Humming a melody, midi repeatability problem

• Jul 6, 2019 - 23:03

I tried a nice audio to midi code, A2M (in Windows 10). Since it can converts an instrument generated voice or whistled or hummed voice to MIDI signal and controls MS Midi mapper accordingly, I tried to get it also inputting notes to Musescore. For that purpose I installed loopMIDI driver. I managed to forward hummed midi signals to MPK keyboard and suddenly several notes to Musescore but couldn't repeat it.

I activated the loopMIDI and A2M plus MPK before starting Musescore, thus getting the loopMIDI available to midi device input dropdown menu. However something doesn't go as expected so the success wasn't repeatable. If anyone has tips how to setup such system permanently, all tips are most wellcome.


Ok, got it working somehow. Trick was not to have any keyboard connected, starting loopMIDI (remember to create one port "loopMIDI Port 1") and A2M having output to loopMIDI Port in this order. Then setting MuseScore MIDI-input to MMSystem, loopMIDI Port. Remember to restart Musescore, it seems to be the only way to get correct settings set up! Notes can be input by realtime manual input with right and left arrows when humming/whistiling. However automatic realtime can't work because this midi setup creates a loop for the midi signals somehow and one gets suddenly maaaaany notes! There is clearly a reverb, it can be heard when humming and also when playing back. Musescore doesn't probably support usage of separate channels (or does it?) to prevent reverb. So any tips from experienced MIDI-gurus are now most wellcome!

Also a tiny wish to Musescore writers, joining together notes to form longer duration notes automatically after manual input would be a wonderful help, just as automatical input does.

In reply to by Marc Sabatella

Thank you Marc, I will try that.

Meanwhile I got some progress in testing. What one has to do, is to define two separate midi ports in loopMIDI, on for input, say Port_in and another for output, say Port_out. Then in A2M define OUPUT to go into Port_in and further in Musescore Settings->I/O MIDI-input to be this Port_in. Use port_out in Musescore for outputting midi-commands, they seem to be automatically forwarded to MS MIDI Mapper. This removes the echo from humming/whistle input and playback, at least with headphones and external microphone. However there is still the annoying loopback which fills staves with machine gun speed if using automatic realtime input. I wonder whether it is something depending on Musescore features and can one somehow prevent it. According to the loopMIDI definition it is looping midi commands among midi devices and virtual devices, it may be that without separate in/out channels this can not be avoided? Any tips still most appreciated!

Just discovered that the machine gun behavior comes from metronome which signal is sent out from Musescore and read in from loopMIDI as a command for next note. Depending on the timing settings, it is necessarily not recognized as next input note. If timing is suitable, one gets lot of B:s to staves because of looping. So, is there a method to force Musescore recognizing it's own metronome ticks and reject them as note inputs? That would probably solve the problem.

Do you still have an unanswered question? Please log in first to post your question.