Tempest Main 1.5.0.2 and Voice 1.5 - Bug Reports

[Roger_Linn]

Besides... the bug is still being misunderstood... the point is that SysEx DO NOT WORK VIA DIN... that's what I've been trying to say all the time... I've even written several times, that SYSEX WORK FINE WITH USB! ... the whole problem is when you want to use DIN for this... it DOES NOT WORK! ... it skips parameters from time to time, inside the dumps for some reason... even if you do not use USB at all.

You could reason, that it would be OK to just use USB instead... and yes! ... if the darn device had a custom MIDI driver that was multi client capable! ... DSI use the standard MIDI drivers of Windows, and they are both buggy, and not multi client capable... this results in it not being useable from more than one program at a time... and if you have both a sequencer and an editor you want to run at the same time, well... then you're fucked!

... last rant from here... over and out... I hope everyone else will get their most irritating bugs fixed... crossing my fingers for you...

Hi Razmo,

If I misunderstood your bug, I apologize. If I now understand you correctly, you are identifying two bugs in Tempest:

1) You mentioned that sysex dumps over DIN MIDI skip parameters from time to time. That’s odd because OS updates over DIN MIDI work fine, and those are very large sysex transfers that won’t work unless they are received perfectly. Perhaps the problem you experienced was that the transferred file did not contain—possible due to a bug—the specific parameter you think was not transferred? Or perhaps the parameter was a global parameter and therefore not affected by the transfer? Perhaps you were loading an older file and the older parameters weren’t being converted correctly to the newer file format due to a bug?

2) You fault DSI for not writing a custom MIDI-over-USB driver for Windows so that you can run both an external editor and an external sequencer at the same time. However, Tempest doesn’t use or need any MIDI drivers because its MIDI-over-USB implementation is USB class-compliant, meaning that no drivers are needed. I’m not aware of any manufacturer that writes a custom MIDI driver for USB, for either Mac or Windows. Are you? It seems to me somewhat unfair to to fault DSI for not writing a custom MIDI-over-USB driver for one customer’s unusual need, and for an OS (Windows) that in my experience is problematic for audio and music, especially considering that DSI is a hardware company that doesn’t write any software drivers for any of its products.

FYI, John the Savage kindly provided me with links to your (as I recall) earlier posts on the forum in which you state that sysex transfers are corrupted if any MIDI messages are received over the other MIDI port (DIN or USB) at any time after powering-on. However, I thoroughly tested this in both combinations (USB messages received before DIN transfer, DIN messages received before USB transfer) and both ways worked perfectly. The only problem I’m aware of is the one identified on the petition, which I’m sorry is not fixed but given the simple workaround of not sending additional MIDI messages to Tempest on the opposite port during a transfer, we felt there were bigger fish that Tempest owners would prefer us to fry at this time.

I do understand your anger because we screwed up in taking so long to fix important bugs. And while we’re late to the game, we’re trying our best now to atone for our past sins. I’m just not sure that the problems you’ve identified are either an actual bug (#1 above) or a reasonable request (#2 above).

If you feel I’ve misunderstood you, you’re welcome to contact me directly at support@rogerlinndesign.com and we can arrange for a call, because a real-time conversation would make it easier for me to understand your point.

[Scoridd]

"2. Note sustain without a duration limit..."

For me, this doesn't work. I hit record then play, then play a sustained note for almost the entire length of the 4 bar sequence. While I'm holding the pad down, the note continues to play for the entire time. But when I play it back via the sequencer, the note is almost always significantly shorter. The actual length that it plays back varies (values I've just recorded 111, 204, 74) - sometime it's around 1 bar, sometimes less - usually never any longer than 2 bars. This is weird because if I go in to the event editor, I can set the duration of that step to a much higher number than was recorded by the sequencer. At maximum duration (254), the note length is nearly 3 bars. It seems like a MIDI off message is received early, at random times.

Note: in the scenario above, I am only playing one pad, one oscillator, open filter/env etc..   basically, it's just a test environment.

