EarQuiz

How to Memorize Frequency Bands?

2023-05-31T00:00:00+00:00

Think of them as of reference points for your mind rather than direct guidance. They may be more or less helpful, depending on different factors: personal skills, experience, peculiarities of perception, kind of audio material, equalization settings, listening conditions, etc.

If you are a musician, and especially if you have a perfect pitch, trying to mentally connect center frequencies of band-pass filters with musical notes, closest to them, may be a good place to start. This works best with pink noise. The higher the Q factor and the more the frequency gain, the stronger the impression of a certain pitch.

Here is a screenshot of the 30-band in-app EQ with labels of musical notes that approximately match the filters’ center frequencies.

Musical instruments manufacturers and software developers often use different octave numerations. We use the Scientific pitch notation (SPN), also known as American standard pitch notation (ASPN) and international pitch notation (IPN) here. So, we consider that C4 is the middle C (~260 Hz), and the exact audio frequency of A4 is 440 Hz.

This method can be applied to some extent to music as well, especially in certain cases when frequency bands contain the fundamental frequencies of musical instruments. You might notice, how equalization changes loudness of different notes or relationship between fundamentals and harmonics/overtones. But for music, the approach would generally work as a positive side effect of timbre/pitch associations.

For pink noise exercises, try to find connotations between different spectral bands and certain “real” sounds, such as working engine, streamlet, waterfall, pouring sand, etc. For music exercises, think of resonating environments and objects which may have an effect, similar to particular equalization: big and small boxes, bottles, jars, cans, pots, cars, rooms, bathrooms, caves, tunnels, anything you have had experience with.

You can associate middle frequencies boost with vowel sounds:

Boosted frequency band, Hz	Vowel sound
250	‘Oo’ /u:/
500	‘O’ /ɔː/
1000	‘Ah’ /ɑ:/
2000	‘A’ /ɛ/
4000	‘Ee’ /i:/

And the highest frequencies can be thought of as sibilants:

Boosted frequency band, Hz	Sibilant sound
8000	‘Sh’ /sh/
16000	‘S’ /s/

When training with music, make (mental) notes of how altering different frequency bands changes timbres of instruments/voices. There are some tables with attempts to verbally describe different spectral ranges. However, using them may not be very helpful because the effect of particular equalization on timbre may drastically depend on a sound source, how it has been recorded, the playback conditions, the amount of frequency gain, and other settings. Moreover, the alteration of the same frequency band may be described as something that brings a positive or a negative quality under different conditions. E.g., fat and groovy vs boomy (low range boost), full and warm vs muddy (low-mid range boost), clear vs thin (low-mid range cut), round, full and rich vs boxy and honky (mid range boost), bright vs harsh (high-mid and high range boost), smooth vs dull (high-mid and high range cut) and so on. Cutting one frequency band may have an effect similar to boosting another frequency band. So, instead of relying on someone’s interpretations, try to make your own descriptions, related to certain sound sources, settings and conditions.

Recognizing cut frequencies is a bit harder than boosted ones. When the EQ is on, listen to the frequencies left trying to figure out what is missing. When the EQ returns to its off state, think of it as of frequencies’ boost.

What about synesthesia? The most primitive and obvious case of it, which you deal with here by training: you hear frequencies, and see different activated sliders, each having its position in space, plus highlighted numbers. But can you see/imagine frequency bands as different colors/textures/material types? Can you feel them? Can you smell them? Can you taste them? Try to note the slightest and subtlest body and mind reactions when listening to equalization examples, and focus on them for a moment.

Try different approaches. Find out, what works better for you in different situations. Stay as relaxed as possible while training. Take pauses. Do not torture your ears and brains too much. Train regularly, not long: 2—4 exercises per session may be enough. Try to build your personal timbre/frequency vocabulary through daily practice!

And, last but not least. Even the small recurring donation to the project not only helps to support and develop the software, but it is a great source of additional motivation for you to keep practicing!

< Back to the Table of Contents

Audio Processing Settings

2023-05-19T00:00:00+00:00

In the version 0.1.7, the possibility to adjust the audio processing parameters of training examples has been introduced. This can be done from the dialog window which is called by selecting Audio | Audio Processing Settings in the main menu.

