This page is read from MIDI Instrument Definitions, a Python library used in the author's other projects, in which every number is traced to a document its maker published. Its source is on GitHub.

Sequential Prophet-10 MIDI implementation

The MIDI implementation of the Sequential Prophet-10: what it answers to, and the 85 controls it offers, on firmware 2.1.0.

At a glance

Firmware

This page describes firmware 2.1.0.

MIDI channel

Any one of 1 to 16.

Voices

10.

Note range

Not recorded.

Velocity

Sent and received. Release velocity is ignored.

Aftertouch

Channel pressure and polyphonic, each answered. It sends aftertouch too.

Pitch bend

12 semitones, and the range can be changed.

Program change

Received and sent, with 400 presets.

MIDI clock

Not recorded.

Start and stop

Not recorded.

Control change

64 controls, in the tables below.

NRPN

Supported, and preferred to CC. 71 of the controls below have one.

System exclusive

Supported.

Controls

In the order the definition gives them, by group. Under each control is the name a program calls it by, once pymidiinstrumentdefs.load("sequential/prophet_10") has loaded them.

OSC A

OSC B

POLY MOD

LFO

FILTER

ENVELOPE

UNISON

PRESSURE

Performance

Global

Global parameters

Where these facts come from

Source documents

• Prophet-5 MIDI Implementation, 1.4, dated 11 March 2021, its MIDI implementation, read on 7 October 2026 (the file read).
• Prophet-10 User's Guide, 1.4, dated 24 February 2021, its user manual, read on 7 October 2026 (the file read).
• Prophet-5/10 OS 2.0 Addendum, 2.0, dated 7 September 2021, its firmware addendum, read on 7 October 2026 (the file read).
• Prophet-5/10 OS 2.1.0 Addendum, 2.1.0, dated 3 May 2025, its firmware addendum, read on 7 October 2026 (the file read).
• Prophet-5/10 Operating System - Sequential, its download page, read on 7 October 2026.
• Prophet-5/10 Documentation - Sequential, its download page, read on 7 October 2026.
• Prophet-5 User's Guide, 1.3, dated 24 February 2021, its user manual, read on 4 October 2026 (the file read).

How they were read

Prophet-5 MIDI Implementation, version 1.4, 13 pages, which Sequential offers under the Prophet-10 as well: the globals that gate everything on p. 1, the received messages on p. 2, the transmitted messages on p. 3, the 59 controller numbers on pp. 4 and 5, the NRPN format on p. 6, the NRPN tables on pp. 8 and 9, and the system exclusive layouts on pp. 10 to 13. Prophet-10 User's Guide 1.4, 80 sheets: the instrument on p. 1, the rear panel on p. 3, the USB port on p. 4, the programs on p. 6, the globals on pp. 13, 14 and 16, the pitch wheel on p. 37, the aftertouch on p. 38, unison on p. 40 and voice assignment on p. 45. Prophet-5/10 OS 2.0 Addendum, for the layers it adds, and OS 2.1.0 Addendum, for what is current. Sequential's operating system page and documentation page, saved as text. Prophet-5 User's Guide 1.3, cited for one thing only, which is what the Prophet-10's guide says that it does not.

ONE DOCUMENT FOR TWO INSTRUMENTS, AND THREE THINGS SAY SO RATHER THAN THE DOCUMENT ITSELF. THE IMPLEMENTATION NEVER NAMES THE PROPHET-10 - swept over all thirteen pages, with the text squashed so that no hyphen or line break could hide it, and it is not there; nor is a layer, a split or a stack. What settles it is outside it. The documentation page has two sections and offers the same "MIDI Implementation 1.4" (listing) in both, the same file by its address. The Prophet-10's guide says "For a list of Prophet-10 CCs and NRPNs, see the Prophet-10 Support page at Sequential.com" (guide p. 13), and that page is where the file is. And the operating system is one release for both: "the latest operating system for the Prophet-5 and Prophet-10 keyboards and desktop modules" (os).

AND THE IMPLEMENTATION'S UNISON ROWS COUNT TO TEN. UNISON VOICE COUNT is printed 0-10 on p. 5, eleven values, and the NRPN table gives UNISON NOTE a span of ten numbers on p. 9 - on a document named for a five-voice instrument. Unison on this one stacks "as many as ten voices" (guide p. 40). The implementation does not say what each value means, and nothing here says it either.

