Decimal places in effect settings

[steve]

Following on from this topic: http://forum.audacityteam.org/viewtopic ... 20&t=86432

Note re: "trailing zeros".
The numeric input boxes allows formatting for how whole numbers are shown. For example, with "trailing zeros = 1", the number "2" will display as "2.0" provided that the maximum number of decimal places is at least 1.

For some effects, the more rigorous validation of user input in Audacity 2.1.1 has reduced the range and flexibility of some effects. To counter this I propose that the following effects are updated to allow more precise setting of parameters.

• Amplify.
  In certain cases, notably when attempting "phase cancellation", high precision is required.
  • Maximum number of decimal places: 3
  • Trailing zeros: 1
• Bass and Treble
  Previously it was intended that the user could enter values beyone +/- 15 dB by text entry. As this is no longer possible:
  • Max/Min boost/cut for bass and treble controls: +/- 30 dB
• Paulstretch
  Very short time resolution can produce interesting effects, particularly on vocals.
  • Minimum Time Resolution: 0.001
  • Maximum number of decimal places: 3
  • Trailing zeros: 1
• Phaser
  In musical applications, it may be desirable to synchronise the sweep frequency with the tempo of the music (requiring high precision).
  • Minimum frequency: 0.001 Hz
  • Maximum number of decimal places: 5
  • Trailing zeros: 1
  Also, there a couple of bugs to fix:
  • The maximum amount of feedback can cause infinite gain. This needs to be slightly reduced to prevent "inf" and "NaN" values.
  • Adjusting the "number of stages" may change the "phase" slider (probably a typo in the code).
• Wahwah
  As with the phaser effect, for musical applications it may be desirable to synchronise the sweep frequency with the tempo of the music (requiring high precision)
  • Maximum number of decimal places: 5
  • Trailing zeros: 1
  .

Any others while we are at it?

I am not inclined to increase the number of decimal places for the Normalize effect as, after three pages of discussion, no-one has come up with a single use case to justify doing so,

I am not inclined to increase the range of the Limiter effect as in my opinion, doing so will only server to encourage bad recording/production practice.

[Gale Andrews]

I am not inclined to increase the number of decimal places for the Normalize effect as, after three pages of discussion, no-one has come up with a single use case to justify doing so

I thought Robert did so (fractional dB can be heard in multiple tracks)?

And as I said, different decimal places in Amplify and Normalize does "look" inconsistent, which is what the user is seeing. Making Amplify and Normalize into one effect would remove the inconsistency.

I am not inclined to increase the range of the Limiter effect as in my opinion, doing so will only server to encourage bad recording/production practice.

Well that's OK if we allow values outside the slider range.

Gale

[steve]

I thought Robert did so (fractional dB can be heard in multiple tracks)?

I thought that was with reference to the Amplify effect (in which case I agree, which is why I propose increasing the maximum number of decimal places in the Amplify effect from 2 to 3).

And as I said, different decimal places in Amplify and Normalize does "look" inconsistent,

Even more "inconsistent" if I increase the number of decimal places in the Amplify effect to 3.
We are even more inconsistent between the built-in generators and the Normalize effect in that the generators allow up to 6 decimal places. I assume that you agree that 6 decimal places is overkill for the Normalize effect?

Making Amplify and Normalize into one effect would remove the inconsistency.

Yes it would, if that ever happens.

Well that's OK if we allow values outside the slider range.

Nyquist plug-ins do not allow input beyond the slider range, However, in this case, that is not a "limitation" (excuse the pun )
Put simply, peaks below -10 dB do not need to be limited. The purpose of a "limiter" is to "limit" excessively high peaks.If the peak level is below -20 dB then there are no excessively high peaks, so the limiter should (and does) do nothing.

[steve]

Also worth noting that the Audacity Limiter has a "Limit to;" setting, not a "threshold" setting. When using the default "Soft Limiter" the threshold (the level at which the effect starts to act) is several dB below the "limit to" level. The reason for using a "limit to" level, is that it is easier to use, If you want your maximum peaks to not exceed -3dB, then all you need to do is to set "limit to: -3 dB".

[Gale Andrews]

And as I said, different decimal places in Amplify and Normalize does "look" inconsistent,

Even more "inconsistent" if I increase the number of decimal places in the Amplify effect to 3.
We are even more inconsistent between the built-in generators and the Normalize effect in that the generators allow up to 6 decimal places. I assume that you agree that 6 decimal places is overkill for the Normalize effect?

Yes.

But as I said in the other thread, for me it depends how related the effects are. Amplify and Normalize are related. If you want for some experiment to bring the peak of multiple tracks to some dB level with more than one decimal place of accuracy, you do not want to be using Amplify to do that.


Gale

[Gale Andrews]

Nyquist plug-ins do not allow input beyond the slider range

But they can, no? In 2.1.0, I can set the High Pass slider range for Hz to 10000 to 20000 Hz but still high pass at 1000 Hz by entering that value in the text box.

In 2.1.1 the validation prevents that.

If a fringe use case of Nyquist plugins is that you can customise them, might disallowing values outside the slider range be a bit of a backwards step?

Gale

[steve]

might disallowing values outside the slider range be a bit of a backwards step?

I see it as half a step back so that we can take a full step forward.

Audacity is the only audio editor that I know of that does not have any logarithmic slider controls. This is not because we "can't" have them, just that we "don't" have them.

Currently we "can't" have logarithmic sliders in Nyquist effects, because the Audacity / Nyquist GUI only has linear sliders, but we could have them in the future, and for built-in effects they are already possible,
Logarithmic sliders would solve many of the problems because they allow a wide range and precision close to the end of the scale where it is most needed,

Consistency between different effects has already been mentioned, but difference between slider range and text input range within the same control is a much more obvious inconsistency. Better in my opinion that we improve our design so that "extended text input" is not required.

[steve]

Are there any other effects that need their range or precision updating to allow for the new validation?

[whbjr]

+1 for the accuracy settings in the original post. "Normalize" is fine as-is for my use, and I rarely-or-never use "Limiter." Thanks!

- Win

[Gale Andrews]

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

If we can specify "trailing zero", that is even less reason not to allow 2 decimal places in Normalize. Then if Amplify changes to 3 decimal places and Normalize has 2, we retain the one greater place of accuracy for Amplify that we have now.

The poster who wants 2 decimal places in Normalize must have a reason for asking, but has not articulated the reason. Clearly there is some potential value in it.


Gale