Gathering requirements for new DSP plugin - please be realistic!

[latimer]

I am running LMS 8.3 on Linux with a squeezebox duet as player.
My hearing has degraded over the past years, and I was looking for an EQ to boost the higher frequencies.
I did try the Squeezebox EQ plugin from Jonathon Fox which is listed on the plugin page, but it did not work at all on my mp3 files, and it completely broke the playback of ogg files.
Thus, I am looking forward to your project.

[foxesden]

Hi - just to be clear I update the Squeezebox EQ plugin too.

If some sources are working and others are not, it is usually an issue with the custom.conf mapping that I have setup, or the file type mappings. The former is a code fix, the latter requires a bit of settings amendment. If you want this fixed within the existing Squeezebox EQ (Inguz) please feel free to raise this issue within that thread, and I will do my best to help. If you are happy to wait for my new updated plugin, I still have a few questions, which will allow me to work on incorporating a fix.

If you could give me a little information regarding the problems you had with ogg and mp3 files. You say it did not work on mp3 files and broke the playback of ogg files. From that I deduce that you got no sound at all on ogg files and reduced functionality vs mp3. Is that correct? If so can you explain what the reduced functionality was?

If I an wrong about that please let me know what was happening.

I also assume that you were playing these from your server rather than a remote stream such as spotify.

[foxesden]

Progress update.
I have a working build and I am now ready to start installation testing - this involves packaging all the components and checking that they correctly install and run on Linux and Windows environment.

I am planning to release this version and then look at an update down the track - as I need a break from it!

Features working and tested:

adjustable pre-amp (This is used to reduce gain to stop clipping - I currently have it set to -14.4dB)
Balance (i.e. if you move the slider left the right volume will decrease by a proportionate amount, same as the physical control)
Delay in ms (if you move the slider left the right channel will be delayed by a proportionate amount)
Loudness - simple off.on at the moment, I will probably make it tweakable in a later release.
Parametric Peaking Equaliser - 2 - 31 bands with adjustable frequency; gain and Q

Optional filters for
High pass
Low Shelf
High Shelf
Low Pass

Room correction via Wav Impulse files

The DSP will correctly adapt to the incoming Sample Rate and Bit Depth and will play gaplessly.

The user interface (UI) has been tweaked to enable the above. In addition I have enable slider reset by double-click (not working on touch devices yet)
The UI will work on a phone, but doesn't look too pretty. I will fix this later.

Other features planned but not included in the initial release are

import of settings from REW
clipping monitor - the info is in the log file.

[Zombie]

Wow! Thanks, looking forward to try out the new functions.

[HiPresto]

I am new to this plugin, but it seems as if it does nothing at all to the wav files I am playing. Is there an instruction manual? I just activated it from the LMS plugins tab. It is version 0.9.39.

Thanks!

[foxesden]

I am new to this plugin, but it seems as if it does nothing at all to the wav files I am playing. Is there an instruction manual? I just activated it from the LMS plugins tab. It is version 0.9.39.

Thanks!

Hi - I think the question that you are raising relates to the Squeeze EQ plugin based off InguzDSP. This thread is talking about its replacement. If you look at post 124 onwards on this thread - https://forums.slimdevices.com/forum...p-plugin/page9 similar questions have been asked. Essentially, you may need to restart the server a couple of times. You may need to change some file type mappings.

If you have any further issues please follow up on the linked thread. NB there was an instruction manual, it needs a rewrite, I am replacing the plugin so any re-rewrite will be vs the new plugin.

Hope this helps

[HiPresto]

Thanks. This is the first time on this forum, and my first install of the plugin. I will research your links.

[foxesden]

Welcome, if you have a problem, just ask. Most people here are friendly and just want to help.

