Re: [Rosegarden-user] Unwanted CCs - update - !!!AHA!!!
Perhaps I misunderstood your setup - are you reading a MIDI file in, it contains bank changes, and when you play back they are messed up? Or they show the wrong banks in the Instrument Parameter Panel? That would make this code suspect, otherwise I'm afraid I don't see a connection. OK, catching up on my email, I read the other thread (without AHA!), so I see that is essentially what's going on. Now it's just nagging at me, if simply copying over the lsb and msb isn't right, what *is* supposed to happen? Or have we overlooked some minor mistake that corrupts the whole thing? Will look further, and look at the MIDI Abrolag sent - thank you for that. Tom Breton (Tehom) -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs - update
On Thu, 5 Feb 2015 20:22:34 + Abrolag abro...@users.sourceforge.net wrote: [...] I've tried attaching the MIDI version of my test file if anyone wants to have a look at it and see what happens. Have looked at your test MIDI and let MidiEvent::print trace it out. Thank you, and also thank whoever created MidiEvent::print years ago. The one wrong-looking thing I see is that we never erase the msb and lsb variables. Once case MIDI_CTRL_CHANGE has set lsb or msb, case MIDI_PROG_CHANGE will consider it set to that value, forever, until case MIDI_CTRL_CHANGE sets another value. We don't aim lsb and msb at a specific channel, either, so other channels' MIDI_PROG_CHANGEs could be affected. My suspicion is that remembering lsb,msb after MIDI_PROG_CHANGE confuses later MIDI_PROG_CHANGEs that don't have bank controllers, and also we probably shouldn't share them between channels. Does this make sense to you? Tom Breton (Tehom) PS: Added a CC to the devel list. Does this still need to go to RG-user? -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs - Long!
On Wed, 11 Feb 2015 14:22:24 -0500 Tom Breton (Tehom) te...@panix.com wrote: On Thu, 5 Feb 2015 20:22:34 + Abrolag abro...@users.sourceforge.net wrote: [...] I've tried attaching the MIDI version of my test file if anyone wants to have a look at it and see what happens. Have looked at your test MIDI and let MidiEvent::print trace it out. Thank you, and also thank whoever created MidiEvent::print years ago. The one wrong-looking thing I see is that we never erase the msb and lsb variables. Once case MIDI_CTRL_CHANGE has set lsb or msb, case MIDI_PROG_CHANGE will consider it set to that value, forever, until case MIDI_CTRL_CHANGE sets another value. We don't aim lsb and msb at a specific channel, either, so other channels' MIDI_PROG_CHANGEs could be affected. My suspicion is that remembering lsb,msb after MIDI_PROG_CHANGE confuses later MIDI_PROG_CHANGEs that don't have bank controllers, and also we probably shouldn't share them between channels. Does this make sense to you? Tom Breton (Tehom) PS: Added a CC to the devel list. Does this still need to go to RG-user? Hi Tom. Here be dragons! :) I've been wrestling with bank program changes since my first hardware synth in 1991 - and I'm still not sure I know what I'm talking about. However, I still have 3 hardware synths (although seldom used now). None use both MSB LSB One uses MSB, the others use LSB The two that use LSB actually send both (zero for MSB). The other one only sends MSB All send a program change whenever they send a bank change. None want a bank change when *getting* a program change they remember their last bank setting (including over a power cycle). One of them, QS300, can accept a program change 'silently' to a non-sounding channel which will then pick up cleanly on the next note, but if you send a bank change as well the entire sound cuts out. They all only transmit on one channel, but will accept MIDI on all 16. Bank is applied to all channels whichever channel it's sent on. Program is channel specific. This makes sense as bank changes are swapping out ROM sets, program changes just select a voice within the ROM. My most recent one is mid 1990s so things have probably changed considerably since then. I would expect them to be even more flexible. You can probably do silent bank changes now. As far as I know there are still none that use both MSB LSB though! There is *no* sensible plan as to what bank numbers are used. 1,2,3,4,5... forget it! With all of this I think the only safe thing for Rosegarden to do is to pass on exactly what it gets without trying to make any interpretation. How it wants to represent banks for display purposes is really up to you, but currently it changes them (seemingly just LSB) and I think that is quite wrong. Since working on Yoshimi, I've developed an even more keen interest in this and in order to try and keep up, our bank changes can be set to receive either MSB or LSB, and work across all channels but have no effect at all until there is a (channel specific) program change. As there is no physical ROM the synth engine only knows about voice patches currently in memory. We effectively have an equivalent to an alternate MSB or LSB as we have a control that sets the entire directory structure that banks are read from (also silently). The test file checks that all these work including trapping invalid bank and program changes. You'll notice there are changes on one channel while others have continuous notes. I would guess that most soft-synths would work in pretty much the same way if they accept MID controls. P.S. I'm not on the developers list just the users one. If this is getting too involved for users I guess I could subscribe, but I'd actually rather stay as I am :) P.P.S. As a matter of interest, I've attached the most complete MIDI chart V2 I can find. It's actually a merge of several! -- Will J Godfrey http://www.musically.me.uk Say you have a poem and I have a tune. Exchange them and we can both have a poem, a tune, and a song. Title: MIDI Standard Controller Numbers MIDI Controller Numbers Summary of the MIDI controller codes (V 2.0) in decimal and hexadecimal DecimalHexController Name 000hBank Select MSB - (rarely implemented) 101hModulation Wheel MSB 202hBreath Controller MSB 303h- - 404hFoot Controller MSB 505hPortamento Time MSB 606hData Entry MSB 707hVolume MSB - (formerly Main Volume) 808hBalance MSB 909h- - 100AhPan MSB 110Bh_expression_ MSB 120ChEffect Control 1 MSB 130DhEffect Control 2 MSB 140Eh- - 150Fh- - 1610hGeneral Purpose Controller 1 MSB 1711hGeneral Purpose Controller 2 MSB 1812hGeneral Purpose Controller 3 MSB 1913hGeneral Purpose Controller 4 MSB 2014h- - 2115h- - 2216h- - 2317h- - 2418h- - 2519h- - 261Ah- - 271Bh- - 281Ch- - 291Dh- - 301Eh- - 311Fh- - 3220hBank Select LSB 3321hModulation
Re: [Rosegarden-user] Unwanted CCs - update - !!!AHA!!!
