Search the Community

RAc makes a solid point, and this is part of why I don't want to use Phil's sound font. It has some harsh high frequencies that stand out to me when using headphones. Another consideration is that I need something that I can publish under an open source license and include with a commercial product. There's no explicit license on Phil's sound font, so I would need to get permission. I also want some extra samples. In addition to wanting to minimize the amount of pitch shifting that's necessary, I've had a request to provide (optional) button noise in my synth for extra authenticity. Thanks for the tip! I'm hoping the SKEDD connectors work well, because they don't require a part on the board, which helps reduce costs (each unique part number adds $3 per board it's used on, in addition to the cost of the parts themselves). If they don't pan out, I'll definitely consider micromatch next.

I'd like to throw in the fun spoiling remark that the main criterion for choosing a sound font in this case should not be what pleases the ear but what it ergonomically best - after all, the main use case will be attached headphones or in-ear plugs, so you may want to look into sound fonts with an overtone spectrum that does not tax hearing at extended usage too bad. Just my 0.02. About the connectors you mentioned earlier: I have switched to micromatch (eg Micro-MaTch 0,050-Steckverbinder-Baureihe - TE | Mouser) and found them to be the best solution for this connection requirement.

Why not stay with Phil’s sound font (as modified by Don)? It’s a perfectly good classic Wheatstone (baritone Aeola, IIRC).

Notice the copyright - 2006 - so nearly 20 years ago. Truth is that this was a spoof, although quite a few folks thought it was genuine at the time. It was produced using Steinberg Cubase, various MIDI arrangements of well known pieces and a sound font created from my C/G Lachenal anglo, although 'Enrico' was me playing anglo. Some unreleased items are still on my hard disc so if there is any interest I will post a few ☺️

Midi players play sampled sounds stored inside sound font files. There does not have to be a sample for every note because Polyphone (see below) will interpolate between samples to give you any missing notes. You could create your own custom instrument, perhaps based upon sound samples from your own concertina. I used the Polyphone sound font editor to tweak Phil Taylor's original sound font, but you could sample your own sounds and create a new sound font. Polyphone is not difficult to use, getting a clean set of samples is more work than editting them into a sound font. Polyphone is a free, open-source program. Phil's sound font is based on very few samples and was especially short of samples for the higher notes and yet it still sounds pretty good to me.

well, yes, any synthesizer can produce almost any sound, given preprovisioned sound fonts. The remaining question, however, is: How to switch between sounds? I also use an RPi (running fluidsynth with the same font that Steve uses) as the synthesizer on my MIDI crane, and it would be max an hour's work to add, say, a dozen other sound fonts to choose from. Yet the interesting question is how to select the font. One would have to come up with some kind of "Midi control surface" (thx Paul) to switch between fonts - eg on my Crane I have a few buttons without a sound mapped to it, so it would be an obvious choice to use those - for example one short press concertina, two violin and so on. Or use any combination of three buttons after a "mode select" button to select one of eight instruments. Or anything else. Then program that scheme into your button processor (in my case an Arduino) to generate the midi select instrument commands from it. So theoretically almost everything can be done, but it needs to be implemented. For my project, I won't bother because I use it for practicing, not music making, so one font (that as a bonus even sounds like a concertina) will be doing perfectly ok.

I'd start with the the sound font in Don Taylor's signature. You could certainly do a lot worse. I'm assuming you'll use a separate MIDI channel for each hand in order to enable your stereo setup? How loud are those speakers? I'll be quite impressed if they can be heard at a session. What microcontroller is your current iteration built around?

Thanks! The sound is produced with the sound font linked in Don Taylor's signature on this forum.

