Speech Property- -Based Based FEC FEC Speech Property for - - PowerPoint PPT Presentation

Speech Property Speech Property-

• Based

Based FEC FEC for for Internet Internet Telephony Telephony Applications Applications

Nguyen Tuong Long Le and Henning Sanneck

GMD FOKUS, Berlin www.fokus.gmd.de/glone

{le,sanneck}@fokus.gmd.de MMCN 2000, San Jose, CA January 24, 2000

Overview Overview

• Voice over IP (VoIP)

– Improved quality for VoIP

• Approach

– Performance of the G.729 loss concealment – Speech Property-Based FEC (SPB-FEC)

• Evaluation

– Reference FEC schemes – Network model – Objective speech quality measurement

• Conclusions

Voice Voice

• ver
• ver IP

IP

• Main drivers:

– current economical incentives (Internet flat rate pricing) ➾ ➾ Internet Telephony – service integration, unified packet-switching infrastructure

• One of the main problems:

– satisfaction of real-time QoS demands in a packet- switched network (fundamental tradeoff: statistical multiplexing vs. reliability ➾ ➾ packet loss)

QoS for Voice over QoS for Voice over IP IP flows flows

–high compression (backward adaptive coding: ITU-T

G.729, G.723.1)

• no further sender adaptation / network adaptation

(transcoding) possible

• amplifies high perceptual impact of burst losses (error

propagation)

+tolerance to isolated losses (speech stationarity ➾

➾ extrapolation of coder state ➾ ➾ loss concealment)

➾ enhance the loss resiliency of high-compressing

codecs with open-loop error control (FEC)

Structure of Structure of an Internet an Internet Audio Audio Tool Tool

Additional components Additional components: Sender : Sender

• G.729 coder used both for the payload and the

redundancy

• Side information available at the encoder is used
• Decoder concealment process is taken into

account

Additional components Additional components: Receiver : Receiver

• No generic (PCM-level) concealment

5 10 15 20 25 30 35 40 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 frame position Signal to noise ratio

1-frame loss 2-frame loss 3-frame loss 4-frame loss

Voiced Unvoiced

Performance Performance of

• f G.729

G.729 loss concealment loss concealment

• Decoder fails to conceal losses at unvoiced/

voiced transition due to lack of state (synthesis filter parameters, excitation)

Speech Property Speech Property-

• Based

Based FEC FEC

• Adjust amount of redundancy adaptively to loss

concealment performance: overhead 40-50%

• Comparison with two reference FEC schemes

Packet (n) Packet (n+1) Packet (n+2) Packet (n+3) Packet (n+4) Packet (n) Packet (n+1) Packet (n+2) Packet (n+3) Packet (n+4) FEC scheme 1 FEC scheme 2 Audio data Redundant data

• verhead

100% 50%

Network Model Network Model

1 p q 1-p 1-q

Network loss condition 1 Network loss condition 2 Network loss condition 3 Network loss condition 4 Network loss condition 5 p=0.05, q=0.2 p=0.1, q=0.3 p=0.15, q=0.4 p=0.2, q=0.5 p=0.25, q=0.6

Results Results: : Application Application-

• level loss

level loss rate rate

0,1 0,2 0,3 0,4 0,5 1 2 3 4 5

Network loss condition Application loss rate

FEC scheme1 FEC scheme 2 SPB-FEC No FEC

0,5 1 1,5 2 2,5 3 1 2 3 4 5

Network loss condition A uditory distance

FEC s c heme 1 FEC s c heme 2 SPB-FEC No FEC

Results Results: : Auditory Distance Auditory Distance

• Application of recent advances in objective

speech quality measurement: ITU P.861A

Results Results: : Perceptual Distortion Perceptual Distortion

• Enhanced Modified Bark Spectral Distortion

(EMBSD; Temple University)

2 4 6 8 1 2 3 4 5 Netw ork loss condition Perceptual distortion FEC scheme 1 FEC scheme 2 SFB-FEC No FEC

Conclusions Conclusions

• SPB-FEC exploits differences in „concealability“

to adjust the amount of added redundancy

• simple network model & objective speech

quality measures showed the reduction of necessary redundancy to maintain a good

• utput quality
• Speech samples:

– www.fokus.gmd.de/glone/products/voice/spb-fec

• end-to-end operation: add network adaptivity
• mapping to network prioritization (DiffServ)