The most important feature here, placed within the Equalization Performance Settings box, is the possibility to change the time proportion of EQ Off vs. EQ On phases. It used to be hard-coded to 30%/40%/30% (EQ Off/EQ On/EQ Off correspondingly). And now a user can increase the length of the part where the equalizer is switched on by dragging the EQ On Time slider to the right, up to 100%.

EQ Always On In Test Mode is another feature that has been added to the Audio menu in the main menu, and it is closely related to the above-mentioned one. This makes sense when doing test exercises with either pink noise or professionally mixed and mastered music tracks with polished spectral balance. When it is checked, the whole game becomes trickier, but it is closer to real-world situations. And the goal becomes not only to recognize the frequency bands that are boosted or cut, but also to develop the ability of detecting the issues that should be fixed, without reference to normal (non-processed) sound.

When EQ Always On In Test Mode is checked, EQ On Time parameter inside the Audio Processing Settings window makes no effect on test exercises.

The Audio Processing Settings window also contains a couple of more parameters.

The EQ On↔︎Off Transition Time parameter, which also refers to the Equalization Performance Settings group, is literally the length of the cross-fade between equalized and non-equalized parts of an example, which prevents possible clicks at the transition places. A user can set a value between 1 ms and 50 ms, with the default one being 35 ms. The less, the faster; the bigger, the smoother.

And the Example Fade In/Out Duration setting has the default value of 5 ms. You might want to set this to zero and do without these fades, e.g., if you decide to make learning files package, which you are going to play back continuously, without pauses between examples within the sequence using your favorite media player. This is a way to do the learning exercises while enjoying the seamless, non-interrupted playback of a music track. Or a user might want to set this value to the maximum (100 ms) to get the smoothest possible starts and endings.

All the adjustable values in this window used to be hard-coded in the previous versions. Now that they have become the default settings, a user can always bring them back by pressing the Reset button.

In order for the adjustments inside the Audio Processing Settings window to take effect, a user should press the Apply button. If a training example is being played back, it will be stopped and reloaded. Checking/unchecking the EQ Always On In Test Mode option has the same behavior, but it is applied immediately when in Test mode.

Adjusting any of the above-mentioned settings not only affects the training examples inside the program, but also changes these parameters for the exported learning/test audio packages correspondingly.

All these values are stored between application launch sessions.

Getting Started - Choosing Audio Playback Backend

2023-05-18T00:00:00+00:00

This option was deprecated in version 0.1.7. The bug in Qt6/PyQt6 library that caused deadlock when using FFmpeg backend and led to the application hanging was fixed. Now, FFmpeg is used as a single possible framework for playing all the supported audio formats on all the supported platforms.

Version 0.1.6

This option was introduced in version 0.1.6 along with the support of FFmpeg, which has now been used by default. Before that, the native system backend (Windows Media Foundation on Windows, and AVFoundation on macOS) had been the only option. The Linux version, which also appeared in this release, uses GStreamer as the underlying framework. To switch between FFmpeg and the native audio playback backend, just select the corresponding option in Audio | Audio Playback Backend menu of the main menu. The program restart is required for the changes to take effect, so the dialog window, asking for your permission to do that, will appear.

The main advantage of FFmpeg is that it is cross-platform and overcomes some limitations of native multimedia backends. It lets the program play audio files of all the audio formats which can be added to the Playlist without the need to use the in-app conversion tool. The playback position timing also tends to be more exact, especially comparing to AVFoundation handling of FLAC files. On Ubuntu (Linux), FFmpeg showed more stable performance than GStreamer when testing the application manually.

However, the app’s GUI (including the audio playback part) is built upon the PyQt6 framework, with all its strengths and pitfalls. Since QT introduced the FFmpeg support as the technology preview, not all the stability issues seem to be fixed yet. While the manual tests showed the stable performance on Windows and Linux, on macOS, it sometimes leads to the program hanging or freezing when a new audio file is being loaded. So, when the crashless performance is the priority, switching to the native backend on macOS is recommended.

< Back to the Table of Contents

Annotations

2023-05-17T00:00:00+00:00

Pink Noise

Pink noise is a kind of random noise which has equal energy in different octave ranges. It derives from white noise which is a random signal with wide frequency range and equal power for each frequency. Since the number of frequencies in octave doubles for each octave increase, the energy of white noise doubles in each higher octave as well, which is 3 dB/oct. boost. In other words, white noise gives the impression of high-frequencies dominance. In pink noise, the energy falls proportionally to frequency. I.e., it is similar to white noise, but with 3 dB/octave attenuation, starting from 1 Hz. This gives a much more even result from a human ear perspective, and the spectrum of pink noise is the closest one to an average music signal.