A recent post on the Session prompted me to re-investigate the use of sound fonts in EasyABC and I have found a way to make them work so now EasyABC can play the concertina (or any other instrument for which you have a sound font - there are lots out there). I have only tried this on Windows and I do not have a Mac so I cannot help with that system. I imagine that there is an analogous process to what follows and if a Mac user can make it work then perhaps they can tell us about it. Anyway, this is what you need to do: 1) Uninstall your existing version of EasyABC and install version1.3.8.6 from Sourceforge - the big green button: Check that it works OK with an abc file and that it plays back with its usual cheap piano sound. 2) Download the Windows 10 version of Fluidsynth 2.2.8 from Github: https://github.com/FluidSynth/fluidsynth/releases/download/v2.2.8/fluidsynth-2.2.8-win10-x64.zip Extract this zip file somewhere temporary as we are only going to need a single file from it, we are not going to install all of Fluidsynth. 3) There is a bin folder inside this zip file, inside this bin folder is the file libfluidsynth-3.dll, copy this file into the same folder that contains your easy_abc.exe file. This is what it looks like on my computer: 4) In that same folder, create a new sub-folder called soundfonts as shown above. Put a copy of the concertina sound font inside this folder. You can get Phil Taylor's concertina sound font from here. 5) In EasyABC go to Settings->ABC Settings... and select the File Settings tab. In the SoundFont entry, browse to the soundfonts folder you just created and select the sound font that you put in there. This is what my installation looks like: 6. Exit EasyABC and restart it. It should now use the sound font you just installed. Issues: When Phil Taylor created his sound font he assigned it to midi instrument 1 (Grand Piano) and there are no other instruments in his sound font so selecting any other midi instrument will still only play the concertina. This means that the other instrument selections, chord playback and bass chord playback, in the ABC2midi tab are not effective - everything is played on the concertina. If this becomes a real problem then let me know and I will investigate adding more instruments to the sound font.

Yes, I think so. This may work in 1.3.7.7 and you cannot do any harm in trying it. I don't see any way in EasyABC to give Fluidsynth a list of sound font files. You could try setting your vlc.exe file as the midi player in File Settings in EasyABC. If you have the concertina sound font installed in VLC (and not in EasyABC, leave the field blank) then this should work without you having to invoke VLC separately. I don't think it will be as fast as using Fluidsynth inside Easy ABC, but it will save you several steps in your procedure. VLC uses Fluidsynth internally. What I think that I can do is to take a complete sound font file that has all of the instruments in it and replace instrument #1 (Grand Piano) with the concertina. For piano players there are several other piano instruments in any complete sound font file but there is not a slot for a concertina so we have to 'steal' another instrument's slot and #1 is the obvious one to use as it is the default slot used by EasyABC, Musescore, etc... You can see the full list of midi instruments here: https://soundprogramming.net/file-formats/general-midi-instrument-list/ I will look at doing this later today, but I will also try to contact Seymour, the maintainer of EasyABC, and suggest that he add the capability of having multiple sound font files in EasyABC. How this works is that Fluidsynth looks for instruments using the sound font files in priority order so if the concertina sound font is top priority then it will find that when asked to play instrument #1, any other instruments (#2-#128) will use a lower priority sound font file. This is the way that Musescore uses Fluidsynth to play sound fonts.

I have added a new sound font to my download folder: "Combined concertina and FluidR3 GM.sf2" This is what I say about this in the readme.txt file: 15-Aug 2022 - added "Combined concertina and FluidR3 GM.sf2" Warning!!! This file is over 150KB, do not download it if you do not need it to use in EasyABC. At this time EasyABC only supports accessing a single sound font file. The original .sf2 files in this folder only contained one instrument - a concertina (actually they contain two other unused variants that I should delete). If you try to specify one of these sound fonts in EasyABC then it will play the concertina whatever other instrument you might have chosen. If this is a problem for you then you can download and use the "Combined concertina and FluidR3 GM.sf2" sound font. This is simply the "FluidR3 GM.sf2" sound font with the "Grand Piano" data replaced with the "Concertina-tuned-v2" data. If you select "Grand Piano" as your playback instrument (which is the default) then you will hear the concertina. All other instruments should sound properly. If you want a piano sound for playback then all of the other piano selections in the Abc settings->Abcmidi->Instrument for playback: list should work. EasyABC needs to have a Fluidsynth .dll added to its executable folder and the sound font location entered into the Abc settings->Abcmidi->Soundfont: box for this to work at all. The README for "FluidR3 GM.html" file is there for licensing reasons.

