Speech Quality in modern Network-Terminal Configurations

Transcription

1 Speech Quality in modern Network-Terminal Configurations H. W. Gierlich HEAD acoustics GmbH ESTI STQ-workshop: Effect of transmission performance on Multimedia Quality of Service June Prague, Czec Republic

2 Outline Overview of configurations and their impact on quality Impact of Delay Coding Noise and noise cancellation Insufficient echo loss/improper echo cancellation Cascading of signal processing Missing signaling information between endpoints and network elements Summary Network - Terminal 2

3 Terminal - Network - typical elements found in modern connections Network - Terminal 3

4 Delay aggregation (I) t MT1 t MNA t IPB t PSTN1 t LDC (10-50ms) ( 80ms) (10-400ms) (1-10ms) ( 240ms, Satellite) ( 20 ms, fiber) Network - Terminal 4

5 Delay aggregation (II) t PSTN2 t IPC t PSTN3 t CT1 (1-10ms) (10-400ms) (1-10ms) (13ms) Network - Terminal 5

6 Delay aggregation (III) t oneway = t MT1 + t MNA + t IPB + t PSTN1 + t LDC + t PSTN2 + t IPC + t PSTN3 Best case = ( )ms = 146ms Worst case = ( )ms = 1113 ms Network - Terminal 6

7 E-model: Influence of delay If no terminal impairment except TCLw: 146 ms/55 db TCLw 1113 ms/55 db TCLw R_ 87,9 MOS (4.5) 4,2 R_ 26,6 MOS (4.5) 1,5 146 ms/35 db TCLw 1113 ms/35 db TCLw R_ 82,1 MOS (4.5) 4,1 R_ 4,3 MOS (4.5) 1 Network - Terminal 7

8 ITU-T G.114: Effect of transmission time 100 Users very satisfied 90 E-model rating R Users satisfied Some users dissatisfied Many users dissatisfied Mouth-to-ear-delay/ms Nearly all users dissatisfied G.114_F01 Network - Terminal 8

9 Terminal - Network - Coding and transcoding AMR/EFR G. 729 G. 723 G. 726 Network - Terminal 9

10 E-Model: The effect of codec tandeming Type Coder 1 EFR (ACELP) 5 Ie Coder 2 G. 729 (CS-ACELP) Coder 3 G. 723 (ACELP, 5.3 kbit/s) Coder 4 G. 726 (ADPCM, 32 kbit/s) 7 => Ie(tot) = = 46 R_ 47,2 MOS (4.5) 2,43 Network - Terminal 10

11 Objective Testing based on ITU-T P.862 and others Instrumental Measures based on Hearing Models: Modeling the Results of Auditory Tests by Comparison of Reference Speech Signal with Processed Speech Signal Typical Processing Steps (Schematic): Results of Listening Tests! Speech Signal Test Object Adaptation processed signal reference signal Hearing Model Hearing Model Comparison, Reference, Reduction, Signal Value S index Q PSQM P.861 PESQ P.862 PSQM99 KPN PAMS BT TOSQA T-Systems PACE ITU-T Ascom P.OLQA VQI Ericsson Network - Terminal 11

12 Example2: Tandeming of Codecs Vehicle BT GSM Preamp. A/D MOST- BUS Hands-Free Algorithm BT chip Mobile Phone GSM Network CVSD & EFR CVSD & FR Microphone & CVSD & FR Microphone (speech & noise) & CVSD & FR Network - Terminal 12

13 Echo loss in IP DECT configurations Assuming G.711 speech coder, no compression Assumimg no transcoding Assuming no VAD, sufficient bandwidth DECT Echo canceller w/o echo, ISDN echo canceller with low level echos Echoes, clipping! DECT terminals, delay. Audible echoes! ISDN GW IAD DECT Coding, packetization, buffer, Delay!! Degrades conversational quality! EC in IAD!... DECT delay, DECT distortions! Packet loss and jitter under network load, PLC, jitter buffer Artifacts, delay! Network - Terminal 13

14 Echo Performance Echo Cancellation in IAD Echo loss 36.7 db Simulated DECT echo Spectrum avg. FFT Size:8192 Overlap:0,0% Rectangle L/dB[V] EC t d f/hz NLP t + Network delay 100, 150, ms!!! Echo loss 31 db t d 30 ms Network - Terminal 14

15 DECT Echoes t d DECT t d t d linear non-linear Spectrum avg. FFT Size:8192 Overlap:0,0% Rectangle L/dB[Pa] -20 Spectrum avg. FFT Size:8192 Overlap:0,0% Rectangle L/dB[Pa] -20 Spectrum avg. FFT Size:8192 Overlap:0,0% Rectangle L/dB[Pa] f/hz f/hz f/hz Echo loss 17.8 db Echo loss 31.0 db Echo loss 36.3 db Network - Terminal 15

16 Tandeming of signal processing Audio playback system EQU Hands-free Terminal Gateway Audio playback system in the echo path of ( both ) ECs EC EC Information exchange between gateway and terminal not possible NR Microphone array signal processing ES NR AGC/ ES NR AGC/ Both EC an NR active Network - Terminal 16

17 Tandeming of noise cancellation Audio playback system Hands-free Terminal Gateway EQU Tandeming of noise cancellation: EC EC Typical result NR Microphone array signal processing ES NR AGC/ ES NR AGC/ Optimized settings Network - Terminal 17

18 Background noise transmission Mobile Terminal Mobile Network Base Station EC EC CN Influence on double talk performance Both EC & ES active ES NR AGC/ ES NR AGC/ Network 1 Network 2 Network - Terminal 18

19 Objective Testing based on ETSI EG MOS for speech quality (S-MOS) noise transmission quality (N-MOS) overall quality (G-MOS) Subjective MOS Objective MOS Noise S N G S N G Car Cafeteria Office Car Road Network - Terminal 19

20 Tandeming of echo cancellation and consequences on double talk performance Mobile Terminal Mobile Network Base Station EC EC Influence on double talk performance Both EC & ES active ES NR AGC/ ES NR AGC/ Network 1 Network 2 Network - Terminal 20

21 Double talk testing Conversational situation Network terminal A terminal B feeding feeding, recording and analysis Without background noise With background noise Analysis methods: double talk measurements based on two uncorrelated Composite Source Signals Network - Terminal 21

22 Double Talk Performance Limits Characterization of HFT acc. to ITU-T P.340 MOS 4,0 4,0 3,5 3,5 3,0 3,0 2,5 2,5 2,0 2,0 TELR DT < 13 [db] A Hsdt [db] > 15 A Hrdt [db] > 12 Type 2 Type 1 Type 3 Type 2a Type 2b Type 2c TELR DT [db] < 21 A Hsdt [db] > 12 A Hrdt [db] > 10 Network - Terminal 22

23 Summary Potential problems in a connection may occur due to tandemed signal processing in terminals and networks Currently no signalling is provided between terminals and networks for information exchange about signal processing active in the different devices Related work: G.mdcspne started in SG16 (establish the best arbitration among Network Signal Processing Equipments and terminal signal processing) G.115.x sub series (Logic and protocols for the control of signal processing network elements and functions) may potentially be enhanced to support this capability Network - Terminal 23