SIXTY-FOUR CONTROLLER NUMBERS, FROM TWO TABLES THAT DO DIFFERENT JOBS. Fifty-nine are the parameter map, "Additional Continuous Controllers Transmitted/Received" (pp. 4-5), each with the maker's own name and its own range. Five more are performance controllers that the received and transmitted message tables name instead (pp. 2-3): the mod wheel, the breath controller, the foot controller, bank select and the damper pedal.

THE TWO MESSAGE TABLES ARE TWO LISTS, SO A NUMBER IN ONE AND NOT THE OTHER IS ONE-WAY. Measured over both: only the breath controller is, "Breath Controller: When assigned to Pedal/CV" (p. 3), transmitted and never received. Everything else that appears at all appears in both.

THE LABELS ARE THE MAKER'S OWN WORDS IN THE MAKER'S OWN CAPITALS, TYPOS INCLUDED. Controller 15 is printed "OSC A SAW ON/FF" (p. 4), which is ON/OFF with a letter missing - and it is printed that way in the NRPN table as well (p. 8), so it is what this maker calls the thing rather than a slip in one place. Nothing here tidies it.

AND NRPN IS THE PREFERRED METHOD, WHICH THE MAKER SAYS IN BOTH DOCUMENTS. "NRPNs are the preferred method of parameter transmission, since they cover the complete range of all parameters, while CCs are limited to a range of 128" (guide p. 13), and the implementation says the same on its first page. "Param Xmit: Off, CC, NRn" (guide p. 13) and "Param Rcv: Off, CC, NRn" (guide p. 14) are globals, so a parameter change is CC or NRPN or nothing, and never both at once. The controller table says the condition itself: "They are transmitted when Param Xmit is set to CC, and recognized/received when Param Rcv is set to CC" (p. 4).

THE NRPN TABLE IS CARRIED, AS IN THE PROPHET-5'S DEFINITION, AND AN NRPN SITS BESIDE A CONTROLLER ONLY WHERE THE MAKER PRINTS THE SAME NAME IN BOTH TABLES. It holds 71 numbered parameters and three more given as spans, from 0 to 4109 (pp. 8-9), and the controller table and the NRPN table number the same parameters differently - controller 3 and NRPN 0 are both OSC A FREQUENCY. So 50 controls carry both numbers, exactly where the two tables print the same name, and what the two tables agree about is every range those 50 share: not one differs. Seven more NRPNs sit in the places the controller table gives to rows of other names - AFTERTOUCH > FILTER where it has PRESSURE FILTER, and six like it - and are carried as controls of their own, because pairing two different names is a reader's judgement and not the maker's, and paired by position one of them would disagree with its controller about a destination: the controller table has PRESSURE FILTER and PRESSURE LFO (p. 5), the NRPN table AFTERTOUCH > FILTER and AFTERTOUCH > AMP (p. 8), and the guide's aftertouch switch chooses "the Filter cutoff and/or the LFO amount" (guide p. 38). The fourteen global parameters are controls of their own too, 85 controls in all, field for field the Prophet-5's. The three spans are not carried: UNISON NOTE is one row over ten numbers, 55-64 (p. 9), and the other two are RESERVED.

TWO NUMBERS ARE REFUSED BY THE VALIDATOR AND ARE LEFT OUT RATHER THAN DROPPED SILENTLY. Controller 123 is "All Notes Off: Clear all MIDI notes" (p. 2) and controller 121 is "Reset All Controllers: Clears all MIDI controllers to 0, MIDI volume to maximum" (p. 2). Both are received; both are channel mode messages, which this corpus does not record as controls.

AND ONE ROW IS LEFT OUT BECAUSE THE DOCUMENT SAYS TWO THINGS ABOUT IT. The transmitted table's fourth controller row is named "Expression: When assigned to Pedal/CV" (p. 3) and its second byte is printed 0000 1101, which is 13. Expression is controller 11. There is no third statement anywhere to settle which the maker meant - the received table has no Expression row at all - so neither number is recorded.

THE DOCUMENT HAS SIX MORE DEFECTS AND NONE OF THEM IS RESOLVED HERE, because resolving one would be inventing a number the maker has not published.

The first is a second typo in a parameter name: NRPN 41 is printed "VELOCTIY > FILTER" (p. 8) where NRPN 42 beside it spells the word correctly.

The second is the one a reader would actually be hurt by. This instrument's system exclusive identity is given two ways. The device inquiry reply gives 0011 0011 on p. 10 as the family byte; the three request and dump chapters give 0011 0001 on pp. 11, 12 and 13, in nine separate rows. 51 against 49, and nothing in the document says which - nor whether the Prophet-10 answers with either.

