TD-3 Format support

[parashu-rama]

Hello!

I want to convert TD-3 seq to Crave seq.

Can you help me with TD-3 seq file support?

[claziss]

Sure, plz let me check the TD-3 format.

[parashu-rama]

Thank You! Da Funk.seq.zip

[claziss]

Hey, I will do first a TD-3 sequencer dumper, then I'll move in creating the convertor to Crave seq. It will take a bit.

[parashu-rama]

Thank You!

[claziss]

Hey @parashu-rama , please can you share with me a simple TD3 sequence of lenght 16, where u have a single C3 note in the first position? I do not have a TD3 and the Behringer tool requires one. I may need several such simple sequences to identify all the td3 seq file info.

[parashu-rama]

Me either :) Found this https://303patterns.com/ Will ask for support!

[parashu-rama]

Bought One! Wait for delivery :)

[bradisbell]

Hello all. I received an e-mail requesting me to check this thread.

I've documented the TD-3 MIDI implementation here: https://303patterns.com/td3-midi.html I haven't looked at the sequence files saved from Synthtribe in awhile, but if I remember correctly, they basically followed the same format as you get via SysEx. If you find this documentation useful, please feel free to consider dropping a link back to 303patterns.com. :-)

Also, if you want to use the Synthtribe tool before your actual TD-3 arrives in the mail, all you need is a MIDI port labeled TD-3. Use your favorite MIDI tool to create such a port, such as Tobias Erichsen's rtpMIDI or loopMIDI: http://www.tobias-erichsen.de/software.html

[claziss]

@bradisbell Thanks for the info !

[claziss]

@parashu-rama I think I've cracked. The listing of DaFunk:

[G1       ]
[A#3          ]
[G2   A   ]
[F3       ]
[D#2    S A   ]
[C2 S     ]
[C1 S A   ]
[C3 S     ]
[D#3      A   ]
[D#1          ]
[F3   A   ]
[G3 S A   ]
[A#1          ]
[G1   A   ]
[C1   A   ]
[G2   A   ]

I still need to clean the implementation up, there are some fields which are not clear to me what they represent. However, we can go further a bit and think how we map the TD-3's slide and accent. Clearly we can map TD3's accent to Crave's accent, but what we do with TD3 slide? Shall we map it to Crave's glide or note length?

[parashu-rama]

Congratulation! My td-3 should be delivered before September 13. We can test the sound of slide in both versions or implement both options to choose from using the command line parameter. Or ignore it :)

eg --slide none/glide/length

[claziss]

That is a good idea.

[parashu-rama]

Manually enter pattern in SynthTribe with glide and try to txtdump it.

1. Unknown header because crave version 1.1.2
2. Change to 1.1.1 in hexeditor - octave mismatch - but sound good

Swing: 50%   Length 16
[G2          64 x1 ####     ]
[A#3             64 x1 ####     ]
[G3   A      64 x1 ####     ]
[F4          64 x1 ####     ]
[D#3    G A      64 x1 ####     ]
[C3 G        64 x1 ####     ]
[C2 G A      64 x1 ####     ]
[C4 G        64 x1 ####     ]
[D#4      A      64 x1 ####     ]
[D#2             64 x1 ####     ]
[F4   A      64 x1 ####     ]
[G4 G A      64 x1 ####     ]
[A#2             64 x1 ####     ]
[G2   A      64 x1 ####     ]
[C2   A      64 x1 ####     ]
[G3   A      64 x1 ####     ]

dafunk-crave.seq.zip

[claziss]

It is an interesting project, but step by step. Let's first finish the the TD3 to Crave (and other way around?) project.

[claziss]

@parashu-rama I've pushed a crude TD3 seq parser. Please give it a try and let me know how does it work. To parse a TD3 seq file just pass -t option. For example:

 ./txtdump -t td3/DaFunk.seq 

The txdump without the -t option will parse the crave specific sequence.

[parashu-rama]

Works like a charm! Thanks!

[parashu-rama]

Probably Crave octave mismatch

CraveSeq % ./txtdump -t Da\ Funk.seq
Parse TD3 seq file ...
Swing: 0%    Length 16
[G1          ]
[A#3             ]
[G2   A      ]
[F3          ]
[D#2    S A      ]
[C2 S        ]
[C1 S A      ]
[C3 S        ]
[D#3      A      ]
[D#1             ]
[F3   A      ]
[G3 S A      ]
[A#1             ]
[G1   A      ]
[C1   A      ]
[G2   A      ]

CraveSeq % ./txtdump xz.seq

Swing: 50%   Length 16
[G2          64 x1 ####     ]
[A#4             64 x1 ####     ]
[G3   A      64 x1 ####     ]
[F4          64 x1 ####     ]
[D#3    G A      64 x1 ####     ]
[C3 G        64 x1 ####     ]
[C2 G A      64 x1 ####     ]
[C4 G        64 x1 ####     ]
[D#4      A      64 x1 ####     ]
[D#2             64 x1 ####     ]
[F4   A      64 x1 ####     ]
[G4 G A      64 x1 ####     ]
[A#2             64 x1 ####     ]
[G2   A      64 x1 ####     ]
[C2   A      64 x1 ####     ]
[G3   A      64 x1 ####     ]

[claziss]

Yes the octave number may be off by one as I am counting from zero. I will check this.