Hi Scoridd,

This is explained in my post about the beta download, in which I describe which of the bugs on the petition are fixed or not fixed. Note the last sentence in bold:

2. Note sustain without a duration limit...

Description: This title is not quite accurate. The problem was that sometimes ADSR sounds in Beats would terminate before their full duration expired, or a performed ADSR sound would terminate before you released the pad. This was because Tempest has only 6 voices, so if all voices are playing, new sounds need to steal voices from playing sounds. Tempest does this by stealing the playing voice with the lowest VCA level, but this can result in recorded ADSR sounds with long durations being terminated before their recorded duration ends.

Status: Mostly fixed. Now held ADSR sounds—whether playing from beats or played live—should never terminate before release.
However, with only 6 voices, there will still be cases where a sound will be terminated. To help, the third post in this thread contains a PDF containing suggestions on how to set parameters in order to best optimize the voice-stealing for specific use cases.
Note: a related bug exists that isn't yet fixed: If while recording, you hold a sound pad for more than the maximum 255 ticks (about 2-1/2 bars), the resulting duration will be (correct duration - 255).

Ah OK, gotcha - thanks Roger. You know, now you mention it, I think I had seen this before but I had forgotten the note at the bottom.

[muleskinner]

I am not on the latest beta yet but on the 1.4 but as this is not in the changelog I presume it exists in the latest beta - it's pretty simple for someone to check.

In LFO 'sync' mode there is pretty bad drift with the timing of the LFOs. Set up something obvious such as a one bar sequence at 120bpm. Simple sawtooth oscillator with filter completely closed. Set one of the LFOs to open the filter all the way, ramp up, sync on, timing 4 qtr (ie one bar). Restart to 'play' or 'beat'.

Now play the sequence. It will sound OK at first but let it play for a while and the LFO timing will drift pretty rapidly and start cycling late so the freq of the LFO must be slightly slower than one bar.

The drift also happens when 'LFO restart' is set to 'beat' which I find even stranger - surely in this case it's logical for the LFO to restart when the beat loops (I guess possibly not for slow evolving FX)? Even if the desired behaviour is for the LFO to restart only when changing from another beat then it should at least be restarted on 'play' as well as on changing beat which doesn't appear to be the case either.

It would be nice to get this fixed as not having tempo synced LFOs reliably sync to tempo is not good!

[Roger_Linn]

I am not on the latest beta yet but on the 1.4 but as this is not in the changelog I presume it exists in the latest beta - it's pretty simple for someone to check.

In LFO 'sync' mode there is pretty bad drift with the timing of the LFOs. Set up something obvious such as a one bar sequence at 120bpm. Simple sawtooth oscillator with filter completely closed. Set one of the LFOs to open the filter all the way, ramp up, sync on, timing 4 qtr (ie one bar). Restart to 'play' or 'beat'.

Now play the sequence. It will sound OK at first but let it play for a while and the LFO timing will drift pretty rapidly and start cycling late so the freq of the LFO must be slightly slower than one bar.

The drift also happens when 'LFO restart' is set to 'beat' which I find even stranger - surely in this case it's logical for the LFO to restart when the beat loops (I guess possibly not for slow evolving FX)? Even if the desired behaviour is for the LFO to restart only when changing from another beat then it should at least be restarted on 'play' as well as on changing beat which doesn't appear to be the case either.

It would be nice to get this fixed as not having tempo synced LFOs reliably sync to tempo is not good!

Hi Muleskinner--
Sorry about that. We are aware of that bug and it's on our internal list.

[muleskinner]

Sorry about that. We are aware of that bug and it's on our internal list.

Great - good to know it's being looked at!

cheers

[dsetto]

I believe this thread is historic. I just got a Tempest. It's wonderful. And I am grateful there's an intelligent community & team of makers fine tuning it.

[bozo]

Sorry about that. We are aware of that bug and it's on our internal list.

Great - good to know it's being looked at!

cheers

+1

[muleskinner]

Here's another one - again I'm still on 1.4 as I'm the the middle of a project and don't want to risk updating atm.

