HF Digital Voice

1
HF Digital Voice
2
Overview

• Current Harris Digital Voice Options
• Advantages Of Digital Voice
• Disadvantages of Digital Voice
• Vocoder Improvements
• Summary

3
Current Digital Voice Options

• 2400 bps LPC-10e based vocoder integrated with
  U.S. MIL-STD-188-110A 2400bps serial tone modem
• Available RF-5022E R/T
• Available RNLA Firm system (RF-5801) SVDT
• Available RF-5811 SVDU
• Available RF-5800H Manpack Radio

4
Current Digital Voice Options

• 600 bps Vector Quantized LPC-10e based vocoder
  integrated with U.S. MIL-STD-188-110A 600bps
  serial tone modem
• Available RF-5022E R/T
• Available RNLA Firm system (RF-5801) SVDT
• Available RF-5811 SVDU
• Available RF-5800H Manpack Radio

5
Current Digital Voice Options

• 600 bps Vector Quantized LPC-10e based vocoder
  integrated with Harris 600bps Serial Tone ECCM
• Available RF-5022E R/T
• Available RNLA Firm system (RF-5801) SVDT
• Available RF-5800H Manpack Radio

6
Advantages Of Digital Voice

• High levels of security achievable with digital
  encryption
• Sound quality mostly immune to fading and
  interference common to HF communications

7
Disadvantages Of Digital Voice

• Lower rate vocoders have noticeable voice quality
  degradation
• Delays introduced by the vocoder / Encryption
  hardware / and HF Modem waveform can add up to
  several seconds
• Vocoder quality can be degraded by background
  noise sources ( Vehicle noise etc.)

8
Improvements

• Harris currently has an IRD task funded to
  examine An Improved vocoder based on MELP

9
Improvements - MELP

• MELP - Mixed-Excitation Linear Prediction - A
  mixed-excitation model that can produce more
  natural sounding speech
• The MELP vocoder is robust in difficult
  background noise such as those frequently
  encountered in military communications systems
• The voice model can simulate frequency dependent
  voicing strength using adaptive filtering based
  on a fixed filterbank
• The multi-band mixed-excitation reduces the buzz
  usually associated with LPC by matching the vocal
  tract better (less residual prediction error)

10
Improvements - MELP

• Pulse dispersion filter spreads the excitation
  energy to reduce the harsh quality of synthetic
  speech
• Aperiodic pulses used as excitation allows the
  synthesizer to reproduce erratic glottal pulses
  without introducing tonal noise (between
  voice/unvoiced segments)
• Adaptive spectral enhancement filter uses the
  poles of the LPC vocal tract filter to improve
  the formant structure in the synthetic speech
• MELP 2400 BPS DRT score 94.0, DAM score 69
  (Current Harris LPCe based 600 DRT 86.8, DAM
  47.1)

11
Improvements - MELP

• Harris MELP Plans
• Fund Advance Development to develop a MELP based
  600bps vocoder (Goal improve DRT and DAM scores
  by 10 points each)
• Port 2400 BPS MELP model to MPR
• Transcode MELP 2400 down to 600
• Evaluate techniques that meet the DRT/DAM score
  improvements
• Develop techniques to improve vocoders tolerance
  to noise
• New vocoder will be fielded only after
  significant improvements of DRT and DAM are
  achieved

12
Summary

• Digital Voice offers significant advantages in
  security and performance in HF systems
• Harris is currently improving its digital voice
  offerings by Investigating higher quality
  vocoders
• Paper will be presented at MILCOM 2001 by Mark
  Chamberlain Discussing the new Harris 600bps
  version Of MELP