< Back to “The Basic Training Method”

Peak Normalization

Peak normalization of an audio file or its part is a linear change of its whole gain that makes the amplitude level of its (PCM) sample with the maximum amplitude equal to a certain value (in dB).

< Back to “Checking Audio Playback”
< Back to “Setting Volume Level”

Supported Audio Formats

Starting from version 0.1.7, the application should flawlessly work with WAVE, AIFF, MP3, FLAC, and OGG files on all the supported platforms.

< Back to “The Basic Training Method”
< Back to “Working with External Audio Files”

< Back to the Table of Contents

Application Data and Settings

2023-05-16T00:00:00+00:00

I tried to make the program store all its data and settings in one place, without leaving a lot of scattered garbage on disk or in registry (on Windows). On macOS, the (parent) data/settings directory is Library/Application Support/EarQuiz/Frequencies under current system’s user folder. On Windows, it is located in EarQuiz/Frequencies folder under the current user’s Application Data folder. And the path on Linux is .config/EarQuiz/Frequencies in a user’s home directory. Therefore, by manually removing it or some parts of it, you can reset the corresponding settings to defaults. Moreover, automatic clean-up of all the application data during the uninstallation process (which is often not desirable) becomes unnecessary.

The config.ini file contains the program settings and states, including checked/unchecked options, windows’/panels’ sizes, placements and visibilities, last used audio source and EQ pattern, locked EQ settings, etc.

The audio tracks’ paths of the current Playlist are stored in current.m3u8 file inside the Playlists sub-folder.

The trimmed ranges (starting and ending points) and the slice length values of audio files are stored inside the SourceRangeLib sub-folder. Each loaded audio file is hashed, and the data about each audio file is saved as a separate JSON format file with the name of md5 hash-sum of the corresponding audio file and the *.afab (Audio File A-B) extension. You can also see the audio file name inside each of these JSON data files. Hashing big files is not a very fast process, but this method allows the program to identify files in a path-independent and platform-independent way. That said, you can even transfer these *.afab files to a similar folder on another computer, and the copy of the application, running on that machine, would be able to apply these settings/data to the identical copies of your audio files.

< Back to the Table of Contents

Setting Default Length of Slice/Example

2023-05-15T00:00:00+00:00

Initially, the default length of a pink noise example is 10 sec. To set another default value:

Switch to Pink Noise in Audio Source window (if it is not selected).
Open Transport Panel (if it is hidden).
Set the target Slice Length value.
Click on the star at the right of the Slice Length value.

The default length of an external audio file slice is 12 sec. To set another default value:

Switch to Audio File (Playlist) in Audio Source window (if it is not selected).
Open Transport Panel (if it is hidden).
Set the target Slice Length value.
Click on the star at the right of the Slice Length value.

However, if the slice length has already been set before for the particular loaded audio file, it will override the default value.

< Back to the Table of Contents

Converting Audio Files

2023-05-14T00:00:00+00:00

The purpose of the in-app conversion tool is to convert audio files between different lossless formats, or to resample audio files. A user might need to downsample audio files with very high resolution (more than 48 kHz) to get sampling rates that are supported by his or her audio system, or just to reduce their size.

To convert audio file(s), first, select it/them. Second, choose the File | Convert Selected Files… item from the main menu or the same option from the right-click context menu of the Playlist.

In the opened dialog menu, you can choose the output format and the action to be done with sampling rate. The available formats are WAV, AIFF and FLAC (since version 0.1.8).

The sampling rate options are the following:

Same as original. The source file(s) are not resampled.
44100 Hz. The sampling rate of the output file is set to 44.1 kHz, regardless of the source sampling rate.
48000 Hz. The sampling rate of the output file is set to 48 kHz, regardless of the source sampling rate.
Auto choose 44100 Hz or 48000 Hz (downsample to multiple where original samplerate is higher). If the file sampling rate is 44.1 kHz or the multiple of 44.1 kHz (88.2 kHz, 176.4 kHz, etc.), the output samplerate is set to 44.1 kHz. If the file sampling rate is 48 kHz or the multiple of 48 kHz (96 kHz, 192 kHz, etc.), the output samplerate is set to 48 kHz.

