Decimal places in effect settings

[steve]

I expect it will be years before we see logarithmic sliders.

Not necessarily

The poster who wants 2 decimal places in Normalize must have a reason for asking,

The user that wants 192 kHz sample rate for audio recording has a "reason", but it is not a reason supported by physical facts.
I'm glad that Audacity does have the ability to record at extremely high sample rates, but not because it offers any benefit for audio recording (it doesn't, in fact there is strong evidence that excessively high sample rates can have a detrimental effect on the quality of audio recordings). I support the high sample rate feature because we have a significant number of users that use Audacity for non-audio purposes such as ultrasonic recording (eg. recording bats).

The user that records at a level "as high as possible" also has a reason for doing so, but mostly this is due to misunderstanding the nature of digital recording (often basing their idea of "good recording level" on experience of tape recording in the 1960's). This we actively discourage through both our documentation, and visual clues such as having a "red zone" in the default meters,

The fact is, that a 0.01dB difference in normalization level is insignificant. Should we (is it responsible) for us to give credence to a misconception by providing multiple decimal places in this effect?

Clearly there is some potential value in it.

If someone gave a practical example for why it was needed, then I could probably be persuaded, but no example has been given and I don't see one, so I'll not be the developer to add a feature for which I see no practical benefit.

Moving on, are there any other effects which you think need updating for control range and/or control precision?
I've left "Compressor" out of this for the moment as I wish to make bigger changes to that effect.

[юра00]

About effects "Leveller" and "Trancate silence (Level)": Noise threshhold is 30, 35, 40 db now, step 5 db. Is it possible to improve to more accurate 31, 32, 33, 34, 35, 36, 37, 38 db... (step 1 db) and for "Leveling degree" put additional values between "Light", "Moderate" and "Heavy"?

[steve]

Thanks for the suggestions юра00.

The Leveler effect is overdue for replacement, so I'll leave that for now.
If you need an easy way to reduce the dynamic range, I would recommend using the new Limiter effect rather than the Leveler as the Limiter can reduce the dynamic range more effectively and with much less distortion than the Leveler effect.
For users that want a distortion effect (the other main use of the Leveler effect) I hope that we will soon have a proper distortion effect to supersede the Leveler.

I'm not sure what to do about the Truncate Silence effect. This effect has always had a unique way of selecting the level (the dropdown menu) and the 5 dB steps are definitely an intended feature. It would have been much easier for the developer(s) to have used a "normal" slider / text box than the dropdown menu, but they purposefully chose to use 5 dB steps. It has been like this for many years, and to my knowledge there have been no complaints about the step size. Also, in my own use of this effect I have always found the 5 dB steps to be perfectly adequate. Consequently I'm hesitant to change the intended design.

[Robert J. H.]

Thanks for the suggestions юра00.

The Leveler effect is overdue for replacement, so I'll leave that for now.
If you need an easy way to reduce the dynamic range, I would recommend using the new Limiter effect rather than the Leveler as the Limiter can reduce the dynamic range more effectively and with much less distortion than the Leveler effect.
For users that want a distortion effect (the other main use of the Leveler effect) I hope that we will soon have a proper distortion effect to supersede the Leveler.

I'm not sure what to do about the Truncate Silence effect. This effect has always had a unique way of selecting the level (the dropdown menu) and the 5 dB steps are definitely an intended feature. It would have been much easier for the developer(s) to have used a "normal" slider / text box than the dropdown menu, but they purposefully chose to use 5 dB steps. It has been like this for many years, and to my knowledge there have been no complaints about the step size. Also, in my own use of this effect I have always found the 5 dB steps to be perfectly adequate. Consequently I'm hesitant to change the intended design.

And that's just what I was going to complain about...

The multiple choice control is fine, especially for vi users.
However, 5 dB steps are not.
The problem is that 5 dB can cover 50 per cent of the audio.
The audio should actually be pre-processed and the threshold be expressed as per centage.

Sorry, I'm not fit at the moment, otherwise I could post some code to support what I'm aiming at.

Robert

[steve]

Thanks for the comments Robert.
As this refers to such a long-standing feature, I'll raise it as a separate feature request.
Hope you're better soon.

[steve]

As this refers to such a long-standing feature, I'll raise it as a separate feature request.

Posted here: http://forum.audacityteam.org/viewtopic ... 44#p281244

[Gale Andrews]

The fact is, that a 0.01dB difference in normalization level is insignificant. Should we (is it responsible) for us to give credence to a misconception by providing multiple decimal places in this effect?

I think the whole case (which is an edge case) probably turns on some scientific use, not practical, or there is some reason the user cannot or does not want to use Amplify.

Wanting to apply offset removal (in one step rather than two) is an obvious reason.

Wanting to bring up multiple tracks to the same level at > 1 decimal place accuracy in one step rather than very many steps is another.

I don't know, perhaps a user wants to maximize and 0 dB distorts in their equipment but -0.05 dB does not?

So I don't agree with you that no examples have been given. Nor do I see any harm in three decimal places for Normalize given Amplify now has three and we could show only one. I think you'll just invite more questions as to the apparent "discrepancy" between the two effects.


Gale

[steve]

Wanting to bring up multiple tracks to the same level at > 1 decimal place accuracy in one step rather than very many steps is another.

It would be two steps.
1) Normalize to the nearest 1 dB
2) Amplify by the decimal part.

You agree that this is an edge case, and as such I don't see a problem in there being one extra step required for this edge case. It's not as if we are preventing the user from achieving the effect.

Another "edge case" is the user that wants to amplify or normalize the channels of a stereo track to different levels, We make no attempt to cater for this edge case, requiring that the user splits the track, amplifies or normalizes each track separately, then joins the tracks together again.

I don't know, perhaps a user wants to maximize and 0 dB distorts in their equipment but -0.05 dB does not?

No problem - Normalize to -0.1 dB. The difference is totally imperceptible.

I think you'll just invite more questions as to the apparent "discrepancy" between the two effects.

but you're not concerned about the discrepancy between Normalize and the Level control in Bass and Treble? or are you saying that that should also have more decimal places?

I think that it's more likely that if we give Normalize more decimal places we will be inviting criticism of "Why on Earth have you given Normalize so many decimal places "

Anyway, I'm working on some changes to the Normalize effect which may end up giving you the additional precision as a side effect of improved usability.

[юра00]

steve:
It would be two steps.
1) Normalize to the nearest 1 dB
2) Amplify by the decimal part.
It is impossible, immediate "Normalize" (if it would have been) to -1,12 db is not the same, that "Normalize" to -1,1 and then "Amplify" to -1,12 db. These effects are nearly, but not entirely identical from the point of achieved result.

[steve]

Let's say that you want to normalize to -1.12 dB.
1) Normalize to -1 dB
2) Amplify by -0.12 dB
The result is that the peak level is -1.12 dB.
In both Amplify and Normalize, the dB measurement is relative to "full scale" (0 dB), so -1.12 dB is -0.12 dB less than -1.0 dB.

A simple example to show that it works:
1) Normalize to -6 dB. The peak level is now -6 dB
2) Amplify by -3 dB. The peak level is now -9 dB.