[foxesden]

Further progress update. I was offered some feedback on the User Interface - and it wasn't that positive . So I have given it an overhaul with the basis of making it as simple as possible and being easy to navigate.
DSP can be disabled, which would allow you to use it for some players and not others,
Frequency sliders operate on a logarithmic scale and snap to "sensible" numbers e.g. it is easy to dial in 11,000 rather than 11,023.16 - this was tricky.
Parametric EQ Bands are added and removed via a plus and minus button. With a max number of bands being 31
I have added some context sensitive help, which can be accessed by mouse hover or screen touch.
I have also spent a long time tuning the gain settings to minimise clipping on various sources, that was not fun.

This week I will be packaging and testing the install. Hope to be released soon

Some screenshots - using Material Skin.


PEQ bands can be added and removed altogether.
And with all features enabled

[boomboxx]

Awesome work! Cant wait to install.

[papaiannis]

Wow! i'm glad your work is so far ahead. can't wait to try this plugin

[foxesden]

I had some serious issues testing installs on Linux last week, namely that I couldn't get CamillaDSP to run without installing an additional library, and CamillaDSP would not shut-down as expected (and as it does on Windows) for example if you skip to the next track. This is not a fault with CamillaDSP, just that I was using it an app rather than a system process. All this meant that I have replaced CamillaDSP with good-ole SoX.

Anyway - I am ready for some beta-testing now, if anyone is interested.

I have tested the install on Windows, Linux-Arm and Linux-x64. The install is now clean and the application runs as intended, I could get it to run on a Pi-3B but the audio stream rapidly became mangled because the load is too much for it's little processor. I have it running fine on an elderly Celeron mini PC (WIndows 10) although I think it would struggle with more than 2 players at once (untested).

As this is a test version if you want to try it you will need to enter the following into the additional repositories section of plugins

https://raw.githubusercontent.com/Foxenfurter/SqueezeDSP/main/test.xml

Once this entered and apply is pressed you should get the plugin SqueezeDSP appear just above the repository name. Select the checkbox and hit apply again. After a few seconds (depending on your internet speed) the plugin should download and you will be asked to restart your server.

If you are using SqueezeEQ I recommend that you uninstall the Plugin - It can be re-installed without you losing settings if you want to revert.

At this point I suggest you do the following. Stop and restart any player, and then use the extras-menu to Select SqueezeDSP.
This will create a config file for the player with default settings - however you need to map the file types to enable it.
Next, stop and restart the LMS again
Now navigate to Settings->Advanced-FileTypes. For most types you will need to enable the entries with SqueezeDSP in them and disable the others. For Spotty it is a bit more complex, you need to enable entries for Spotty-Sox and for SqueezeDSP - flac. I suggest take it a step at a time and try switching one source over first.

Tips
One limitation to be aware of is that settings only take effect when a track is started.
To get an effect that is easy to hear - set highpass to a frequency of 1000 Hz and Low pass to a frequency of 3000Hz. This should give you a nasty transistor radio sound.

[grimmace92]

I had some serious issues testing installs on Linux last week, namely that I couldn't get CamillaDSP to run without installing an additional library, and CamillaDSP would not shut-down as expected (and as it does on Windows) for example if you skip to the next track. This is not a fault with CamillaDSP, just that I was using it an app rather than a system process. All this meant that I have replaced CamillaDSP with good-ole SoX.

Anyway - I am ready for some beta-testing now, if anyone is interested.

I have tested the install on Windows, Linux-Arm and Linux-x64. The install is now clean and the application runs as intended, I could get it to run on a Pi-3B but the audio stream rapidly became mangled because the load is too much for it's little processor. I have it running fine on an elderly Celeron mini PC (WIndows 10) although I think it would struggle with more than 2 players at once (untested).

As this is a test version if you want to try it you will need to enter the following into the additional repositories section of plugins

https://raw.githubusercontent.com/Foxenfurter/SqueezeDSP/main/test.xml

Once this entered and apply is pressed you should get the plugin SqueezeDSP appear just above the repository name. Select the checkbox and hit apply again. After a few seconds (depending on your internet speed) the plugin should download and you will be asked to restart your server.

If you are using SqueezeEQ I recommend that you uninstall the Plugin - It can be re-installed without you losing settings if you want to revert.

