MIDI Over Ethernet (IP)

There is substantial current activity related to using MIDI over Ethernet (more accurately: MIDI over IP: the Internet Protocol). This includes:

  • Two independent standards efforts (IETF and IEEE).
  • At least one commercial product (NetMIDI) for show control applications.
  • At least one commercial product (MIDIOverLAN+) intended for music applications.

Interest in MIDI over IP is driven by a number of factors. These include:

  • Ethernet technology is cheap, mature, high bandwidth, mass-market.
    An 8-port 100-BaseT switcher costs $40. A PC card (NIC) costs < $10.
    A 3x8 MIDI "switch box" costs $80. (US street price)
  • More and more people are using multiple computers for recording or performance.
    In the near future, audio and synthesis “server farms” (already used at facilities like Hans Zimmerman's Media Ventures) will be commonplace, much as the movie industry has adopted Linux-based graphics server farms as the primary platform for computer-generated effects creation and digital image processing.
  • The potential to solve several long-standing problems of MIDI: bandwidth, routing and addressing, multiple MIDI streams, and the expressive limitations of the MIDI 1.0 model of music representation.

However, MIDI over IP also poses a number of risks. These include potential performance problems, lack of interoperability and potential obstacles to further evolution of the MIDI protocol.

Performance may suffer due to excessive transport jitter, latency or other quality-of-service problems (which can damage the rhythmic integrity of a MIDI stream or provide sluggish response). The MIDI 1.0 transport layer is capable of supporting fairly good quality-of-service characteristics, but not all MIDI systems and products have delivered on this potential. Performance problems could make a particular transport unsuitable for discriminating users. Quality of Service and Temporal Fidelity discusses performance-related issues in more detail.

End users expect different MIDI products to work together, simply and without problems. Historically, this has been a major advantage for MIDI: things just work. Along with increased capabilities, new transports also have many interoperability issues. Incompatible connectors, device addressing schemes, flow control policies, synchronization mechanisms and other factors can all cause problems. Additional quality-of-service problems can occur when bridging between two different kinds of media transports. In fact, quality-of-service problems can occur even when connecting two devices over the same host-based transport, due to the additional host-based processing involved. Configuration issues for multi-transport systems are another area of concern. It is important to maintain the consistency and compatibility that has helped MIDI flourish on many platforms and areas for 20 years.

Finally, architectural issues should be considered to avoid creating obstacles to the further evolution of the MIDI protocol or similar protocols. Individual products must, of course, be able to transform and process MIDI data in various ways in order to deliver desired functionality to their users. However, the underlying transports should not make any assumptions about the allowable byte sequences or bit patterns that may occur within a MIDI stream. The MIDI stream delivered by a transport should be essentially identical to the MIDI stream submitted as input to that transport.

The other pages in this section discuss many of the issues related to MIDI over IP:

  • Overview of current activity
  • Comparisons of current protocols
  • Personal assessments
  • Supplemental Material
    • Node Addressing and UDP Protocols
    • A very brief primer on IP technology
    • Standards Bodies: Who "Owns" What
    • References
          Selected Patents and Prior Art
          Standards and Work-in-Progress
          Tutorial References

We hope to work closely with other parties interested in this area to help produce a good result for the various constituencies involved.