Speech Coding Techniques

1
Speech Coding Techniques
These slides are adapted from the following
book Daniel Collins as well as lecture notes
from Professor Ai-Chun Pang of National Taiwan
University Please send any comment,
enhancements, corrections to Frank.Lin_at_sjsu.edu
THANK YOU
2
Introduction

• Efficient speech-coding techniques
• Advantages for VoIP
• Digital streams of ones and zeros
• The lower the bandwidth, the lower the quality
• RTP payload types
• Processing power
• The better quality (for a given bandwidth) uses a
  more complex algorithm
• A balance between quality and cost

3
Voice Quality - MOS

• Voice quality is subjective
• MOS, Mean Opinion Score
• ITU-T Recommendation P.800
• Excellent 5
• Good 4
• Fair 3
• Poor 2
• Bad 1
• A minimum of 30 people
• Listen to voice samples or in conversations
• P.800 recommendations
• The selection of participants
• The test environment
• Explanations to listeners
• Analysis of results
• Toll quality
• A MOS of 4.0 or higher

4
Voice Quality - PSQM

• Perceptual Speech Quality Measurement
• ITU-T Recommendation P.861
• Algorithm approach
• Compare coding output against known input
• type of speakers (male, female, child)
• loudness
• delay
• silence ratio
• clipping
• environmental noise
• PSQM score can be converted to MOS

5
About Speech

• Speech
• Air pushed from the lungs past the vocal cords
  and along the vocal tract
• The basic vibrations vocal cords
• The sound is altered by the disposition of the
  vocal tract ( tongue and mouth)
• Model the vocal tract as a filter
• The shape changes relatively slowly
• The vibrations at the vocal cords
• The excitation signal

6
Speech sounds

• Voiced sound
• The vocal cords vibrate open and close
• Quasi-periodic pulses of air
• The rate of the opening and closing the pitch
• Unvoiced sounds
• Forcing air at high velocities through a
  constriction
• Noise-like turbulence
• Show little long-term periodicity
• Short-term correlations still present
• Plosive sounds
• A complete closure in the vocal tract
• Air pressure is built up and released suddenly

7
Sampling Quantization

• Sampling rate is based on desired frequency
  range/cutoff
• 300-3800 Hz human speech
• Nyquists theorem gt Sampling rate 8K
• Quantization how many bits per sample
• quantization noise (more bits gt less noise)
• uniform quantization
• favors loud speaker (11.2 11)/11.2 1.8, (2.2
  2)/2.2 9
• SNR is better for loud speaker
• non-uniform quantization
• smaller quantization steps at smaller signal
• try to achieve uniform SNR

8
Types of Speech Coders

• Waveform codecs
• source codecs (a.k.a vocoders)
• model vocal tract, model parameters are sent
• hybrid codecs

high Q simple large BW
low bit rate low Q
9
G.711

• The most commonplace codec
• Waveform coder
• Used in circuit-switched telephone network
• PCM, Pulse-Code Modulation
• If uniform quantization
• 12 bits 8 k/sec 96 kbps
• Non-uniform quantization
• 64 kbps DS0 rate
• u-law
• North America
• A-law
• Other countries, a little friendlier to lower
  signal levels
• An MOS of about 4.3

10
ADPCM(adaptive differential PCM)

• Waveform codec, no algorithmic delay, relative
  high BW
• Each PCM sample is independent of each other
• DPCM differential PCM
• assumption voice changes slowly
• predicts the next sample and only sends diff
  between actual and prediction (quantization
  error)
• simplest DPCM no prediction, just send diff of N
  and N1
• ADPCM adaptive differential PCM
• Adaptive Prediction based on past samples
• Adaptive Quantization not fixed bits
• G.721 ADPCM speech at 32Kbps.
• G.726 (A-law or u-law)
• 16,24,32,40Kbps
• MOS 4.0 , at 32Kbps

11
Analysis-by-Synthesis (AbS) Codecs

• Hybrid codec
• Fill the gap between waveform and source codecs
• The most successful and commonly used
• Time-domain AbS codecs
• Vocal tract prediction filter model same as LPC
  vocoder
• Not a simple two-state, voiced/unvoiced
• Different excitation signals are attempted
• Closest to the original waveform is selected
• MPE, Multi-Pulse Excited (first AbS codec ..
  1982)
• RPE, Regular-Pulse Excited
• CELP, Code-Excited Linear Predictive

12
G.728 LD-CELP

• CELP codecs
• A filter its characteristics change over time
• A codebook of acoustic vectors (1024 vectors)
• A vector a set of elements representing various
  char. of the excitation
• Transmit
• Filter coefficients, gain, a pointer to the
  vector chosen
• Low Delay CELP
• Backward-adaptive coder
• Use previous samples to determine filter
  coefficients
• Operates on 5 samples at a time
• Delay lt 1 ms (sample rate 8K, 125us 5 .625
  ms)
• Only the vector pointer (10 bits) is transmitted
  for every 5 samples
• G.728 LD-CELP rate is 16K bps (8K/510 16K)
• MOS score 3.9
• Process intensive

13
G.723.1 ACELP (algebraic)

• 6.3 or 5.3 kbps
• Both mandatory
• Can change from one to another during a
  conversation
• The coder
• A band-limited input speech signal
• Sampled at 8 KHz, 16-bit uniform PCM quantization
• Operate on blocks of 240 samples at a time
• A look-ahead of 7.5 ms
• A total algorithmic delay of 37.5 ms other
  delays
• A high-pass filter to remove any DC component

14
G.723.1 Annex A

• Silence Insertion Description (SID) frames of
  size four octets
• The two lsbs of the first octet
• 00 6.3kbps 24 octets/frame
• 01 5.3kbps 20
• 10 SID frame 4
• MOS 3.8
• At least 37.5 ms delay

15
G.729, G.729A, G.729B

• 8 kbps
• Input frames of 10 ms, 80 samples for 8 KHz
  sampling
• 5 ms look-ahead gt Algorithmic delay of 15 ms (10
  ms 5 ms)
• An 80-bit frame for 10 ms of speech
• G.729 is a complex codec with MOS of 4.0
• G.729A is a simplified version with MOS of 3.7
• G.729B
• VAD, Voice Activity Detection
• current frame plus two preceding frames (i.e. 3
  silence frames)
• DTX, Discontinuous Transmission
• Send nothing or send an SID frame
• SID frame (15 bits) used to generate comfort
  noise
• CNG, Comfort Noise Generation

16
GSM Adaptive Multi-Rate (AMR)

• Eight different modes
• 4.75 kbps to 12.2 kbps
• 12.2 kbps, GSM EFR
• 7.4 kbps, IS-641 (TDMA cellular systems)
• Change the mode at any time
• Offer discontinuous transmission
• The coding choice of many 3G wireless networks

17
Tones, Signal, and DTMF Digits

• The hybrid codecs are optimized for human speech
• Other data may need to be transmitted
• Tones fax tones, ringing, busy, congestion, etc.
• DTMF digits for two-stage dialing or voice-mail
• G.711 is OK, G.723.1 and G.729 can be
  unintelligible
• The ingress gateway needs to
• Intercept the tones and DTMT digits
• Use an external signaling system to relay the
  info
• Easy at the start of a call, difficult in the
  middle of a call
• Encode in RTP packet with tone name and duration
• Encode in RTP packet containing frequency, volume
  and duration
• Encode in RTP payload format as redundant audio
  data and send both types of RTP payload

18
RTP Payload Format for DTMF Digits

• Payload format

E end of the tone R reserved