It is possible to make a generally accepted data format “Frequency tab Gain” in the Filter Curve plugin. This format:
0.0 0
1000 8
20000.0 0.0
Previously, there was an xml format, now txt of incomprehensible formatting. Whereas in all programs of frequency correction, import and export are in the format “Frequency tab Gain”.
If you want, at the beginning or end of the file make a line with additional parameters (although why are they needed?), but it is better to do the basic data about frequency and level through tabulation and not xml or f0 = “1000” v0 = “- 6” … this is extremely uncomfortable!
The “Frequency tab Level” format is used in programs: LoudSpeaker LAB, SpectraPLUS, HOLMImpulse … why complicate simple things from version to version at the expense of convenience is completely incomprehensible? That remains is to make a binary format, that finally no one can do neither import nor export freq data.
Hopefully in next version Filter Curve export/import format will be a compatible with other applications, and simple.
The format is simple. Using a format as you describe would require the effect to convert it to its own format to be able to use it, and that is added complexity that a developer would need to write. The exported settings files are not intended for manual editing by end users. End users are expected to use the graphic interface. Expert users should have no problem working out the format just by looking at the text.
The change was necessary for the effect to work correctly in Macros and with scripting.
You can still use the old XML format if you prefer. There is a tool shipped with Audacity to convert the old format to the new format (EQ XML to TXT Converter - Audacity Manual)
Is it possible to make a converter from the generally accepted txt format to the txt format of the latest version of Audacity (f0 = “1000” v0 = “-6”, etc.)? xml → in txt Audacity is still limiting, everything revolves around Audacity.
I used to be able to convert txt from other applications to xml Audacity (wrote a small script). Now, with the new txt Audacity format, import impossible again.
I wrote a script for the new format, it’s not easier, if only because f0 and v0 digits are increasing, added a counter, there are more letters in xml, but simpler. I’m not a programmer.
I found out another problem, the Filter Curve plugin imports from txt and displays no more than 200-300 values on the curve (~ 8 kB file), and in total, for example, I have 2000 frequency values and 2000 levels in the curve. Previously, XML could open and display as large a curve file as desired. Can make the Audacity Filter Curve plugin be able to open large files with equalization as before? I really need.
Other programs give such frequency response. Of course, they can be reduced to 200 or 10 dots, but what for? If before everything loaded without problems, but now it suddenly turned into an overkill? It is necessary to develop the program along the path of increasing functions and capabilities, and not cutting back, for some, half of the functions are overkill, let’s remove them?
Instead of adding an Excel-compatible format, you say that it is not convenient for programmers. I’m not a programmer, and I made a converter. Programmers are somehow uncomfortable with the Excel format txt of 2 columns with numbers? More than 200 points on the curve are also somewhat inconvenient and unnecessary for them, absurd?
Calm down StrongBond.
The 200 point limit is already logged on the Audacity issue tracker - I logged it at the beginning of February when I first encountered it. None of the developers have picked it up yet.
In my bug report I suggested that:
200 points does not seem very generous, perhaps that could be increased a bit (I suggest 512 points as a generous maximum number of points
You are now saying that you need thousands of points.
I can update the bug report, but I would need to justify why support for thousands of points is necessary. Can you help me to justify the need for thousands of points in the EQ filter?
Can you help me to justify the need for thousands of points in the EQ filter?
Why do need an equalizer - for equalization in great detail. For example, to load such a curve. It is desirable in the original.txt file format, because it is the most common and easily processed one in Excel and other programs. Audacity 2.3.2 and past loads this file into xml easily. Of course, the number of points there is still limited to 8192, but still loads the curve itself. We need to make sure that the latest version of Audacity can load large files with curves. It’s bad that the slider for selecting the number of EQ points was removed, it was always set to max, but still.
And in general "filterLength =“8191” (sometimes even this is not enough for me to experiment), but only 200 are available?
Updated the converted.zip file converted.zip (54.3 KB)
One of the reasons that the slider was removed was because very few users understood to pros and cons of different window size settings, and consequently would frequently use sub-optimal settings. 8191 provides a good balance between frequency resolution and temporal resolution. Personally I’d like to see the “Size” setting supported by scripting / Macros so that “advanced” users have more control, without confusing general users.
