CC#120 all sound off not implemented

[jpcima]

It does not handle this controller. It is supposed to mute a channel's sound output. In my own fmidi sequencer, I send CC120 to silence the audio when there is a change of track. TiMidity++ recognizes it.

Indeed code reveals it is not a handled cc number. There is however 123, a similar one. (all notes off). From this, I guess it's probably not a thing on my side to implement, but an omission of ADLMIDI.

cf. MIDIplay::realTime_Controller, adlmidi_midiplay.cpp

[Wohlstand]

I think, I need to add that, useful event that must be supported!

EDIT: Yeah, seems to implement I will need to do same as CC123 but also shut up all chip channels.

[Wohlstand]

Just now I have added that! :fox_face: The test MIDI file I made to test both CC123 and new CC120! сс120-123-test.mid.zip

[jpcima]

This is not right. Playing back a normal midi I can hear nothing except percussions now. I have pulled your commits into adljack. Until this, I just changed libADLMIDI to make cc120 do the same as 123.

For reference this is how I use the cc120 (which worked with timidity in alsa midi server). midi_reset I send this event sequence to reinitialize the instruments between tracks. In reality I don't take this event from the midi files.

[Wohlstand]

I have checked the behavior of those events on FluidSynth until I began to implement it. Anyway, can you send me the MIDI file to let me polish my stuff with it? The CC120 must shut up all playing notes completely without giving them to play release while CC123 just sends NoteOffs to all playing notes in a current channel.

EDIT: Mine a bit updated example where I have been added some notes following CC120 in the second channel: сс120-123-test.mid.zip

EDIT2: I have checked Timidity too, that works fine at me, or what do you mean "not right"? Can you illustrate the results?

[jpcima]

It's not a problem of one MIDI file, it's a problem of all of them. The only thing is that I work in realtime playback.

Before I pull the last commits => I have sound After => only percussion

And the stranger thing is, it does not even happen in relation with all sound off events. On a fresh instance of an ADL player, where cc120 is not sent, the sound channels still start muted.

[jpcima]

Here is a picture. Maybe it shows what I mean better.

[jpcima]

Ok here's breaking news. It must have to do with the new commits about emulation cores. Because without these but only the cc120 patch, the sound does play.

edit: and cc120 does its intended thing, I shall add

[Wohlstand]

Anyway, I checked out your adljack, and seems it works unstable, and plays few notes, I ran RtMidi thing I built from sources together with RtAudio. Even I playing MIDI file which has no any CC120 events. Also, I have reduced minimal CMake into 3.5 from 3.6 as I have 3.5 installed, and C++14 instead of C++17 as my compiler GCC 5.4.1 doesn't supports it as I can see. However, everything was built with no errors.

[Wohlstand]

P.S. I'll aply CC120 patch to OPNMIDI too as it's confirmed it working right.

[jpcima]

I checked out your adljack, and seems it works unstable

(ah! Maybe I know why, now you mention it. I have to test alsa on motherboard audio, because I'm on a firewire card)

Is the new NukedOPL 1.8 the origin of the sound problems?

[Wohlstand]

Oh, yeah, Nuked OPL3 1.8 had some sounding issue I have been resolved!

[jpcima]

I possibly fixed adlrt, please check. The default latency of rtaudio was just set way too low. I add the flag [-L latency-ms] to choose a latency value. default 20ms

[Wohlstand]

I hope, this thing seems useful, Ill try with bigger latency a bit later, for now I gonna sleep and I have powered down my PC.

Отправлено с Samsung Mobile

-------- Исходное сообщение -------- От: JP Cimalando notifications@github.com Дата:10.04.2018 5:08 (GMT+03:00) Кому: Wohlstand/libADLMIDI libADLMIDI@noreply.github.com Копия: Vitaly Novichkov admin@wohlnet.ru,Assign assign@noreply.github.com Тема: Re: [Wohlstand/libADLMIDI] CC#120 all sound off not implemented (#48)

I possibly fixed adlrt, please check. The default latency of rtaudio was just set way too low. I add the flag [-L latency-ms] to choose a latency value. default 20ms

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub, or mute the thread.

[jpcima]

Ok thanks for all the things. Good night

[jpcima]

Just for the record so far here's what I know about this sound problem. If you can run adl-rt, and pull latest libADLMIDI, it's a decent chance you will experience what I speak about.

• the problem has appeared in the timeframe after 0ad695e : from 8a0a939 up to cd09e38
• it does not matter which is the emulator selected. same with Nuked and Dosbox
• it's only in realtime playback. adlmidiplay is OK
• it manifests as if MIDI events are skipped or delayed, but only a few of them pass. It's a best way I can describe this bizarre behavior. It results as if a lot of the music is muted.

[Wohlstand]

@jpcima, Okay, just now I figured out for reason why that thing wasn't worked: https://github.com/Wohlstand/libADLMIDI/commit/ff0982dd98cd7a828730a3498e3e7b9cd1daf868 the mistake was so stupid :man_facepalming: OPLChipBase::generate() does replace of content of given buffer, and I needed to use OPLChipBase::generateAndMix() call which mixes a given buffer with new-generated data.

When I fixed that, now latest state playing fine at my machine, and with Latency 5 milliseconds that playing fine :smile:

P.S. @jpcima, THANKS for that thing! :+1:

[jpcima]

Oh thanks god for finding this!!!! I was going to look into it more after doing some details in the rt player.

[Wohlstand]

Okay, as the main topic of this issue resolved, I'll close it. For RTMIDI discussion let's create a new issue :wink:

[Wohlstand]

Discussion for RT-MIDI is moved to here: https://github.com/Wohlstand/libADLMIDI/issues/53