1 E-Model & MOS Speaker: Cheng-lin Tsai Adviser: Quincy Wu Date:2009/07/02
2 Outline Introduction E-model Transmission rating R factor E-model implementations in VoIP network R-factor invert to MOS Reference
3 Introduction Speech quality mainly impaired by packet loss, delay, and delay jitter It is determined by listener’s perception MOS (Mean Opinion Score) - subjective Compare reference and degraded signal PAMS, PESQ - objective A computational model E-model - objective
4 E- model E-model (ITU-T G.107) is a transmission planning tool that provides a prediction of the expected voice quality Mouth-to-ear transmission quality Use Transmission impairment factors that reflect the effect Transmission rating R factor (0 ≦ R ≦ 100) R factor can be transformed into other quality measures Such as MOS It considers a wide range of telephony-band and VoIP impairments
5 Reference connection of the E-model
6 Transmission rating R factor (1/4) Ro: basic signal-to-noise (SNR) - circuit noise and room noise Is : simultaneous impairment factor - combination of all impairments which occur more or less simultaneously with the voice signal Id : delay impairment factor - impairments caused by delay
7 Transmission rating R factor (2/4) Ie-eff :effective equipment impairment factor - impairments caused by low bit-rate codecs A : advantage factor - allow for compensation of impairment factors when there are other advantages of access to the user ex:
8 Transmission rating R factor (3/4)
9 Transmission rating R factor (4/4) If all parameters are set to default values, the rating factor of R =94.2
10 E-model implementations in VoIP network (1/2) The connection between CODEC and telephone handset may be transient R = Id(Ta) - Ie(codec,loss) Id(Ta) about one way delay Id = 0.024*d+0.11*(d-177.3)*H(d-177.3) Ie about codec type and packet loss rate Ie= λ 1 + λ 2 ln(1+ λ 3 e) Codecλ1λ1 λ2λ2 λ3λ3 G G
11 R-factor invert to MOS (1/3) Categories of speech transmission quality
12 R-factor invert to MOS (2/3) R<0 : MOS =1 0<R<100 : MOS = R+R(R-60)(100-R)7*10 -6 R>100 : MOS = 4.5
13 R-factor invert to MOS (3/3)
14 Reference ITU-T G.107 (E-model) ITU-T G.113 appendix 1 Modeling the Effects of Burst Packet Loss and Recency on Subjective Voice Modeling the Effects of Burst Packet Loss and Recency on Subjective Voice An E-Model Implementation for Speech Quality Evaluation in VoIP Systems An E-Model Implementation for Speech Quality Evaluation in VoIP Systems Voice over IP Performance Monitoring Speech Quality Prediction in VoIP Using the Extended E-Model Speech Quality Prediction in VoIP Using the Extended E-Model