On Tue, 10 Feb 2015, D. Michael McIntyre wrote: I was wondering about that, now that you mention it. Controller banks? Does anything support more than 127 controllers? Are bank changes Controllers have standardized numbers - e.g. 7 is master volume - so the main issue isn't really that a single device has more than 127 controllers (although yes, some DO), but that there are more than 127 different kinds of controllers in the world. It's made worse by the fact that a single controller number only handles a 7-bit value, so if you need finer granularity than that, you need to assign two or more of the 127 controller numbers to the more and less significant chunks of a finer-grained value. And it's also true that controller numbers ended up being used for other things, such as selecting banks for the separate program change messages. The namespace of 127 controllers ran out very fast. -- Matthew Skala msk...@ansuz.sooke.bc.ca People before principles. http://ansuz.sooke.bc.ca/ -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs - update - !!!AHA!!!
I just had a nose around in MidiFile.cpp - without full hazmat too :) At line 1221 you have the start of a section that interprets bank changes. This is very risky as almost every synth maker has a different way of handling them. Commenting out both the msb and lsb code appears to cure the problem. Can someone more knowledgeable have a look and see what other implications there may be? -- Will J Godfrey http://www.musically.me.uk Say you have a poem and I have a tune. Exchange them and we can both have a poem, a tune, and a song. -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs - update - !!!AHA!!!
On 02/10/2015 01:46 PM, Abrolag wrote: I just had a nose around in MidiFile.cpp - without full hazmat too :) At line 1221 you have the start of a section that interprets bank changes. This is very risky as almost every synth maker has a different way of handling them. I was wondering about that, now that you mention it. Controller banks? Does anything support more than 127 controllers? Are bank changes used for something else? Was that code meant to tackle a known case? The answer to all those questions is probably but yet I somehow doubt our good-intentioned code even works anyway. The fastest way to find out whether anybody is using something or not is to disable it and wait for the complaints. H. -- D. Michael McIntyre -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs - update
On 02/09/2015 05:10 PM, Abrolag wrote: I've tried attaching the MIDI version of my test file if anyone wants to have a look at it and see what happens. I'm piddling around... The bank changes (CC32) in order should be: CH1, 105 CH1, 105 CH1, 7 CH2, 110 CH3, 105 CH1, 5 105 110 105 105 Lots of weird things going on here. First, there was no 7 or 5 in the data. Second, we map either 110 or 105 to 255, sure enough. I didn't sort it out before hitting the limit of my attention span. Stuff is created in MidiFile.cpp, but who what where when and why remains a mystery to untangle, if anyone is in the mood for untangling mysteries. I, apparently, am not as much in that mood as I first thought. -- D. Michael McIntyre -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs - update
On Thu, 5 Feb 2015 20:22:34 + Abrolag abro...@users.sourceforge.net wrote: Just tried build 13868 and can confirm that it now sends the correct data. However on *importing* a midi file CC32 values are set as either 0 or 255. This is bank LSB and should be settable 0-127. A value of 255 should be quite impossible! The actual values I've set (and correctly sent) are 105, 7, 110, 5. Just to confirm this is designed as a test file, but it's supposed to be testing Yoshimi! :) Just tripple checked this with aplaymidi, ardour3 and muse2. Rosegarden 13868 is *definitely* exporting the correct MIDI data and it is purely the import side that is now having problems. P.S. after using those last two I was *extremely* grateful to be able to return to Rosegarden, for all their bells and whistles they simply are not musician friendly :( I've tried attaching the MIDI version of my test file if anyone wants to have a look at it and see what happens. The bank changes (CC32) in order should be: CH1, 105 CH1, 105 CH1, 7 CH2, 110 CH3, 105 CH1, 5 This is intersped with program changes and actual (loosely described) music. -- Will J Godfrey http://www.musically.me.uk Say you have a poem and I have a tune. Exchange them and we can both have a poem, a tune, and a song. RootBankProgChangeTest.mid Description: MIDI audio -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs
On Wed, 4 Feb 2015 20:20:11 -0500 Tom Breton (Tehom) te...@panix.com wrote: On 02/04/2015 05:49 PM, Abrolag wrote: Yes tried that. It still sends CC 121 (reset all controllers) and it sends it at the start of every segment of every track! ChannelManager::setControllers() appears to be the function doing this. Oh, my. It's right where I wrote: // This was the old logic, but it's not clear that it is still // desirable and I think I ifdef'ed it out after we let it fly for a while and nobody seemed to be using it. Obviously that's not the case, so I'll put it back in immediately. Tom Breton (Tehom) Just tried build 13868 and can confirm that it now sends the correct data. However on *importing* a midi file CC32 values are set as either 0 or 255. This is bank LSB and should be settable 0-127. A value of 255 should be quite impossible! The actual values I've set (and correctly sent) are 105, 7, 110, 5. Just to confirm this is designed as a test file, but it's supposed to be testing Yoshimi! :) -- Will J Godfrey http://www.musically.me.uk Say you have a poem and I have a tune. Exchange them and we can both have a poem, a tune, and a song. -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs
On 02/04/2015 08:03 AM, Abrolag wrote: Every time Rosegarden starts up, and every time it starts a run from the beginning it sends reset CCs. I can't seem to find a way to stop this - they are not mentioned in any filter options that I can see. It makes a mess with the new version of Yoshimi which responds to these messages and tries to set banks, programs etc. that don't exist :( If I disable response to banks and program change, then I can't embed the CCs I *do* want into a RG file! Maybe this is related to how RG always resets Yoshimi to its default voice when I start playing a track for which I've already set a voice in Yoshimi? -- David W. Jones gn...@hawaii.rr.com authenticity, honesty, community http://dancingtreefrog.com -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs
Every time Rosegarden starts up, and every time it starts a run from the beginning it sends reset CCs. I can't seem to find a way to stop this - they are not mentioned in any filter options that I can see. Sounds like the normal setup stuff. Have you tried unchecking all the checkboxes in Instrument Parameters? Basically, that tells it not to try to set up an instrument, because setup will be handled externally, which sounds like what you are doing. Tom Breton (Tehom) -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs
On 02/04/2015 05:49 PM, Abrolag wrote: Yes tried that. It still sends CC 121 (reset all controllers) and it sends it at the start of every segment of every track! ChannelManager::setControllers() appears to be the function doing this. Ted. -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs
On Wed, 4 Feb 2015 17:00:52 -0500 Tom Breton (Tehom) te...@panix.com wrote: Every time Rosegarden starts up, and every time it starts a run from the beginning it sends reset CCs. I can't seem to find a way to stop this - they are not mentioned in any filter options that I can see. Sounds like the normal setup stuff. Have you tried unchecking all the checkboxes in Instrument Parameters? Basically, that tells it not to try to set up an instrument, because setup will be handled externally, which sounds like what you are doing. Yes tried that. It still sends CC 121 (reset all controllers) and it sends it at the start of every segment of every track! What is worse is that it then can't reload what it sent. CC 32 bank LSB always comes back with a value of 0. It is also creating spurious CCs but I can't pin down whether that is happening on export or reload. BTW I've also tried using arecordmidi to capture the actual midi stream. -- Will J Godfrey http://www.musically.me.uk Say you have a poem and I have a tune. Exchange them and we can both have a poem, a tune, and a song. -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs
On Wed, 04 Feb 2015 17:59:05 -0500 Ted Felix t...@tedfelix.com wrote: On 02/04/2015 05:49 PM, Abrolag wrote: Yes tried that. It still sends CC 121 (reset all controllers) and it sends it at the start of every segment of every track! ChannelManager::setControllers() appears to be the function doing this. Ted. Would it help If I made my test rg file available? it's only 23k. I can't remember, are attachments accepted on this list? -- Will J Godfrey http://www.musically.me.uk Say you have a poem and I have a tune. Exchange them and we can both have a poem, a tune, and a song. -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
Re: [Rosegarden-user] Unwanted CCs
On 02/04/2015 05:49 PM, Abrolag wrote: Yes tried that. It still sends CC 121 (reset all controllers) and it sends it at the start of every segment of every track! ChannelManager::setControllers() appears to be the function doing this. Oh, my. It's right where I wrote: // This was the old logic, but it's not clear that it is still // desirable and I think I ifdef'ed it out after we let it fly for a while and nobody seemed to be using it. Obviously that's not the case, so I'll put it back in immediately. Tom Breton (Tehom) -- Dive into the World of Parallel Programming. The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net/ ___ Rosegarden-user mailing list Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-user