When the original sampling rate of a file is changed, it is labeled as Resampled in the output filename.

After checking out or adjusting all the settings, press the Convert button to proceed. When done, the new file(s) would be added to the Playlist after its/their source file(s).

< Back to the Table of Contents

Making Training Audio Files

2023-05-13T00:00:00+00:00

Learning and test exercises can be made in the form of audio file collections. The possible export formats are WAV, AIFF, MP3, FLAC, and OGG. So, you can use them either for self-studying or teaching even without this application, just with any playback device or software which supports these common audio formats.

To make learning or test exercise audio file collection, select File | Make Learning Files… or File | Make Test Files… from the main menu, correspondingly. Both options call the same dialog window, but with different Type setting selected, which can be changed afterward.

For the exercise files, the current audio source, EQ pattern and EQ settings will be used. The default parent output directory (Exercise Folder) is Documents/EarQuiz Frequencies/Exercises in the current system user folder, but the dialog has an option which lets you change it.

By default, for each new exercise a new sub-folder with Exercise Name is created, and this name is used as the prefix of each name of exported files, but you can switch off these options by unchecking the Use as Folder Name or Use as Prefix of Filenames settings. A default exercise name is generated automatically on each dialog window call, starting with Exercise1 and increasing the ending number (Exercise2, Exercise3, etc.) as needed so that the name would not repeat the name of an existing folder.

The number of learning audio files equals to the number of possible equalization options/combinations within a current EQ pattern. The applied equalization (center frequency, boosted or cut) is indicated in the learning audio files’ names. The order of EQ bands is determined with the current settings of the Controls | EQ Bands Order in Learn Mode menu of the main menu. The examples are enumerated by default, but you can deactivate this option by unchecking the Enumerate Learning Examples option.

The number of test audio files in an exercise is always 10, and the examples are always enumerated, but you cannot get the answers from the filenames.

There is also a text file added to audio files. In case with learning exercises, …Info.txt includes the common information (audio source information, EQ pattern, EQ settings, peak normalization level, etc.) The text file …Answers.txt of a test exercise collection in addition provides the right answers.

The sampling rate of exercise audio files is the same as the sampling rate of the audio source they are made from. The default bitrate of lossy formats (MP3 and OGG) is 320 kbps, but you can choose another one from the corresponding box.

After checking out or adjusting all the settings, press the Make button to proceed.

< Back to the Table of Contents

Creating, Editing and Saving Playlists

2023-05-12T00:00:00+00:00

Basically, within the software, the single playlist (and the corresponding Playlist widget inside the Audio Source window) is used. You can add the tracks you are going to use as audio sources for your exercises to it, reorder them and remove them. The Playlist widget is a table with 3 columns: Filename, Duration and Folder Path. The playlist contents are saved and restored on each program startup unless cancelled by user. You can also export them to separate playlist files, which is highly recommended every time you make a collection of tracks that would be reused in the future.

Whether the application reloads the internal playlist or imports data from external playlist files, only the existing audio files of the supported formats are added as items.

Adding and Removing Tracks

Different ways can be used to add audio tracks into the application Playlist:

Select File | Open Files… from the main menu, or press Ctrl+O (on Windows and Linux) or ⌘O (on macOS), or click on the + button in the Audio Source window. Then choose audio files (WAV, AIFF, FLAC, MP3 or OGG) or playlist files (M3U, M3U8, PLS or XSPF) in the file dialog and click the Open button.
Select File | Open Folder… from the main menu. Then choose the folder containing audio files and click the Open button. In this case, the program will not only add audio files of the supported audio formats within the selected folder, but it will also search for audio files in all the sub-folders. The playlist files of the supported formats within the folder and the sub-folders will be found and parsed as well.
Drag & Drop files/folders to the Playlist from anywhere possible. Just like in the previous point, the program will also search for audio files and parse the playlist files in all the dragged & dropped folders and their sub-folders.
If the application is not opened yet, you can alternatively select audio files in File Explorer (on Windows) or Finder (on macOS) or File Manager (on Linux) and use the Open with option in the right-click context menu to open them with EarQuiz Frequencies.