The third: the RPN reset message is given the wrong controller numbers. The received NRPN chapter prints 0010 0101 and 0010 0100 on p. 7 for the registered parameter number's two halves, which are 37 and 36. They are 101 and 100. Every NRPN controller number on the same two pages is right, so it is these two rows and not the chapter.

The fourth: nothing in the document says how to save the edit buffer. The table headed Save Edit Buffer (p. 12) is byte for byte the Request Program Dump of the page before, command byte and all.

The fifth: a program is addressed three incompatible ways. "Bank Number, 0 - 7" (p. 11) with "Program Number, 0 - 127" (p. 11); "Group Number: 0 - 9" (p. 12) with "Program Number: 0 - 39" (p. 12); and the received program change's own banks on p. 2.

The sixth: the NRPN tables are headed "Control NRPN Data" (p. 9) and introduced with "It is received and transmitted but not saved as part of a program" (p. 9) - over the parameters the document later says are saved, "The 128 'packed' parameter bytes in the program dump follow the order of the NRPN list" (p. 13).

THE SYSTEM EXCLUSIVE CHAPTER IS FOUR LAYOUTS, not a mention. A universal device inquiry, a request program dump, a program data dump and a global parameters data dump, each byte by byte (pp. 10-13), and a global chooses the port they travel on.

TEN VOICES, AND NINE WITH A PLUG IN THE GATE IN JACK. "The Prophet-10 is a ten-voice, polyphonic analog synthesizer with analog voltage-controlled oscillators, filters, and amplifiers" (guide p. 1). And "Whenever a ¼ inch connector is inserted into the rear-panel gate in jack on the Prophet-10, its keyboard polyphony is reduced to 9 voices" (guide p. 3), which no field here can make depend on what is plugged in.

THE GUIDE IS THE PROPHET-5'S EDITED, AND SOME SENTENCES WERE NOT. The gate in jack gates "a single voice on the Prophet-5" (guide p. 3); MIDI In and USB together "may cause the Prophet-5 to respond unpredictably" (guide p. 4); the pedal mode "Sets the Prophet-5's response to an expression connected to either the amp jack or filter jack" (guide p. 15), and "Either type can be used with the Prophet-5" (guide p. 16). Osc B in lo freq mode "creates five separate modulation LFOs since Osc B on each of the Prophet-10's ten voices" (guide p. 34) is the edit made halfway through a sentence. And unison is "as many as ten voices" (guide p. 40) in one chapter and "(now you can stack 1-5 voices)" (guide p. 45) in the next. The ten is what the instrument does and the rest is the Prophet-5's guide showing through.

SO WHICH VOICE THE GATE IN JACK TAKES IS NOT SETTLED. "The envelopes of the fifth voice are triggered by the gate in signal" (guide p. 3), and "voice 5 is no longer available to play from the keyboard" (guide p. 55). Both are the Prophet-5's guide word for word, so whether the Prophet-10's jack takes its fifth voice or the sentences were never edited is not something the documents can say.

SINCE OS 2.0 IT IS BI-TIMBRAL, AND NOTHING SAYS HOW MIDI REACHES THE SECOND LAYER. "With this OS, the Prophet-10 (or a Prophet-5 with a voice expansion card) becomes bi-timbral and allows you to play two different sounds/programs at once" (addendum_2_0 p. 1), as "Layer A and Layer B" (addendum_2_0 p. 2). Stacked, "the 10-voice Prophet-10 functions as a 5-voice instrument" (addendum_2_0 p. 3); split, "with five voices available on Layer A and five voices on Layer B" (addendum_2_0 p. 5); and a global "limits available polyphony to five voices" (addendum_2_0 p. 8) on request. The addendum's only mention of MIDI is the cable an owner updates the operating system through, and so is the 2.1.0 addendum's: not a channel, not a controller, not a program change. The implementation is six months older than the layers. So no part is recorded here, because a part is something a sender addresses, and nothing published says how.

IT RECEIVES BOTH KINDS OF PRESSURE AND ITS OWN KEYBOARD SENDS CHANNEL PRESSURE, which the two fields and this paragraph say. The received table has "Polyphonic Key Pressure" (p. 2) and "Channel Pressure" (p. 2); the transmitted table has "Channel Pressure" (p. 3) and no polyphonic row at all. And "The Prophet-10 provides monophonic (or 'channel') aftertouch, which means that applying pressure to any key within a chord will apply modulation to all notes currently held" (guide p. 38). aftertouch: both records what it answers to and aftertouch_transmits: true what it sends, and this paragraph is the rest of it.

