Hi. I'm working on a vst3 editor for the OB-6 and I'm trying to confirm a specific technical detail before release.
I've got the sequencer working reliably for 5 voices via sysex, but it doesn't seem that all 6 voices are addressable through the sequencer data in the sysex dump.
Does the OB-6's sysex program format allocate space for 6 voices worth of sequencer note data, or does it cap it at a lower number? I can't find a definitive answer, and Sequential support is giving me non-answers (e.g., "No one has had a problem using 6 voices in the 10 years it's been on the market." etc.). Has anyone worked with sysex for the OB-6 or even P6 that could answer?
Could someone share with me a dump of a program that uses all 6 voices? Maybe it's my specific synth that has an issue. Also, it seems like you can record and play 6 voices atop a preset, but that's not the same as saving 6 voices to sysex as part of the preset itself.
Note: It seems like even the Soundtower standalone editor doesn't do 6 voices via sysex, but maybe I'm using it wrong (one reason I'm making one is I dislike the UI of that thing).
Thanks!!
Following up on my own question.
I had Claude Opus Max do a deep technical investigation across the manual, my own hardware (firmware 1.8.0), and ~2,300 programs from third-party and factory sources.
Conclusion: the OB-6 sequencer SysEx stores only 5 voices, not 6, despite the manual and marketing copy claiming otherwise.Reasons:Controlled A/B test on my own OB-6 (FW 1.8.0): Saved a baseline patch with no sequence, then saved the same patch after entering 6-note chords across the sequencer. Diffed the SysEx dumps. The 6th-voice region (bytes 896-1023, the NRPN range the manual lists as "Seq Note/Vel 6") was byte-for-byte identical between the two dumps. The hardware silently dropped the 6th note of every chord.
Bytes 896-1023 are byte-identical across 7 single-program dumps from my own hardware regardless of patch content — a fixed non-sequencer pattern, almost certainly firmware metadata.
Six commercial third-party soundsets (~700 patches), including one specifically built around the OB-6's polyphonic sequencer: zero programs use Voice 6. Some use Voice 5, proving the encoding for non-V1 voices is well-known to designers. They stop at V5 because V6 doesn't store.
Sequential's own 500-patch factory bank: zero use of Voice 6.
My complete 1000-program hardware dump: zero programs use Voice 6.
Firmware release notes from 1.6.6 → 1.7.4 → 1.8.0 describe MPE additions, aftertouch curves, and filter calibration fixes. No sequencer changes since 1.6.6. So this isn't a recent regression and isn't unit-specific to me.
Total dataset: 2,307 programs across 7+ independent sources, on multiple OB-6 units, multiple firmware versions, including a sequencer-focused soundset where any designer would have used V6 if it worked. Zero V6 sequencer data anywhere.
The manual page 95 entries for NRPN 896-1023 ("Seq Step 1-64 Note 6 / Vel 6") and the "up to 6 notes per step" marketing copy both describe an unimplemented spec.
The actual sequencer seems to be 5-voice in storage and 5-note per step in chord input.If anyone from Sequential can confirm this is correct (or correct me if there's a way to reach the 6th voice that I'm missing), it would help me finalize an unofficial editor design decision, specifically whether to hide the V6 row permanently (current path) or keep it visible. Thanks!
Recent Posts