- Pads don't always light up correctly in 16 timesteps mode.

The most reliable rep I can get for this is as follows...

1. Go to an empty beat
2. Select 16 timesteps mode with quantize on 1/16
3. Add events for the first four notes using the pads
4. Change quantize to 1/32 - pad lights will space out as expected
5. Select the events screen (you don't have to do this, it just makes it clearer something screwy is going on)
6. Use the (very useful) event copy feature to copy one of the existing events to one of the 'gaps'

...I get the pasted event appearing as expected in the events screen and hear it when playing back, however the pad does not light up on 'paste', neither does it light up when going out of 16 timesteps mode and back. It's like the event is there but not there!

Additionally if, after pasting the event, you click the 'delete step' softkey it actually inserts a step with default parameters before deleting it, again indicating that the step is somehow there but not there.

I have also had it where events can be inserted on the events screen but the relevant timesteps aren't lighting up, however i can't replicate this at the moment.
 
     

[soidthezoid]

Here's another one - again I'm still on 1.4 as I'm the the middle of a project and don't want to risk updating atm.

- Pads don't always light up correctly in 16 timesteps mode.

The most reliable rep I can get for this is as follows...

1. Go to an empty beat
2. Select 16 timesteps mode with quantize on 1/16
3. Add events for the first four notes using the pads
4. Change quantize to 1/32 - pad lights will space out as expected
5. Select the events screen (you don't have to do this, it just makes it clearer something screwy is going on)
6. Use the (very useful) event copy feature to copy one of the existing events to one of the 'gaps'

...I get the pasted event appearing as expected in the events screen and hear it when playing back, however the pad does not light up on 'paste', neither does it light up when going out of 16 timesteps mode and back. It's like the event is there but not there!

Additionally if, after pasting the event, you click the 'delete step' softkey it actually inserts a step with default parameters before deleting it, again indicating that the step is somehow there but not there.

I have also had it where events can be inserted on the events screen but the relevant timesteps aren't lighting up, however i can't replicate this at the moment.
 
   

I can confirm this one - I reported it to DSI some months ago now.  J

[soidthezoid]

Hey Guys,

Do we have a rough estimate of date for next OS update please?  Great work so far, by the way.  Very impressed.

Cheers, J

[Roger_Linn]

Here's another one - again I'm still on 1.4 as I'm the the middle of a project and don't want to risk updating atm.

- Pads don't always light up correctly in 16 timesteps mode.

The most reliable rep I can get for this is as follows...

1. Go to an empty beat
2. Select 16 timesteps mode with quantize on 1/16
3. Add events for the first four notes using the pads
4. Change quantize to 1/32 - pad lights will space out as expected
5. Select the events screen (you don't have to do this, it just makes it clearer something screwy is going on)
6. Use the (very useful) event copy feature to copy one of the existing events to one of the 'gaps'

...I get the pasted event appearing as expected in the events screen and hear it when playing back, however the pad does not light up on 'paste', neither does it light up when going out of 16 timesteps mode and back. It's like the event is there but not there!

Additionally if, after pasting the event, you click the 'delete step' softkey it actually inserts a step with default parameters before deleting it, again indicating that the step is somehow there but not there.

I have also had it where events can be inserted on the events screen but the relevant timesteps aren't lighting up, however i can't replicate this at the moment.

Hi muleskinner--
I have verified this bug and have added it to our internal bug list. Thank you.

[muleskinner]

Hi muleskinner--
I have verified this bug and have added it to our internal bug list. Thank you.
[/quote]

Great - thanks!

[Roger_Linn]

Hey Guys,

Do we have a rough estimate of date for next OS update please?  Great work so far, by the way.  Very impressed.

Cheers, J

Hi soidthezoid--
Sorry for the late reply. DSI hasn't given an estimated release date, but it's definitely being worked on. And I can say that much of the work has already been done in order to create the beta release. Tempest is a complex beast with 9 computers inside, which makes for a lot of surprises with things taking longer than expected, so it's probably best to take the time to get it right. I'm sorry I don't have a more specific answer.

