Field Guide · technology

Also known as: AMBE+2

AMBE+2 is the efficient successor to IMBE and AMBE from DVSI.1 It powers P25 Phase 2, DMR, and NXDN, and supports half-rate operation — the feature that lets two conversations share the airspace one analog channel used to occupy.

slot 1 · AMBE+2 slot 2 · AMBE+2 one 12.5 kHz channel half-rate coding fits two voices where one used to go
AMBE+2 is a more efficient successor used by P25 Phase 2, DMR, and NXDN, supporting half-rate streams.

How it works

AMBE+2 keeps the Multi-Band Excitation analysis core — pitch, per-band voicing, harmonic magnitudes on a 20 ms frame — but refines the parameter quantization and forward error correction so that audio quality at a given bit rate is noticeably better than IMBE, or equivalently that the same quality survives fewer bits. That efficiency is what enables half-rate modes: DMR’s per-slot vocoder runs about 2450 bps of voice + 1150 bps FEC (3600 bps total), and P25 Phase 2 uses a 7200×4400 half-rate configuration. Two such streams interleave in the two TDMA slots of one 12.5 kHz channel, doubling call capacity without new spectrum.

Variants

AMBE+2 is configurable: DVSI ships it as a family of “rate tables” so a system designer picks a total bit rate and FEC split to fit the channel. DMR, NXDN, and P25 Phase 2 each specify a different configuration, so their frames are not directly interchangeable even though they share the same codec engine (the AMBE-3000 chip). It is the third generation of the MBE line — IMBE → AMBE → AMBE+2 — and, like its predecessors, is not bit-compatible with them.

In practice

AMBE+2 is proprietary and license-gated; radios embed DVSI silicon and early PC decoders required a USB dongle. The core patents underpinning independent software decoding have matured enough for open reimplementations, which is what a modern SDR relies on. Where an open, royalty-free path is required — as in M17 — designers reach for Codec 2 instead of AMBE+2.

Relevance to SDR

GopherTrunk runs AMBE+2 decoding to produce audio from the most common modern digital voice systems: DMR (Tier II/III), NXDN, and P25 Phase 2. It parses the per-slot frames, unwinds the FEC, and synthesizes voice in pure Go. Encrypted traffic (for example DMR’s RC4 Enhanced Privacy or P25 AES) still produces AMBE+2 frames, but their payload is scrambled and cannot be resynthesized to intelligible audio without the key.

Sources

  1. Multi-Band Excitation — Wikipedia, on the MBE vocoder family that includes AMBE+2. ↩

See also

Learn it step by step

From the blog