Add support for MIDI SDS and TM-1

[dagargo]

The goal is to transmit samples to Machinedrum via TM-1.

TM-1 is a standard USB to MIDI (DIN5) interface that can also speed up MIDI transmission if the Elektron devices connected to them allow that.

As it seems that the underlying protocol is standard MIDI SDS, perhaps we could extend this functionality to any other device that supports it too.

Here we have some useful links taken from here.

Format:
http://elektron-users.com/index.php?option=com_openwiki&Itemid=43&id=waveforms_samples 94

Batch conversion:
http://www.elektronauts.com/t/best-software-for-batch-sample-rate-conversion/510

Discussion:
http://www.elektron-users.com/index.php?option=com_fireboard&Itemid=28&func=view&id=30234&catid=9&limit=10&limitstart=0 27

[dagargo]

More official information at Elektronauts but not related to samples.

[dagargo]

Working MIDI SDS pushed to branch sds.

As I don't have a TM-1, I suggest you to work on the connector_handshake_midi_sds function in here.

elektroid/src/connector.c

Line 2506 in bdddae8

connector_handshake_midi_sds (struct connector *connector)

[dagargo]

SDS support completed.

[dagargo]

sds branch merged into master and deleted.

[dagargo]

A lot of refactoring has been going on lately so the instructions above are obsolete.

In order to send a custom SysEx message to the TM-1, the code should be added at the end of this function. Other devices will ignore it so it will be safe.

elektroid/src/connectors/sds.c

Lines 763 to 785 in aff3aab

	gint
	sds_handshake (struct backend *backend)
	{
	//We send a dump header for the highest number allowed. Hopefully, this will fail on every device.
	GByteArray *tx_msg = g_byte_array_sized_new (sizeof (SDS_DUMP_HEADER));
	g_byte_array_append (tx_msg, SDS_DUMP_HEADER, sizeof (SDS_DUMP_HEADER));
	tx_msg->data[4] = 0x7f;
	tx_msg->data[5] = 0x7f;
	//In case we receive anything, there is a MIDI SDS device listening.
	if (sds_upload_tx_and_wait_ack (backend, tx_msg, 0) == -ETIMEDOUT)
	{
	return -EIO;
	}
	//However, we cancel the upload just in case the sampler sent an ACK and is waiting for data.
	sds_tx_handshake (backend, SDS_CANCEL, sizeof (SDS_CANCEL), 0);
	backend->device_desc.filesystems = FS_SAMPLES_SDS;
	backend->fs_ops = FS_SDS_OPERATIONS;
	snprintf (backend->device_name, LABEL_MAX, "sampler (MIDI SDS)");
	return 0;
	}

Notice that we are assuming that TM-1 has only one MIDI interface and it's reading and bypassing the messages.

The code should look like this.

static const guint8 SDS_TM1_TURBO_ON[] = { 0xf0,..., 0xf7 };
[...]
backend_tx_raw (backend, SDS_TM1_TURBO_ON, sizeof(SDS_TM1_TURBO_ON));

[erroreyes]

I am going to try to test this out tomorrow (PST). I emphasize "try" because my TM-1 is really unstable in my Linux/Lenovo laptop because I have to do this plug/unplug/plug-while-holding-the-turbo-button dance to get it to stay recognized and connected by the machine/OS, and even then, I haven't been able to get a simple command like amidi -p hw:1,0 -s some_file.syx -i 20 to work with it, while it works fine with an equally as old Native Instruments Audio Kontrol 1 audio/MIDI interface.

PS, this is in response to your recent post on Elektronauts. I didn't want to reply there as I think a technical conversation may be more relevant here.

[dagargo]

PS, this is in response to your recent post on Elektronauts. I didn't want to reply there as I think a technical conversation may be more relevant here.

Abosulutely! Fully agree.

I forgot to mention that, for now, you should change the following line.

#define SDS_BITS 16

I think it'a a 12 there in your case. We should set a variable depending on the MIDI identity reply but we'll do it later.

[dagargo]

Any news on this?

I've just pushed a few commits that should improve the SDS functionality. It's been tested and works well with my Digitakt (only upload)) and my E-Mu ESI-2000 (except when the device doesn't send the last byte).

Now, there are 4 filesystems; for 8, 12, 14 and 16 bits transmissions.

I'm still trying to figure out a couple of issues but hopefully this won't affect you if the Machinedrum MIDI SDS code is the same it's in the Digitakt.

One of the issues is the Receive SysEx option in the menu, it's not working ATM.

[erroreyes]

Hey, sorry about the delay due to unforeseen personal things.

I meant to test this exclusively with my TM-1, but with my current Linux setup, the TM-1 just doesn't want to connect permanently. I will keep trying since I have in fact been able to connect it to the laptop well by doing a bunch of brute force connecting and disconnecting maneuvers. That said, in the times that I have managed to keep it connected, Turbo mode hasn't worked.

With that in mind, do you need me, or is it possible and beneficial, to test the new changes to Elektriod on the Machinedrum without having to use the TM-1, i.e., can I use it with, say, my Native Instruments Komplete Audio 6 MKII?

[dagargo]

Hey, sorry about the delay due to unforeseen personal things.

No worries. First things first.

With that in mind, do you need me, or is it possible and beneficial, to test the new changes to Elektriod on the Machinedrum without having to use the TM-1, i.e., can I use it with, say, my Native Instruments Komplete Audio 6 MKII?

Testing the TM-1 should be the last thing to do. Right now, I need you because I'd like you to test the SDS against the Machinedrum and the integration in Elektroid and using Komplete Audio 6 MKII is perfectly fine. Later, we'll see how to integrate the TM-1 and its turbo mode.

As a side note, Elektroid offers direct sample copying with the GUI (and CLI) but also SysEx sending via the app menu (soon via the CLI too). The latter can be used to send the SysEx files found in these sample packs. BTW, SysEx sending only allows to send a single file but I'm working on sending multiple files at once.

Regarding a better integration with the Machinedrum, could you post here the output of elektroid -vv when selecting it in the device combo? I'd like to add a few tweaks if possible.

TIA.

[dagargo]

FYI, the bugs I mentioned a few days ago are fixed. Also, multiple SysEx uploading is now possible as well as using the CLI to send and receive SysEx.

[dagargo]

A couple of bugs have been fixed.

Also, an occasional error with an E-Mu ESI-2000 when downloading has been addressed with a little hack. Apparently, the sampler was not transmitting the last packet if it had only a single sample which cancelled the whole download. Now, Elektroid uploads an additional 0 sample if this is the case and then downloading just works. This affects all MIDI SDS devices but a single sample in this scenario is not much.

Edit: The additional sample was removed in b5a320f as another workaround was found.