Multi Channel Audio Digital Interface, defined in 1989 and introduced in 1991, the AES10 standard defines the digital transmission of more than two channels of linearly represented digital audio data in a professional audio environment. It is actually an extension of the AES3 standard. The MADI format is capable of transmitting 28 AES/EBU channel pairs (56 digitally encoded audio signals) on a standard 75ohm coax or fibre cable with a sample rate of 44.1 or 48kHz. MADI is a low jitter and sample accurate format, thus ensuring the maximum sound quality for digital audio environments.
Each MADI channel consists of 32 bits, 24 bits are allocated to audio or to other data as defined by the audio/nonaudio status flag, four further bits represent the validity (V), user (U), status (C), and parity (P) bits of the two-channel AES3 interface, and four more bits are allocated for mode identification.
The MADI standard was initially developed by AMS Neve, Solid State Logic (SSL), Sony and Mitsubishi.
The MADI interface can also be configured with a capacity of up to 64 channels, sample rates are 32 to 96kHz. The 64 channel mode was introduced officially in 2001. It allows for a maximum sample rate of 48 kHz, corresponding to 32 channels at 96 kHz, without exceeding the maximum data rate of 100 Mbit/s.
MADI is a uni-directional transmission and provides a ‘point to point’ interface between a source and a destination. The 56 audio channels are multiplexed with Time Division Modulation (TDM). The overall data rate is 125 Mb/s. Maximum cable length for coax cable: 50 meter.
Glasfiber transmission: fibre interfaces should be used as specified according to ISO/IEC 9314-3, graded-index fibre with core diameter of 62,5 nm, nominal cladding diameter of 125 nm and a numerical aperture of 0,275, wavelength of 1300 nm, connector ST1 (ISO/IEC 9314-3). This specification can provide a range of up to 2 km.
MADI can also be used to transmit audio signals with higher sampling rates (ie. 96kHz). There are two alternate methods specified for higher sample rate, but in both methods the number of audio channels is reduced. For 96kHz it is reduced to 26 mono channels.