DNR Dev results (sneak peak)

[Overture]

Ahoy there all,

Been writing R+D reports for various submission requirements and thought I'd share the results of the realtime DNR. Because we all love visual representations of audio results.
This is taken straight from the report. The content used was an old recorded tape of The Goon Show from decades ago.

Initial Signal (Goon Show)

Cleaned signal (Goon Show)​

As can be observed quite clearly above, the cleaned signal has significantly reduced high-frequency noise content. It is important to note that the gap between the first two sections of speech is not intended to be silent. There is studio noise and quiet laughter here. This demonstrates both the effectiveness of using a highly multi-band system, and the way the system compares against the average signal content in each individual band separately. The result here is nothing short of fantastic, the noise is being dramatically filtered out without compromising the audio. The strength of the noise reduction was at about 50% on the adjustment knob – this adjustment is important as it allows for fine tuning the compromise between clarity and range.


This can also be observed in the spectrogram of the same section of audio played twice, on the left untouched and on the right with noise reduction. The amount of noise is clearly reduced down to well below audible level. This measures as anywhere from a 10 to 40dB reduction in noise.

The left is the original audio, the right is the cleaned audio in realtime. Scale is 20-20K on the vertical axis, time on the horizontal axis.
Note that it’s not merely a removal of those upper frequencies, as there is still content there across the audio sample.
We have done tests with music as well, and it sounds fantastic - it's just more obvious visually using voice and broadcast recordings.

Hope you're all having a pleasant time.

~D (Captain)

 

[kfbkfb]

If the music/content is only in the lower midrange and treble with noise in the midrange, will this system reduce the noise in the midrange with a minimal effect on the lower midrange and treble?


Kirk Bayne

 

[Overture]

If the music/content is only in the lower midrange and treble with noise in the midrange, will this system reduce the noise in the midrange with a minimal effect on the lower midrange and treble?


Kirk Bayne

Indeed it should.

 

[chucky3042]

OK I am biased but I remember back in the cassette era, Philips were dead against Dolby compression/ expansion as it meant the cassette would no longer be universal for machines that were not Dolby B. So Philips released their own DNL but unfortunately you could hear it restricting the top end.

I honestly think if we had a time machine and had this technology back in the late 60's Dolby would never have happened.....as much as I like Ray Dolby RIP.

 

[barfle]

I have about 100 Edison Diamond Discs, and about the same number of 78s that all have the ravages of age and abuse in them. I also have an Esoteric Surface Noise Reducer in my system, but it’s only effective on stereo recordings.

It looks like this is effective on a single channel. I’d love to try this discovery on those old, old (some well over a century) recordings.

I’m amazed at the research you guys are doing. I never worked at a place that allowed such depth of study. I hope it makes you millions.