I can see reasons why some users may want more than 200 points, but I still don’t know why you need more than 200 points.
To correct or adjust the tonal balance.
With high precision. Otherwise, there are other instruments such as a 2-32 band equalizer.
One of the reasons that the slider was removed was because very few users understood to pros and cons of different window size settings, and consequently would frequently use sub-optimal settings. 8191 provides a good balance between frequency resolution and temporal resolution. Personally I’d like to see the “Size” setting supported by scripting / Macros so that “advanced” users have more control, without confusing general users.
I cannot imagine a user who does not understand that 8191 is better than 10, especially since the expected curve is plotted on the graph. You have a bad opinion of users’ intelligence.
I can see reasons why some users may want more than 200 points, but I still don’t know why you need more than 200 points.
For the same reasons that other users may sometimes need more than 200 points. I gave examples, take the trouble to understand.
In the image below, the first track shows an impulse (no effects), the other two tracks have had EQ applied. The EQ settings are the same for the second two tracks except that the second track has a small “Size” and the third has a large “Size”. Notice that there is considerably more “ringing” in the third track. The ringing is an unavoidable consequence of the larger window size / filter length. To retain good transient response the filter length should be kept as short as possible, but for good frequency resolution the filter length needs to be long: Filter length (window size) is a compromise between frequency resolution and temporal resolution.
The same effect can be seen in the track spectrogram view.
The first track below has a window size of 128. Notice the discontinuity in the waveform is quite narrow in the spectrogram, but the steady tone is poorly defined.
The second track has a window size of 1024. Notice that the steady tone is fairly clear white line near the bottom of the spectrogram, but the discontinuity is poorly defined.
Notice that there is considerably more “ringing” in the third track. The ringing is an unavoidable consequence of the larger window size / filter length.
This is not more ringing, this the frequency resolution has increased. There is no bell on a good convolver.
To retain good transient response the filter length should be kept as short as possible, but for good frequency resolution the filter length needs to be long
Insufficient frequency resolution can be easily heard. Ringing - I have never heard any increase in THD or other abnormalities in the sound.
How is the flabby “transient response” expressed? I have never heard any deviations in sound with a window size of 8191 with a good convolver. If there is ringing or some other distortion, use another convolver.
Showing sound imperfections on a single impulse is like analyzing sound quality on a square impulse - visual but useless.
And then I wrote about fixing defects, not to argue about the size of the windows and the width of the frequency correction. Let the user decide what he wants 32, 8191 or 32768.
I realize I’m resurrecting an old thread here, but I just wanted to add my request for greater than 200 curve points to be supported in the Filter Curve EQ. I use Audacity for virtual instrument design, and I often generate advanced filter curves for morphing tonal characteristics between recorded velocity layers. These generated filter curves feature nearly 2,000 curve points, which works well with the old Equalizer plugin, but obviously fails with the replacement Filter Curve EQ. This means in order to complete my work, I cannot upgrade beyond Audacity 2.3.2 unless this limitation is fixed.
In the meantime, does anybody happen to know of any other filters or audio editors that are comparable to the old Equalizer plugin and allow import/export of curve points? Thanks in advance
mrbumpy409
The topic is not old, actual. Forget it, they won’t help you here. You will write 10 pages about simple things that needed, with zero result. Audacity develops archaic, good features kill.
Use version 2.3.2. For yourself modified it a little, became:
0.1s
44 31299 dot
48 31169 dot
96 29569 dot
176.4 26889 dot
192 26369 dot
0.2s
44 29829 dot
this gave more resolution at frequencies <500Hz, the lower the better.
Who do you mean by “they”? This is a community forum, run by members of the Audacity community, for the benefit of the Audacity community.
It appears that you are trying to offer some help to mrbumpy409, so that makes you part of the Audacity community, helping others.
200 points is not enough resolution to accurately shape an instrument sample with all its overtones and other noises (e.g., flute with natural overtones and breath noise). 2,000 points is sufficient resolution for my work, based on the testing I have done. There is an audible difference in the result between 200 points and 2,000 points. None so much beyond 2,000 points, and I haven’t tested to see exactly how much less than 2,000 points I can use without impacting the audible result.
