Midi clock/shift/aeros loops

MIDI synchronization is not done correctly. When I sample a loop at 77 bpm, I hear that it shifts little by little from the master. I have the impression that the aeros does not resynchronize each time it restarts the loop with the midi clock.
I have the impression that the loop length is made during the first recording but that after, nothing recalibrates it compared to the midi clock that feeds the aeros.

Wonder how this works with the midi it sends out, does that drift out with the loop too

Here a video to show you the problem of midi sync (and who does not look to midi sync)

As you can see between 0:10 and 4:45, the aeros is offset from the midi synchronization. The midi synchronization does not work.

1 Like

Hi Singular Support,

Can this be prioritized? I can confirm that this is a real bug. Here are specific steps to reproduce it.

Setup:

  1. External MIDI clock
  2. External transport MIDI start
  3. Aeros receives MIDI IN
  4. Mono Synth plays a 16 note sequence through the Aeros
  5. A recording device captures the master audio (from Aeros)

Problem / Steps to Reproduce:

  1. After I record, the playback timing is correct
  2. If I stop the Aeros and press START, while the external sequencer continuously plays, the screen shows that the loop will start exactly on-the-measure, except that the timing is completely incorrect
  3. I can start and stop the Aeros any number of times and the playback timing is always incorrect
  4. If I stop the external sequencer and start it again (sending a new transport MIDI start) the timing is correct again

Why is this Important?
Because I have a lot of gear, including two Boss RC-505s. It is not practical for me to stop the entire system/show if I want to clear or start a new song on the Aeros.

Below is a link to a 1.7MB MP3 file where I follow the steps to reproduce:
[https://drive.google.com/file/d/1YMX1dgYkJz3617Jg-FcjTYU3kKI9TJLy/view?usp=sharing](http://Sample Recording Demonstrating the Timing Bug)

  1. The recording starts with the initially recorded loop
  2. I stop and start the Aeros several times and it sounds like a train wreck
  3. At the end of the recording, I stop the external sequencer then start the external sequencer and subsequently the Aeros - the timing is correct

Best regards,
Eric

Followup: I am running the same test as described above, except this time I had recently started up my studio equipment. The timing is now much more accurate when I stop and start the Aeros. Yesterday, I had been running my sequencer uninterrupted for at least 10 minutes and was able to prove that there is a problem (please listen to my audio recording link in the previous post).

My assumption is that the Aeros is not using the MIDI clock pulse in the correct way, as specified in the MIDI specification, i.e. it is probably using the pulse to count the BPM. If so, this is entirely incorrect and will never produce accurate timing.

  1. The MIDI clock pulse is sent 192 times per second
  2. Every second pulse, the slave device is supposed to not only adjust its BPM, but it is also supposed to align its timing
  3. In terms of MIDI notes, every 2nd clock pulse is supposed to represent a MIDI step frame where notes are aligned (with +/- 5 resolution)

If the Aeros is simply counting the clock pulses to set the BPM then the following will happen.

  1. The first batch of clock pulses will be miscounted and the Aeros will start play at the wrong time
    a) The reason is that a master sequencer uses IO buffers and when play is pressed is not exactly aligned to the clock pulses
    b) The correct behavior is easy to observe with a drum machine
    c) The first few beats are slower and the tempo will quickly speed up to the matching clock synchronization
  2. MIDI Jitter - all hardware, including a DAW will produce timing jitter
    a) The reason for the clock pulse (PPQ) is to align the jitter so that timing among different machines does not become out-of-phase
    b) jitter is typically 1 to 5 milliseconds per machine in the midi chain
  3. The temperature will affect the MIDI clock because a quartz crystal is used

For the reasons stated, if the Aeros is simply capturing a BPM, this will quickly drift from the master MIDI clock - a lot like two vinyl records with the drag of the needle causing drift.


EDIT

I have run another series of experiments and my findings are that the timing is correct and does not drift after reasonable periods of time > 10 minutes. And there is a bug with stop/start timing after the Aeros has been reading a MIDI clock for > 5 minutes.

Because the timing is accurate and consistent, I really don’t care how this was accomplished so what I wrote above is irrelevant.

Experiment Setup:

  1. External MIDI master sequencer
  2. Drum machine for kick and clap
  3. RC-505
  4. Aeros

Steps:

  1. Create a step quantized hi-hat pattern (using Novation Circuit)
  2. Record the hi-hat audio on the RC-505 for 4 measures
  3. Immediately after, record the same hi-hat audio on the Aeros for 4 measures
  4. Play both hi-hat loops from the RC-505 and the Aeros at the same time
  5. Wait… see if the hi-hats go out-of-phase

Aeros Hi-Hat Incorrect Timing - Stop/Start mp3

In my first experiment, I followed the steps above and the loops did not go out-of-phase or miss their timing in any way. I allowed them to play together for greater than five minutes. Next, while the master sequencer had been planing throughout, I pressed STOP on the Aeros and then START. This caused the hi-hats to play at a completely incorrect time, with an obvious double hit because the timing was off by almost a complete step. I hit record on my main recorder to capture this. In the recording you will hear that I press STOP/START several times and I mute/unmute the RC-505 so that the Aeros plays only it’s loop.

Workaround:
Don’t press STOP on the Aeros unless you are prepared to stop your master sequencer.

Experiment 2:
In this round of experiments, I wanted to test whether the Aeros would have correct and consistent timing after a reboot. I used the same setup described above but restarted the Aeros many times and started and stopped the master sequencer many times. The timing was correct. I then played the Aeros hi-hat loop with the same RC-505 loop for more than ten minutes and created a main recording at the end to show that the timing was correct and had not drifted.

Aeros with RC-505 Correct Timing mp3

2 Likes