[soidthezoid]

Thanks Roger, that's great news.  I understand completely, I am a Software Engineer myself.  I frequently undertake tasks which would appear to take just an hour or two which end up taking a day or more!  The unforeseen is often around the corner...  Congrats to you and Dave on fantastic lineages.  I am in awe and full admiration.  J

[natrixgli]

I've noticed since updating, the arpeggiator behaves differently depending on if the sequencer is running or not. When it's not running, it seems like notes get lost particularly when in 16 turnings mode and playing overlapping notes. When the sequencer is running, overlapping notes are fine. Also when stopped, if the sound is locked to a voice, the sound is very short, like almost inaudible.

Sorry if this description isn't the best, I can try and elaborate more if needed.

[Roger_Linn]

I've noticed since updating, the arpeggiator behaves differently depending on if the sequencer is running or not. When it's not running, it seems like notes get lost particularly when in 16 turnings mode and playing overlapping notes. When the sequencer is running, overlapping notes are fine. Also when stopped, if the sound is locked to a voice, the sound is very short, like almost inaudible.

Sorry if this description isn't the best, I can try and elaborate more if needed.

Hi natrixgli--
I'm afraid I do need more detail. Perhaps a video is easiest, especially if I can see the settings in the display.
You mentioned "Also when stopped, if the sound is locked to a voice, the sound is very short, like almost inaudible." I assume you're aware that if arpeggiating a chord using only one voice, each new note must cutoff the previous note because there is only one voice, correct?

[natrixgli]

Hi natrixgli--
I'm afraid I do need more detail. Perhaps a video is easiest, especially if I can see the settings in the display.
You mentioned "Also when stopped, if the sound is locked to a voice, the sound is very short, like almost inaudible." I assume you're aware that if arpeggiating a chord using only one voice, each new note must cutoff the previous note because there is only one voice, correct?

Hi Roger -

Thanks for replying (and also for revolutionizing drum machines & inspiring lots of great hip hop/dance music). As I was recording the video it occurred to me this might be expected behavior that I just never noticed before. I think when the sequencer is stopped, the arpeggiator starts every time you hit a pad, whereas when the sequencer is running maybe it's start time is quantized to the beat. Noticing it after the update may be pure coincidence.

Video Link: https://youtu.be/LzsN8qkfIq8

I apologize for the quality of the video, with my limited video capabilities I wasn't able to get a solid capture of the screen. However if you determine this is bug-list worthy, I will be happy to provide more details about settings.

Thanks!

[dsetto]

When I sync a slaved Tempest to my DAW or hardware sequencer via MIDI Port, Playlist doesn't play as programmed when the DAW or hardware sequencer is begun on or after measure 9.

For this observation, Tempest was always configured as Slave, and the DAW's & hardware sequencer were Master. For both DAW's, MIDI Beat Clock offset was configured both aurally & visually. Once the offset was established, sync between Tempest & Pro Tools remains tight, reliable & consistent within the framework of a 5 minute song. I.e., successful sync. I.e., no perceived jitter, drift. And no latency after configuring the offset in Pro Tools. (I didn't test sync beyond 5 mintues. For my brief tests in Logic, I did get tight sync. I didn't bother testing for sync tightness with the hardware sequencer, as I didn't have audio connected.)

Playlist played as programmed when:
- DAW is started at beat 1 or 3 of any measure before measure 9 (via slaved MIDI Port sync)
- Tempest slaved to a master via USB, regardless of where the Master was started
- Tempest is Master

I am brand new to Tempest, but I have read the pertinent parts of the manual, this forum, and the last 2 years of the older Tempest forum. I believe this 'Slaved Playlist' irregulatiry is not as intended. When I synced via USB, the slaved Playlist did play as programmed regardless of where I started the DAW. I am also brand new to drum machines & employing Song Position Pointer to align a linear DAW to a pattern-based machine. I mention this to highlight the possibility of user error. Although, as I've run this by tech support, I no longer think it's user error.