Yes, I'm using Jim van Donsel's current algorithm verbatim for C/G (with his permission, contacted him first) as well as his button labeling system. I tried replacing the "P" and "D" under the notes with arrow glyphs as I do for my B/C and C#/D box tab bellows direction indications, but they were much harder to read, so I stuck with his letter-based labeling. I also found that if I used arrow glyphs next to the button numbers, things got very crowded visually, so I just stuck with his "under the note" direction labelling. I can always easily revise the button and bellows direction labelling conventions in the future. I've built my own standalone batch pre-processor for large collection of tunes using his algorithm, with addition code derived from my own ABC tool to do the font and MIDI directives injection required for the play links. The end result is a final processed ABC file I can then bring into my ABC tool for final Interactive PDF generation. At this point, I don't intend to integrate the solution directly into the ABC Transcription Tools, but may in the future. I may make the batch pre-processor available as a standalone tool. if the loading of my concertina sound font is what is slow on your old system, you can follow the instructions (with explanation video) on the tune book page in the section called "Changing the Instruments Used for Playing Tunes from the Tunebooks", that describes how to change the default settings on the ABC tool to always override the MIDI instrument selection in the play links to an instrument sound font that is much smaller, like piano. If I do further optimizing of his algorithm, most likely around my own preferred C/G fingerings, I can regenerate all the PDF books in about an hour, thus the "highly experimental" label.

The same holds true for pitches which are hard coded by the target sound font. While a MIDI concertina would allow a key press action to, say, sound like a mouth harp, the ability to bend reeds (which I believe is essential to mouth harp players) can not be emulated by the "input side" of any keypads driven instrument. Of course, one can define MIDI extensions that open up the pitch range which, along with more refined sound fonts and elaborate "input measurement techniques," can approximate microtonal effects, but it is a moving target; apt players will always be able to conjure sound effects of an "analog" instruments that any range of sounds defineable by MIDI generated sounds will not be able to map. I would suggest to add these limitations to Paul's contra list. None of that contradicts anything on Paul's Pro list; the ability to practice with a head set would certainly open a wide range of opportunities for ambitious players. Yet practicing, say, 90% of your time on a MIDI instrument - no matter how technically sophisticated - will help quite some with dexterity and muscle memory, but very likely not in exploring the sound opportunities a "real" concertina can offer.

The SF2 sound font that has been in my sig for years now (see below) contains a set of sounds sampled from a baritone EC courtesy of Phil Taylor (no relation). If you want the raw .wav files then you could extract them with the free Polyphone sound font editor: https://www.polyphone-soundfonts.com/ However, not every note is sampled, far from it, as sound font player software interpolates for missing notes. You might be better off installing Musescore and then installing Phil's sound font in Musescore. You can then play each note in turn and capture the sound. Warning, some the really high notes are a bit iffy as the software occassionally interprets the note a fifth higher than than it should because the second (I think it is) overtone is really strong in the sound from a concertina.

If I am understanding your question correctly, that’s what a sound font IS (or at least one way to make a sound font). This one certainly is. Phil Taylor digitally recorded each note of his concertina and used those samples to create the font. Here is an example of MuseScore playing my (Hayden Duet concertina) arrangement of “South Australia” using Don’s tuned version of Phil’s Concertina Sound Font. It doesn’t quite sound like natural live playing (and I wouldn’t inflict it on my Morris Dancers), but it sounds more like a concertina than anything else the software might come up with. I could make it sound more natural by adding sliver rests for articulation, but that would make the notation more difficult for a human musician to read. [Aside: Just noticed Don and Phil, the (unrelated) Taylor “brothers” were also the names of the Everly Brothers.]