[claziss]

@parashu-rama Please check the latest push. I've added the '-d' option which is parsing the TD3 sequence file, and dumps it as a Crave sequence file. The Crave sequence file has .crave.seq extension. It is still a very crude implementation, but the output looks alright.

I've also corrected the octave missmatch.

[parashu-rama]

It working! Thank You! https://youtu.be/Q4KkNYIL8iE

[parashu-rama]

At last a simple TD3 sequence of lenght 16, where u have a single C3 note in the first position. singleC3.seq.zip

[claziss]

It looks like the last 4bytes from seq file are set. I need to figure out their meaning. The TB 303 (and TD3) should have something about note length, maybe this is the one.

[parashu-rama]

TD-3 gate length 50 and 100 c3l.seq.zip

Crave

[claziss]

When slide is on, the TD-3 editor shows the note as 100% open.

[parashu-rama]

oops, sorry

Any other test?

[parashu-rama]

what we do with TD3 slide? Shall we map it to Crave's glide or note length?

I think 100% length better parser.c.diff.txt

[claziss]

what we do with TD3 slide? Shall we map it to Crave's glide or note length?

I think 100% length better parser.c.diff.txt

Updated as you suggested. Plus I don't check for TD3 version any longer.

[parashu-rama]

Thanks!

Are you interested in parsing the uno synth patch sequence or I close this issue?

[claziss]

I need to fix the obvious snafus for this implementation first. Probably it is a good idea to use something more highlevel like Python, and after that to parse the Uno too. I'll close the issue my self.

[parashu-rama]

How I can help You?

[claziss]

For the time being nothing to do on your side, I will have an update soon, and then I would like to ask you to compare the sound from Crave and TD3 ;) Still decoding the last bits of TD3 seq format, and I think I've got a good idea what every filed does.

[parashu-rama]

https://youtu.be/CyGYl5CtBpI

[claziss]

@parashu-rama Thanks for the demo, any chance to use your TD-3 and record only the DaFunk seq in a wave and share it? Also, please find my latest update, with the missing link ;) namely the sustain note of TB-3. I do not think, I do a good job when creating the crave sequence with sustain.

[parashu-rama]

140921.m4a.zip

[claziss]

I've made some mods in how to handle the TD3' sustain pitch feature. I hope it sounds good on a Crave. Anyhow, the above dafunk track looks all right although I couldn't get to much acid from my Crave. @parashu-rama please make a new issue with the Uno sequencer, I'll check what is its format in a shot while. Cheers.

[parashu-rama]

Last note length bug? ./txtdump -d I\ feel\ love.seq Parse TD3 seq file and convert... Swing: 50% Length 8 01:[C2 A #### ] 02:[C3 A #### ] 03:[C2 A #### ] 04:[C3 A #### ] 05:[G2 A #### ] 06:[G2 A #### ] 07:[A#3 A #### ] 08:[A#3 A ######## ]

I feel love.zip

[claziss]

@parashu-rama fixed

[parashu-rama]

Thanks!

[parashu-rama]

Probably octave mismatch 02 step?

./txtdump -t Da\ Funk.seq Parse TD3 seq file ... Swing: 50% Length 16 01:[G1 #### ] 02:[A#4 #### ] 03:[G2 S ######## ] 04:[F3 #### ] 05:[D#2 S A ######## ] 06:[C2 A #### ] 07:[C1 S A ######## ] 08:[C3 A #### ] 09:[D#3 S ######## ] 10:[D#1 #### ] 11:[F3 S ######## ] 12:[G3 S A ######## ] 13:[A#2 #### ] 14:[G1 S ######## ] 15:[C1 S ######## ] 16:[G2 S ######## ]

[claziss]

C2 (TD3) is 0x18 (24 decimal), the second position has 0x2e value (46 decimal), hence, there are 22 pitches between the two notes:

Note	Value
C2	24
C#2	25
D2	26
D#2	27
E2	28
F2	29
F#2	30
G2	31
G#2	32
A3	33
A#3	34
B3	35
C3	36
C#3	37
D3	38
D#3	39
E3	40
F3	41
F#3	42
G3	43
G#3	44
A4	45
A#4	46
B4	47

To me, the second pitch should be A#4.

[parashu-rama]

I think that octave begins with C, not A .

[claziss]

Yes, that is true, if you are considering C major scale. I use A major scale for printing. However, the seq file should be correct. I will revert this patch

[parashu-rama]

Thank you for the clarification!