.

The total bandwidth is limited to a bit less than 4 kHz. What this means is that the shifted carrier and all audio must fit within this bandwidth and that the amount of allowed carrier shift is around 1 kHz. When a carrier is tuned in "exactly", the receiver on the WebSDR side is actually offset by 150 Hz to allow a bit of frequency variation of the signal. While this is typically enough for SWBC signals, you may need to offset tuning a bit - or switch between SAM-U and SAM-L (upper and lower sideband) when listening to an AM roundtable as many users are operating older radios that may be several hundred Hz off frequency.

When in synchronous AM mode, the status of reception appears just below the frequency display bar (and above the "Mode" indicator) showing whether the synchronous carrier recovery is locked, the apparent frequency of the received carrier (any offsets/drifting of the receive hardware at the server are included here) plus the "NCO" frequency - an internal, software oscillator that is used to shift/lock onto the audio: The NCO must be between 50 and 1000 Hz for proper locking.

Heavy/fast QSB or QRM can cause momentary instabilities in the signal lock.

Beacuse the synchronous AM is decoded on the user's computer, certain on-screen functions (waterfall zooming, shifting, etc.) can cause the synchronous AM detector to lose lock momentarily owing to software pre-emption.

1 Hz tuning steps:

This is another feature that is actually DSP . Natively, the WebSDR tunes only in 31.25 Hz steps so extra code was written to run on the client side to "shift" the audio by the residual amount (e.g. that between the 31.25 Hz steps) to effect precise 1Hz tuning.

CatSync users - NOTE: If you use "CatSync" it MAY not know what to do with the extra (1Hz) digit and work improperly. In this case, add "?10hz" to the URL (web address) and reload the page to force 10 Hz tuning resolution. For more information about URL-based commands see the "URL Parameters" page - link.

Any problems that might be caused by these features?

The above functions are client side - that is, they are run on your computer rather than the WebSDR server itself. What this means is that they will increase processor load on your end - something that may cause intermittent audio issues on slower computers - particularly if you have minimized the browser window in which you are running the WebSDR. In early 2022, these features were rewritten to dramatically reduce computer load and improve performance, but they can still task slower computers.

It has also been observed that in general, the Firefox browser seems to work better than most other browsers. The Chrome browser seems to have more trouble on some computers running the WebSDR's web page, causing issues such as audio drop-outs. If this happens, check to see if the sound card's speaker output sample rate is set to 48000 Hz (in Windows) rather than something higher like 96000 or 192000 - or simply try using FireFox.

Where did these features come from?

These added features (DSP Noise Reduction, Notch 2 and High Boost) have been added to the client side of the code, written by the folks at the Northern Utah WebSDR by adding software "hooks" to intercept the audio stream and not with the direct participation (but with the knowledge of) the original author of the WebSDR software, P.T. de Boer, PA3FWM. There are no modifications of any sort to the websdr binary files: All processing - including the increased tuning resolution - are being done to the original audio stream produced by the WebSDR software.

If you have questions about these added features, or if you operate a WebSDR system and are interested in adding them to your system please direct inquiries to sdrinfo@sdrutah.org and not PA3FWM, the original author.

Updates:

As of 28 May, 2022 a complete rewrite of the above features was implemented on all of the Northern Utah WebSDR's servers, reducing processor load on the users' computers, improving performance and adding freatures, including the "CW Peak", "Vari-Notch", "Synchronous AM", and finer tuning resolution modes - See the "FAQ" page under the heading "Features of the WebSDR".

Additional information:

For technical information about this WebSDR system, go to the technical info page (link).

For the latest news about this system, visit the latest news page (link).

For answers to possible questions, visit the FAQ page (link).

For more information about the WebSDR project in general - including information about other WebSDR servers worldwide and additional technical information - go to http://www.websdr.org

Go to the Northern Utah WebSDR landing page