I have had a request to explain how to add a new sound font to Musescore 2.0. Maybe other folks want to know this too, so here goes. These instructions should work for PCs, Macs and for Linux - just adjust the folder paths accordingly. First, you need a new sound font file: Say the concertina.sound font that I posted earlier: https://www.dropbox.com/sh/kx6ude5uwzoan15/AAAy6ZJJMRRuwHVxgx16tfDZa?dl=0 You need to put the sound font file (the .sf2 or sf3 file) somewhere useful. It does not really matter where. I put mine in the soundfonts folder inside my Musescore 2.0 install folder: ...\MuseScorePortable\Data\soundfonts Now you need to tell Musescore where to find your new sound font file: Start Musescore 2.0 and goto Edit->Preferences. A multi-tab panel should open. You need the 'General' tab. There is a list of folder preferences, one of which is labelled 'SoundFonts:' with a little microphone button next to it. Click on the microphone button and a new panel should open with a list of places where Musescore will look to find sound font files. If the folder you used earlier is already in this list then you can just cancel out of all of these dialog boxes. If not, then you will need to "Add" your folder to the list. If you did add a new folder then I think it would be a good idea to close and restart Musescore just to be sure that the new folder preference is used. I don't know if this is really necessary, but it was needed in an earlier beta version of Musescore 2.0. Finally, you need to tell Musescore which sound font to use by default to play your scores: View->Synthesizer (check the box) will bring up another multi-tab dialog box. You need the 'Fluid' tab. This tab lists the sound font files that Musescore will actually use and they are listed in priority order. Sound font files may contain fonts for many instruments, each numbered from 1 -128. When a score calls for instrument 1, for example, Musescore will look first at the highest priority sound font file in this list. If this file contains a font for instrument 1 then that is what it will use. If not then it looks at the next sound font file in the list and so on. So you need to get your new sound font file in this list and then make sure that it is high priority so that it is chosen before any other sound font. Use the "Add" button and then the "Up" or "Down". When you are done you can just close the Synthesizer dialog box. If you installed the concertina.sf2 sound font file as the highest priority sound font then when you play a score that uses instrument 1 (which should really be a grand piano, but never mind) then you should hear the concertina. If you want to switch back to hearing a grand piano for instrument 1 then lower the priority of the Concertina.sf2 file in the Fluid list. There are other ways to change instruments, but this is the brute force way that works for me. If you do get into changing instruments through staff properties then just remember that as far as Musescore is concerned the concertina in concertina.sf2 is a grand piano. The midi standard did not define an instrument number for the concertina so whoever created this font decided that this would be the simplest thing to do. The only down side to this choice is that if you have a multi-part score and you want one voice to be a concertina and another to be a grand piano then you would have to settle for one of the other piano sounds defined in the midi GM instrument list: http://www.ccarh.org/courses/253/handout/gminstruments/

Works like a charm - except the sound does not work on my MS Surface tablet (Edge browser), regardless of whether I am online or not (the player starts and proceeds as expected, but there is no sound. The online version works flawlessly). I chose to use GitHubs downlad zip file option and unpacked it instead of cloning, is that a problem? Edit: Never mind, I found it! The Grand Piano sound font used as the melody default requires an internet connection. Apparently all sound fonts are accessed over net internet. Is there a way to specify the root for the sound fonts so I cab redirect the directory to a local one? Sorry for the noise! 😉

Thanks! Yes - the load cell/strain gauge is pretty well symmetric. I'm using a 3kg load cell. However... because the bandoneon technique is more about using gravity than push/pull, I'm only playing on the pull - because this fixed bellows system (it has a small amount of movement controlled through rubber mounting grommets). For a good bellows emulation, and bi-directional playing, something more like real bellows is needed, I think. I'm using a sampled bandoneon sound font by Jörg Bleymehl - it includes the starting transients, and then turns into a loopable clip. They're intended for score playback, so don't really account for the overlap region - where the octave around middle C is available on both sides, but sounds much more mellow on the left. I'm using some EQ processing to emulate that, but it's not ideal!

I have been experimenting with an ESP32 microcontroller sending midi messages over BLE (Bluetooth Low Energy) to an Android phone running the Fluidynth app loaded with Phil Taylor's concertina sound font. I can't confirm that the latency is as low as 10 ms but I think it is less than 25ms, I cannot sense any latency, "Traditional' midi audio has a very large latency because it has to compress and decompress the audio for transmission. This also reduces the quality of the audio. BLE midi only transmits simple commands to the receiver (e.g. turn note 56 on at volume 42) which do not need to be compressed and are quick to transmit. Audio quality depends only on the receivers capabilities.

Search through concertina.net on “sound font” and you’ll find many references to a high quality concertina sound font that has existed for years.

