slide1
Download
Skip this Video
Download Presentation
World Telecommunications Congress 2006 30 April - 3 May 200 6 , Budapest, Hungary

Loading in 2 Seconds...

play fullscreen
1 / 12

Outline - PowerPoint PPT Presentation


  • 245 Views
  • Uploaded on

Perceptual QoS Evaluation Model for Audiovisual Communication Services. T. Hayashi , K. Yamagishi, and H. Yoshino NTT , Japan. World Telecommunications Congress 2006 30 April - 3 May 200 6 , Budapest, Hungary. Outline. Goal of our study Concept of perceptual QoS evaluation model

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about 'Outline' - Philip


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
slide1

Perceptual QoS Evaluation Model for Audiovisual Communication Services

T. Hayashi, K. Yamagishi, and H. Yoshino

NTT, Japan

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

outline
Outline
  • Goal of our study
  • Concept of perceptual QoS evaluation model
  • Subjective quality assessment method
  • Proposed model
    • Effect of individual audiovisual quality
    • Effect of audiovisual delay
  • Conclusion

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

slide3
Goal

To find answers to the following questions:

  • How can we evaluate, design, and manage users’ perceptual QoS in order to provide comfortable audiovisual communication services?
  • How can users assess overall quality of audiovisual communication services, taking into account the balance among individual audiovisual quality, delay, and media synchronization?

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

perceptual qos evaluation model
Perceptual QoS evaluation model

Audio qualityestimation function

Auditory source

MOSA

Multimedia

quality

integration

function

Multimedia quality

(MOSMM)

Audio delay (DA)

Video delay (DV)

DA

DV

Visual

source

Video quality estimation function

MOSV

Target of this study

Task

MOS: Mean opinion score

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

subjective quality assessment 1 3
Subjective quality assessment (1/3)
  • Experimental system

17-inch

PC monitor

10-inch video window

Camera

Headphone

Network emulator

Microphone

Desktop PC

Audio/video IP packet loss rate and one-way audio/video IP packet transmission delay were controlled.

Speech coding: G.722 (64 kbps)

Video cording: MPEG4 part 2 (4 Mbps, VGA, 30 fps)

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

subjective quality assessment 2 3
Subjective quality assessment (2/3)
  • Experimental conditions

60 conditions: combinations of packet-loss rate and delay

  • Task
  • We used the task referred to as the Name-Guessing Task in ITU-T Rec. P.920.
  • The interactivity of conversation in this task was similar to a free conversation.

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

subjective quality assessment 3 3
Subjective quality assessment (3/3)
  • Assessment method
  • ACR (Absolute category rating) method
  • Subjects talked for one minute per condition and evaluated the quality using the five-grade quality rating scale.

5: Excellent, 4: Good, 3: Fair, 2: Poor, and 1: Bad

  • Each subjective quality was represented as a MOS calculated by averaging the scores of 32 subjects (non-experts).
  • The audio quality: MOSA, video quality: MOSV, and overall multimedia quality: MOSMM were individually evaluated in different rating sessions.

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

audiovisual quality estimation
Audiovisual quality estimation

Audiovisual qualityMOSAVcan be evaluated from MOSA, MOSV, and a multiplicative interaction term.

MOSAV: MOSMM for DA=DV=167 ms

4.0

CD: 0.96

RMSE: 0.14

MCI: 0.31

RMSE < MCI

3.0

Subjective MOSAV

2.0

CD: Coefficient of Determination

RMSE: Root Mean Square Error

MCI: Mean of 95% Confidence Interval for subjective MOS

1.0

1.0

2.0

3.0

4.0

Estimated MOSAV

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

effect of audiovisual delay
Effect of audiovisual delay
  • Influence of quality degradation for media synchronization was larger than that for delay.
  • Quality degradation at DA < DV was larger than that at DA > DV.

1200

MOSD: MOSMM without IP packet loss

1000

800

2.6

One-way

video delay: DV [ms]

600

2.8

Experimental conditions

3.0

400

3.2

3.4

200

200

400

600

800

1000

1200

One-way audio delay: DA [ms]

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

estimation for quality of audiovisual delay
Estimation for quality of audiovisual delay

Quality degradation due to audiovisual delay can be estimated by functions considering absolute audiovisual delay and differential delay.

MOSD: MOSMM without IP packet loss

4

CD: 0.87, RMSE: 0.16, MCI: 0.32

3

Subjective MOSD

DA = DV

DA < DV

DA > DV

RMSE < MCI

2

1

1

2

3

4

Estimated MOSD

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

conversational quality estimation
Conversational quality estimation

Multimedia quality MOSMM can be accurately evaluated from MOSAV, MOSD, and their interaction term.

CD: 0.90

RMSE: 0.16

MCI: 0.31

4

3

Subjective MOSMM

RMSE < MCI

2

1

1

2

3

4

Estimated MOSMM

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

conclusion
Conclusion
  • Perceptual QoS evaluation model for audiovisual communication services is proposed.
  • This model takes into account not only the individual audiovisual qualities but also the audiovisual delay and media synchronization.
  • The evaluation error of the proposed model was less than the statistical reliability of the subjective score, so the quality evaluation accuracy was sufficient for practical use.
  • Evaluating the effects of tasks and experimental parameters on the coefficients of the proposed model are for further study.

World Telecommunications Congress 200630 April -3 May 2006, Budapest, Hungary

ad