I upgraded to Debian LXQt 12.6.0 last month, having Audacity 3.2.4. There is a little problem with the version, the horizontal slider doesn’t appear unless the window size is altered. But no issues with this.
I have been using the Distortion / Leveller for a long time. But even with 100% Levelling fine adjustment and the max 5 for the Degree Of Levelling, lower volume areas aren’t restored well.
Also, the Bass sounds are enhanced more than the middle section and the treble.
Could I use the Nyquist Prompt with a code to piece-wise amplify, section by section to be amplified to a 1dB peak, irrespective of their original peaks?
Either
(a) say each of 1ms segment for an audio file -
Or
(b) for the max peak for each of the segments between two zero audio signals as creating a boundary -
to be amplified to a 1dB peak.
Is there already a Nyquist code available?
Of course, I am aware that the result might not be audibly optimal. But I want to experiment to later fine-tune the code further.
Apologise for the belated reply, Mr. Steve.
Thank you.
I will try and get back with my feedback.
[follow up]
Used once. But consumes a lot of computational time (resource), easily more than 10 times for a 30 min mono audio file.
The output isn’t optimal even with Squelch Threshold set at -60dB, and Attenuation set at -30dB doesn’t yield a result better than my current approach: distortion → Leveller (max, i.e., Noise Floor: -80dB, Fine Adjustment: 100% and Degree of Levelling: 5) → Amplify → Equalizer (only Bass reduced to -8dB).
Thank you, Mr. Trebor, for replying to my post. I checked the page to understand and visualise Compressor and Limiter.
I also looked at the illustration, with scales for the Compression Curve and Smoothing, and a curve in the third quadrant of an x-y coordinate frame.
I must admit that I couldn’t understand the idea behind Compression. Smoothening I can intuitively understand.
Should I create a new post to receive clarifications and explanations on the aspect of Compression?
As a physicist, I am aware of the mathematical basis behind the Attack and Release as well as the other terms like Decay and Sustain involved in this framework.
But I am completely at sea for the terms like Threshold, Make-up gain, Ratio (could be between the Threshold and Make-up gain) as I don’t understand the basis behind these terms.
On my computer it takes much less than 30 seconds to process a 30 minute track. Ensure that you are using the “latest version”. Also, the “Music” setting is faster than the default setting as it disables the optional filtering.
I downloaded from the link that you’d given. I used the “Music” setting.
I just posted my first experience. I will get back to the utility once the tasks I have are over.
Thank you for your feedback. Such inputs help a great deal, Mr. Steve.
Thank you very much for the link, Mr. Trebor. Your promptness and kindness to support is admirable.
[postscript]
It is a long paper in details, Mr. Trebor. Editing sound files is a necessity arising out of my hobby to record talks for later reference, not my profession. My interest is in HEP & Cosmology.
So I would have preferred a one page explanation. But thank you.
The code bit enhances the Bass. So I used Bass and Treble... filter and applied -8dB bass reduction twice to get a desirable sound file.
The processing computation is very fast. Finishes in less than a minute in my old computer.
Thank you, Mr. Trebor.
In the coming days I would thoroughly test Mr. Steve’s suggestion also, as he said that his suggestion takes a very short time to compute.
[postscript]
No wonder I couldn’t understand the terms in the afore-mentioned screenshot of the option window!
The correct meaning isn’t reflected by the word compression alone. The sub-phrase is Range-Compression, here the word range means a complete set of values from zero amplitude to max amplitude for a peak or an RMS sampling bit. The further qualifying word for Physical, Real Time Electronic devices is Dynamic.
I briefly read the paper, a link to which was posted by Mr. Trebor. To me, it appears neither very well composed, nor very attractive, nor optimally compact. But I would definitely thank Mr. Trebor for bringing it to my attention, as it provided a step for me to begin my exploration. Better something than nothing.
Aren’t there good books (illustrated, with diagrams, rather than word jungles) written on acoustical physics, particularly on these advanced acoustical aspects?
Ok. Yes …, that should be intuitively apparent, as the DRC is on a logarithmic sound sample, not a linear sample, and only an approximation limited by the carrier wave frequency.
I have a proposal/suggestion to make: wouldn’t DRC be better rationally represented by DARC? A represents Amplitude.
Also there are already terms like Slope,Gradient, inflection, maxima, minima, etc., in use for curves since 1700s. So don’t the introduction of new terms like Knee to represent the same ideas in curves for acoustics lead to bloating/obfuscation?
The graph shows you what it’s doing except the axes are not labeled (X is input dB and Y is output dB). And it’s only 2-dimensional so it doesn’t tell you any thing about attack and release.
My quick description of dynamic compression:
Compression reduces the dynamic range (or “dynamic contrast”) by making loud parts quieter and/or quiet parts louder. In practice, usually it “pushes down” the loud parts and then make-up gain is used to bring-up the overall volume, making everything louder.
Limiting is a fast-kind of compression. Clipping is a BAD kind of limiting.
Automatic gain control, automatic volume control, or leveling are a slow-kind of compression.
These aspects of approximation, sampling, carrier frequency, logarithmic scale, etc., ensure that both the positive and the negative peaks that are pushed down to the 0dB level can’t be restored from the final result. What is lost after application of a filter remains lost, never to be retrieved back from the resultant sound sample.
Though the matter may appear intuitively apparent, on hindsight it isn’t very simple to visualise. This compelled my return to this aspect once again to try a better clarity.
This issue isn’t a simple shifting of the amplitude axis to reduce the amplitude for a sound sample. While the amplitude peaks are attenuated, the 0 of the amplitude axis doesn’t shift to a new lower value. While 0dB is silence, marked by the speaker membrane at absolute unstretched rest, if the peak +ive amplitude means a maximum push forward of the stretched speaker membrane, the -ive amplitude is a maximum push inward, again stretched in the opposite direction.
Mr. Trebor, it seems that this time I have done justice to clarify your comment.
Silence has an amplitude (not dB) of zero (zero volts if an electrical signal or a numerical value of zero in a digital file, etc.).
But decibels need a reference and pure silence is minus infinity dB. Minus infinity is easy in digital but it doesn’t exist acoustically or electrically. There is always some noise in the real-analog world.
If you try to calculate dB for an amplitude of zero your calculator or spreadsheet can’t do it and it will report an error.
Digital dB levels are measured in dBFS (decibels full-scale) where the 0dBFS reference is the “digital maximum” so digital dB levels are usually negative.
The meters in Audacity show negative dB with 0dB as the maximum.
The scale to the left of the waveform that goes from -1 to +1 with zero (silence) in the center and it’s showing amplitude. + 1 and -1 represent +/-100% or 0dB.
Acoustic levels (loudness in the air) is measured in dB SPL where the 0dB reference is approximately the quietest sound humans can hear so yes, with SPL, 0dB is “silence” and SPL levels are positive.
With analog electrical signals there are several different 0dB references and you have to specify which one you are using (dbV, dBm, and dBu are common).
There is no automatic calibration between digital and acoustic levels. It obviously depends on the volume control and other factors. But there is a direct correlation and if you turn-down the digital level (more negative) by 3dB, the SPL will also go down by 3dB.
Yes, correct!
dB needs a reference, a sound signal of a standard value from a standard distance.
I would have edited the post, replacing the dB of “0dB” in the said post with Amplitude.
Thanks, Mr. DVDdoug, for pointing out.