VELOCITY ARRIVES AND IS IGNORED ON NOTE OFF. "Note Off. Velocity is ignored" (p. 2) and "Note On. Note off if vvvvvvv = 0" (p. 2), and what it transmits on note off is a literal 0000000 (p. 3). Two global curves shape what the keyboard produces, "Vel Response: 0-6" (guide p. 16) and "AT Response: 0-7" (guide p. 16), and neither says anything about what an incoming velocity does.

THE BEND RANGE IS PER PROGRAM AND THE PANEL AND THE CONTROLLER COUNT IT DIFFERENTLY. "A setting of 1 equals a semitone. A setting of 12 equals an octave" (guide p. 37). Controller 70 is "PITCH WHEEL RANGE" (p. 5) with a range of 0-11 - twelve values either way, by one, and the implementation does not say that the controller's 0 is the panel's 1.

FOUR MORE RANGES THE GUIDE GIVES DIFFERENTLY, CARRIED AS THE IMPLEMENTATION PRINTS THEM. Controller 114 is UNISON DETUNE over 0-7 (p. 5), eight values, and the guide says "A setting of 0 is minimum detuning. A setting of 8 is maximum detuning" (guide p. 40), nine. Controller 31 is RESONANCE over 0-120 (p. 4), and the guide's example of the pot modes says "the resonance parameter has an internal value range of 0 to 127" (guide p. 14). And two of the global NRPNs: VEL RESPONSE is printed over 0-7 (p. 9), eight values, where the guide's global is "Vel Response: 0-6" (guide p. 16), seven; and ALT TUNINGS over 0-15 (p. 9), sixteen, where the guide offers normal tuning and sixteen more (guide p. 16). The ranges below are what a message carries, by the implementation's word; neither document says how the other's numbers map onto them.

FOUR HUNDRED PROGRAMS, AND A PROGRAM CHANGE REACHES FORTY OF THEM. "The Prophet-10 contains a total of 400 programs. 200 are permanent 'Factory' programs and 200 are 'User' programs" (guide p. 6). The implementation's received table gives "Program change, 0-39" (p. 2) and bank select picks the rest: "Bank Select: 1 - 5 select user banks 1 - 5; 6 - 9 select factory banks 1 - 4" (p. 2). And that program change row contradicts itself: after the 0-39 it enumerates four spans of eight, naming Banks 1, 2, 4 and 5 - thirty-two values across four banks, where the number before it is forty and the bank selector cycles through five. 40 is the figure the row states outright, and nothing here resolves the rest.

AND BANK SELECT NAMES FIVE USER BANKS AND ONLY FOUR FACTORY ONES, where the guide's selector "increments and cycles through Groups 1-5" (guide p. 6) in either set of 200. As printed, no bank select value reaches a fifth factory group, and the transmitted table says only "Bank Select: 1 - 5" (p. 3). The program data dump does reach one, in the guide's own word: "Group Number: 0 - 9" (p. 12) with "Program Number: 0 - 39" (p. 12), ten groups of forty. Neither document says that a bank select value is a group.

THE USB PORT IS NAMED THREE WAYS. "Under Mac OS, 'Prophet-5/10' will appear as a MIDI port" (guide p. 4), and under Windows the same page says it is installed as Prophet-10.; the operating system page says "a MIDI port named 'Prophet-5' or 'Prophet-10'" (os). Which a given computer shows is not something these documents settle.

NOT RECORDED: the clock and the transport. The words clock, start, stop, continue, song position and active sensing appear nowhere in the thirteen pages of the implementation - not as a no, as nothing at all. This instrument has no sequencer and no arpeggiator.

NOT RECORDED: anything the two addenda add beyond the layers - a high note priority, a layer link, per-program velocity and aftertouch amounts and filter Q compensation (2.1.0), and poly unison, a five-voice limit and round robin allocation (2.0). Neither addendum gives a number for anything it adds, so the tables above are what is published and are known to be incomplete for the instrument as it ships. One of them visibly so: the guide names four key priorities, "low, last, low retrigger, last retrigger" (guide p. 45), and controller 71, RETRIGGER AND UNISON ASSIGN, has four values (p. 5) - which value is which is not said - and NRPN 87, RETRIGGER AND UNISON, the same four (p. 9). 2.1.0 adds two more priorities, high note and its retrigger: "High Note gives note priority to the highest note played, with a retrigger option" (addendum_2_1_0 p. 3), which no value reaches.