David Just to followup on my earlier comment about transcribing 'by hand' from a score to ABC. You do not have to be a fluid music notation reader to do this, God knows I am not one and never will be. But it is quite easy to read and transcribe, rather than to read and play, and in the process you do learn to read sheet music a little. I suggest that you print a copy of a simple tune and simply go through it note by note and pencil the ABC value for each note underneath the note. Once you have got a few bars notated then enter those into an ABC program and play it back - does it sound OK? Does it look the same as the original score? Yes, then carry on transcribing. If not then try to find the problem and fix it before going much further. Just transcribe four or five bars at a time, test, fix and repeat. At least initially, do not try to transcribe an entire tune in one go as this is harder to correct if you make a mistake. Just a warning about song books - I have a few and have found that I often disagree with the score in the book. I think that this is often because a professional transcriber has tried to reproduce the subleties of the human voice - and failed! You say that your objective is to be able to play a score so that you can learn a tune by ear. This is mostly what I do although I can slowly and haltingly read a score and play the tune. Funny thing is that once my fingers know a tune I cannot then read it at all! Anyway, I am going to put in a plug for Phil Taylor's concertina sound font (see below in my sig) and Musescore. Once you have a tune that works in one of the ABC players then try it in Musescore with the concertina sound font. I think that you will find that the sound is much, much better than that coming out of an ABC player and there are some handy looping and speed-changing tools, plus a metronome and a lead-in. The latest version of Musescore can also add a nice chordal accompaniment using a grand piano sound font - the same chordal accompaniment played by a concertina sounds awful!

Don, thank you for that! Brilliant! I have a question relating (I think) to your "Issues" point: I use the 'ABC settings>Voices' menu to explicitly select Flute (MIDI 74) and Fiddle (MIDI 111) as the 2nd and 3rd voices for my multi-voice tunes. Am I right in thinking that this will not work using your scheme, and that the 2nd and 3rd voices will play back using the concertina sound font? Or will the explicit voice selections over-ride the concertina font for these voices? I haven't tried this yet because (a) I'm currently using EasyABC 1.3.7.7; (b) I'm currently up to the earballs in other music programming tasks. If you are in a position to answer that question, I'd be grateful for your thoughts before trying this out. Thanks! Roger. PS: I use Tango Accordion as my 1st voice - of the available MIDI instruments, it's the one which sounds most like a concertina, to my tin ear, at least...

Yes. With mine he did four voices (sound files). 1, the 'as is'. (Precise samples of all notes as they are, worts and all) 2, the 'balanced voice' (balanced voice, volume and response) 3, The complete scale (stretched from G2 - D7 with all notes on the scale, including those not found on an anglo) 4, Complete scale, balanced, including breather hiss. He did it because, of the sound files he found, he found none that sounded convincingly like an anglo concertina when put into his compositions. (4 above sounds pretty amazing when a tune is played). I must admit I had not thought to direct him to concertina net, which I shall do now, Thank you David. But never-the-less, if anyone has the desire to capture the exact sound of their favourite concertina in a sound font, it can be done, which I must admit, until yesterday I had absolutely no idea was possible. (Please excuse my ignorance.)

The sound font, which should still be in my sig below, is in SF2 format which was the standard midi format for many years and should still be accepted by a midi player. (The SF2 format has been superceded by the SFZ format which gives more human sound music). There is one problem with this sound font in that it gets confused with very high notes and plays a note a fifth higher than played, but I have never found to be a problem with music that I want to play. If you play concertina and use Musescore then this font works very nicely in Musescore.

I've updated my free Anglo Concertina tunebooks with fingerings with slightly larger and easier-to-read fingering indications and added a Table-of-Contents along with the existing Index. Click any tune name in the Table of Contents or Index to immediately jump to that tune. On the tune pages, click tune title to play the tune in my ABC Transcription Tools using my high-quality Concertina samples. Click the << on any tune to jump to the Table of Contents. Click the >> on any tune to jump to the Index. Get the latest versions here: http://michaeleskin.com/tunebooks.html I also spent some time optimizing reducing the size of the concertina sound font by limiting the individual note samples to 3 seconds. That seems to be typical for these kind of sound fonts. Reduced the size by about half, which should dramatically improve playback loading times. If you need a note longer than three seconds, just re-articulate it in the ABC.