The audio items of valid (10 sec. and more) duration and sampling rate (44.1 kHz and more), which can (potentially) be opened, have normal (black) font color. If an error occurred while trying to read a file or load it to the player, or its duration is less than 10 sec., or its sampling rate is lower than 44.1 kHz, it would be colored with gray. If a file, having already been added to the Playlist, cannot be found anymore, it becomes red.

You can change the items’ order with the internal drag & drop of the Playlist.

To show any of the tracks from the Playlist in File Explorer (on Windows) or Finder (on macOS) or File Manager (on Linux), select and right-click on it (or select it with right-clicking) and choose the corresponding option in the context menu.

Clicking the Clear button above the Playlist removes all the items from it. To remove certain tracks, select them and click on the - button, or choose the Remove Selected option from the right-click context menu, or press the Backspace or the Delete key. There is also an option to remove the unavailable (gray and red) tracks, selecting Remove Unavailable from the right-click context menu.

Exporting Playlists

To export your Playlist contents to a separate file, select one of the two items in the File | Export Playlist… menu if the main menu. But which one to choose?

With Export with Absolute Paths… option, all the saved paths will be absolute. This is suitable when you are going to open the audio files from the same machine, without changing their location or renaming drives, parent folders, etc. The playlist file itself can be stored locally anywhere.

However, if you want to make a movable library of audio sources and organize it with playlists, you should have a separate parent folder for it, choose Export Using Relative Paths for Subfolders… option, and select this folder as your playlist file location in the file dialog. In this case, the paths to the audio files in the same folder as the playlist file or below, would be stored as relative ones to the playlist file path. The other audio files’ paths would be absolute.

The file dialog, which is opened on choosing any of the two options above, lets you name your playlist file, choose its future location, and select its format/extension. Currently, you can save your playlist as M3U or M3U8 file. These are very human-readable text file formats which may contain only audio file paths/URLs at different lines. The only difference between them is that M3U8 forces the use of UTF-8 encoding. On macOS and Linux, you can use them interchangeably, but on Windows the system encoding, which may be different, is used for M3U files, which may cause problems with non-Latin characters. So, M3U8 is more recommended in general case.

< Back to the Table of Contents

Getting Started - The Test Mode

2023-05-10T00:00:00+00:00

The Test mode is activated by clicking the corresponding button on the main window or by selecting Controls | Test Mode from the main menu. The keyboard shortcut for it is Ctrl+T (on Windows and Linux) or ⌘T (on macOS).

The application will normalize the audio if needed. When you use an external audio file as a source, the program will read and crop it before that. Then the playback of the first example will start automatically.

To complete a test, you should do ten examples in a row. Use sliders’ handles to input the guessed answers. This can be done during or after the playback of each example. Once your answer is accepted, the right answer will be labelled and highlighted in green immediately. The Next Example control/button will become enabled so that you could proceed. Unlike in the Learn mode, the sliders with changed handles’ positions (the user input) may differ from the highlighted ones (the right answer).

You can play and stop a current example as many times as you want with the Play and the Stop buttons/controls correspondingly. Any time you do not give the exact right answer, it is highly recommended to play the example once again for better training effectiveness.

When the mode is activated, the Exercise / Score Information window is shown. It contains useful information about a current test: examples count, test status (in progress, finished with pass/fail or canceled), the user and the right answers, and scoring.

The score calculation is quite straightforward. The maximum score for an example is 10, and the maximum total score for a test is 100. When a user’s answer is equal to the right answer, he or she gets 10 out of 10. Each 1-octave error subtracts 1 score point. Each 1/3-octave error subtracts 0.33 points. The boost/cut mistake takes off 2 points. In case of dual-band exercise, the scores for each frequency band in an example are count separately, with 5 being the maximum and 0 being the minimum for a frequency band. And then a maximum possible sum is taken as the score for the example.

The following table shows the expected score range for different types of exercises (EQ patterns):

Exercise / EQ Pattern Type	Expected Score
1-octave bands, single-band	85—95
1/3-octave bands, single-band	75—90
1-octave bands, dual-band	70—85
1/3-octave bands, dual-band	65—80

To pass the test, you must get a total score which is 5 points below the lower end of the expected score range or higher. If your score is higher than the upper end of the range, you get the passed+ mark.

Changing EQ settings is disabled in this mode. Loading another audio source or switching to another mode would cancel the current test. Selecting another EQ Pattern would reset it as well.

< Back to the Table of Contents