The slaved Playlist irregularity was observed with the following hardware, software, & conditions:
TEMPEST:
- Main OS 1.4 (Voice 1.4)
- Main OS 1.4.0.15b. (Voice 1.5)
- Initialized Project, followed by loading sounds.
- With 16 sequenced Beats sequenced. 2, 4, & 8 measure long. (And also, with 16 empty Beats.)
- Sync setting: Slave. Sync via MIDI Port.

MIDI INTERFACES:
- Digidesign 003 (connects to computer via FireWire 400)
- MOTU MIDI Express 128 (connects to computer via USB)
- MPC 4000

HARDWARE/DAW's
(Most of my testing/observations have been with Pro Tools 9 & Digidesign 003.)
- Macbook Pro Early 2008, Mac OS 10.6 Snow Leopard
- Pro Tools 9 (most tests)
- Pro Tools settings: 120 bpm, 44.1kHz sample rate, H/W Buffer Size: 512 samples & 256 samples.
- Logic Pro 9 (some tests)
- MPC 4000 (some tests)
- All were set to transmit Midi Beat Clock.


---
I am really appreciative to have a Tempest. I am eager to test out the ability to "program" Tempest Beats via Pro Tool's MIDI sequencer. If it works & things remain tight, this will be helpful.

The reasons I would like to be able to have Tempest play from the proper place in playlist when Pro Tools is started on or after measure 9 are:
- I intend to get an external clock that synchronizes to a DAW's audio track to further improve the sync. (It is surprisingly good without it.). In that setup, I may be limited to syncing (as slave) via MIDI Port.
- It is a pleasure to find a Beats arrangement using Tempest' pads in Beats Mode, while playing back the DAW tracks.
- I could imagine it will be very nice to have Tempest record a Beats "arrangement" into a Playlist while in Playlist mode, while playing back to DAW tracks. (I suppose I can do this now, starting Pro Tools from measure 1.)
- It would be nice to be able to start the DAW on or after measure 9, for a 5 minute song, when focuing on a certain section.

---
Thank you Dave, Roger, Chris, Carson, Andy, and Team DSI.

[Roger_Linn]

Hi Roger -

Thanks for replying (and also for revolutionizing drum machines & inspiring lots of great hip hop/dance music). As I was recording the video it occurred to me this might be expected behavior that I just never noticed before. I think when the sequencer is stopped, the arpeggiator starts every time you hit a pad, whereas when the sequencer is running maybe it's start time is quantized to the beat. Noticing it after the update may be pure coincidence.
Video Link: https://youtu.be/LzsN8qkfIq8
I apologize for the quality of the video, with my limited video capabilities I wasn't able to get a solid capture of the screen. However if you determine this is bug-list worthy, I will be happy to provide more details about settings.

Thanks also for your kind words. I try.

Thanks!

Hi natrixgli---
It's a little tricky to see because normally Roll/Arpeggiator is used by holding the pads. But I understand. The problem you're seeing is that if multiple pads are held while the sequencer is stopped, the repeated notes aren't in sync with each other. Yes, this is a bug we're aware of but I thank you for letting me know.

[Roger_Linn]

When I sync a slaved Tempest to my DAW or hardware sequencer via MIDI Port, Playlist doesn't play as programmed when the DAW or hardware sequencer is begun on or after measure 9.

Hi deetto--

Thanks for buying a Tempest. I'm glad to hear that you're enjoying it.

Can you tell me how Playlist doesn't play as programmed when the DAW or hardware sequencer is begun on or after measure 9? For example, is it late or plays the wrong notes or ? A short video would be best.

Also, you mentioned that you'd like to record playlists in real time from a live Beat arrangement, in sync to external MIDI clock. You can already do this. Set System > MIDI Clock to Save. Select 16 Beats, Play List and Events, then press Shift and turn on RT Rec (soft knob 4). Then turn on Record mode and start external sync from anywhere within the song. Tempest will begin recording your live beat arrangement from that location.