At this point I suggest you do the following. Stop and restart any player, and then use the extras-menu to Select SqueezeDSP.
This will create a config file for the player with default settings - however you need to map the file types to enable it.
Next, stop and restart the LMS again
Now navigate to Settings->Advanced-FileTypes. For most types you will need to enable the entries with SqueezeDSP in them and disable the others. For Spotty it is a bit more complex, you need to enable entries for Spotty-Sox and for SqueezeDSP - flac. I suggest take it a step at a time and try switching one source over first.

Tips
One limitation to be aware of is that settings only take effect when a track is started.
To get an effect that is easy to hear - set highpass to a frequency of 1000 Hz and Low pass to a frequency of 3000Hz. This should give you a nasty transistor radio sound.

For the life of me couldn't get Inguz to work. This works right out of the box first try. Very cool!

Can somone provide a quick way to get started with the PEQ bands? How many bands and what frequencies are best to try for a beginner to test out?

[foxesden]

Very simple question, almost impossible to answer.

Why would you want to apply EQ in the first place?
You listen to headphones and they don't sound too great in certain frequencies
You listen to speakers and the speaker/room interaction is giving you boomy bass or other serious anomilies
You generally like your speakers / headphones but want to tweak for taste.

There are a number of articles which explain in more detail than I could the mechanics (and even the mathematics) of how they work. Here is a reasonably friendly one

https://github.com/jaakkopasanen/AutoEq

In order to see what it can do, just try the "transistor test" I outline above, this was suggested by another member as a good way of seeing if the installation is working and gives you a feel for what you can do.

It is much easier to start with headphone correction, than speaker correction - because measurements of most headphones have already been done and converted to parametric EQ records.
See https://github.com/jaakkopasanen/Aut...aster/results/
Under the headings Oratory, Crinacle etc you will see lists of headphones, and behind that you will see the parametric EQ settings to use. My intention is to write an importer for these settings but it is simple enough to do manually.
The PEQ settings for Headphones will not normally correct to a flat response, as research has indicated that us humans have a preference for more bass and the most popular target is known as the "Harmon Curve", following research at Harmon. Interestingly younger males tend to prefer even more bass (who'd have thunk it!) and older females prefer less bass, but this is a generalisation based on listener response so will vary from person to person.
So having selected and applied an peq curve to your headphones you may find you want to tweak to taste.

In general terms it is generally considered more effective to remove peaks than boost troughs. If you do decide to boost any signals above 0 then you should be aware of clipping (i.e. if you play a sine wave the peaks will be clipped) this results in unpleasant distortion, to counteract this, you will need to reduce the gain using the pre-amp.
The maths is quite complex in this space, so a simple rule such as for any boost of 1 db reduce preamp by 1 db may not work if you have multiple frequencies boosted that are close together

For corrections to speakers, you normally must include the room in the equation hence measuring the speakers in the room is normally advisable. I have some links I can post if you are interested, but it can be quite time-consuming to do, usually involves some additional expense and the sounds used for measurement can be very annoying to others!

[boomboxx]

Have it working, however, I cannot get it working with mp3 files. Is this currently only working with flac? Am I missing something? There are no options for mp3 other than native for mp3. Also, I could not get this working without saving a preset and then selecting the preset. Looks like it kept turning off if I did not use and have a preset saved. The other issue is that I cannot scroll forward in the song when using FLAC files. It just skips to the next song if I try to skip forward in the track. Actually, it looks like everytime I go back to the plugin it defaults back to no preset selected. It will keep working anyways though, just have to select the preset everytime you go into the dsp settings. Will continue to mess with it and let you know if I see anything else. Hopefully I can get mp3s working, just not sure why its not currently. Great work btw, seems to be working great otherwise!

Is there anyway to manually input PEQ settings? For instance, I noticed that Freq is in intervals of 5. I would prefer to be able to input values all the way down to a few decimal places. Even gain and Q would be nice to manually input specific amounts to a few decimal places.