Also known as: RALCWI
RALCWI (Robust Advanced Low Complexity Waveform Interpolation) is a low-bitrate speech vocoder developed by CML Microcircuits and sold as a chip-based alternative to DVSI’s AMBE for digital radio.1 It encodes voice at roughly 2000–3600 bps and has been used in some DMR and dPMR handsets and modules, giving radio makers a vocoder they could license and integrate as a dedicated IC without depending on the AMBE supply chain.
How it works
RALCWI, as its expansion states, is a waveform-interpolation coder. It analyses each speech frame into a linear-prediction spectral envelope and a pitch estimate, then represents the excitation as a slowly evolving prototype waveform sampled once per pitch period. Splitting that prototype into slowly and rapidly changing parts lets the model capture both the periodic (voiced) and noise-like (unvoiced) character of the sound. Only the envelope, pitch, gains and the compact excitation description are transmitted; the decoder interpolates between successive prototype waveforms to rebuild a continuous excitation and drives the synthesis filter to recover speech.
The design emphasises low computational complexity and robustness — the “RLC” of the name — so it can run in a small fixed-function IC and tolerate the bit errors of a mobile radio channel. CML packaged it as vocoder chips (and later soft/DSP implementations) presenting a simple digital interface: audio in, a few kbps of voice bits out, ready to hand to the radio’s forward-error-correction and modulator.
In practice
RALCWI’s role, like TWELP’s, was to offer a second source for low-rate digital voice. Because AMBE and AMBE+2 are widely licensed but come from a single vendor, an independent vocoder let some DMR and dPMR product lines ship digital voice without that dependency. The trade-off is interoperability: radios using RALCWI encode a different bitstream than AMBE radios, so at the vocoder level they are not directly interchangeable even when they share the same multi-band-excitation-era rate budget and the same air interface framing.
Relevance to SDR
For a decoder like GopherTrunk, the vocoder matters as much as the modulation. The DMR and dPMR standards themselves define the radio layer, but the voice payload can be AMBE+2 or, in some equipment, RALCWI — and a receiver can only turn voice bits back into audio if it implements the matching vocoder. The bulk of trunked and conventional digital traffic on the air uses the AMBE family, which is what GopherTrunk focuses on; RALCWI-encoded voice from those specific radios would need a separate RALCWI synthesiser to render. GopherTrunk does not implement RALCWI decoding. Knowing it exists explains why occasional DMR/dPMR voice may not decode even when the signalling does.