MELP (Mixed-Excitation Linear Predictive) is a speech coding algorithm standard designed for low bandwidth, secure voice communications. The MELP vocoder was selected for MIL-STD 3005 and later, as a NATO STANAG 4591.
VOCAL’s extensive background in the design and implementation of mobile, radio, and telephony solutions will help you meet your low bandwidth and low power needs. The use of MELP 2400 bps with RTP and SDP is described in detail in RFC8130 – RTP Payload Format for MELPe Codec.
VOCAL’s MELP codec source code (C and assembly) is optimized for performance on DSPs and conventional processors from TI, ADI, ARM, Intel and other leading vendors. MELP is available either standalone, as a library, or with a VoIP stack that can be easily integrated with your application. Custom designs are also available to meet your specific needs. Contact us to discuss your voice application performance requirements.
- MIL-STD-3005 and STANAG 4591 compliant
- Performance exceeds reference implementation
- Quality verified through interoperability and proprietary test vectors
- Scalable real-time multi-channel implementation
- 2400 bps with optional post-filter
- Customizable configuration for optimal memory requirements
- Optimized source code (C and Assembly)
- Mobile VoIP and secure voice
- Ground mobile and software defined radio
- Joint Tactical Radio System (JTRS)
- JTRS Handheld, Manpack and Small Form Fit (HMS)
- Soldier Radio Waveform (SRW)
- Compatible with TSVCIS wideband coder and narrowband TSVCIS waveforms
- UAV VoIP Gateways
- PDAs, tablets and other mobile devices
The MELP vocoder evolved from improvements and modifications to another code excited linear predictive (CELP) coder known as LPC-10. An enhanced version of the MELP algorithm is available, the mixed-excitation linear predictive enhanced (MELP enhanced or MELPe) vocoder.
The MELP voice compression algorithm specifies the 2400 bps mode of operation with an optional postfilter available for the decoder. The MELP frame size is 22.5 ms in duration with each frame containing 180 digital voice samples at a sampling rate of 8,000 Hz. Recommended analog requirements are for a nominal bandwidth ranging from 100 – 3800 Hz. Although the MELP encoder audio compression algorithm can operate with a limited low bandwidth signal, there is some degradation in performance.
MELP encoding is robust in difficult acoustic environments with significant background noise and reverberation such as those frequently encountered in commercial and military communication systems. The MELP encoder speech compression algorithm is very efficient in its computational requirements. This translates into relatively low power consumption, an important consideration for portable systems.
The MELP decoder uses extensive lookup tables and models of the human voice to extract and regenerate speech. The MELP algorithm is tuned to regenerate the English language and, as a result, speakers of non-Germanic languages generally rate MELP speech intelligibility more poorly than English speakers.
MELP vocoder audio quality is achieved using a mixed-excitation model that can produce more natural sounding speech than traditional pitched-excited LPC speech coders because the MELP algorithm can represent a richer ensemble of possible speech characteristics. The LPC vocoders use either a periodic pulse train or white noise as the excitation for an all-pole synthesis filter. These vocoders do produce intelligible speech at very low bit rates, but they sometimes sound mechanical or buzzy and are prone to annoying thumps and tonal noises. These problems arise from the inability of a simple pulse train used in LPC algorithms to reproduce a complete range of voiced audio.
Tactical Secure Voice Cryptographic Interoperability Specification (TSVCIS) started with MELPe and introduces the Wideband Waveform, which provides the framework for improved speech intelligibility, security, and robustness of defense radio communications. This newer codec is designed to be backward compatible with MELPe.
- DAA interface using linear codec at 8.0 kHz sample rate
- Direct interface to 8.0 kHz PCM data stream (A-law or μ-law)
- North American/International Telephony (including caller ID) support available
- Simultaneous DTMF detector operation available – (less than 10 talkoff hits on Bellcore test tape set)
- MF tone detectors, general purpose programmable tone detectors/generators available
- Data/Facsimile/Voice Distinction available
- Common compressed speech frame stream interface to support systems with multiple speech coders
- Dynamic speech coders selection if multiple speech codecs available
- Can be integrated with Acoustic Echo Canceler, G.168 Line Echo Canceller and Tone Detection/Regeneration modules
- Multiple ports can be executed on a single DSP
- Available with VoIP stack