1 TTC‘s Standardization Work Network Management Committee “A Method for Speech Quality Assessment of IP Telephony”

2 Establishment of WG on Speech Quality of IP Telephony ◆ Work Group on IP Network Technology" of the Ministry of MPHTP pointed out the need for standardization of IP telephony speech quality. ◆ The TTC formed Special Working Group 7 under Working Group 4 (WG4-7)to study a method for speech quality assessment of IP telephony.

3 Mission of SWG2 ◆ Standardize speech quality assessment method of IP telephony ◆ Downstream ITU-T Recommendations ・ G series (Transmission systems) ・ P series （ Telephone transmission quality ） ◆ Contribute to ITU-T Recommendations ・ P series （ Telephone transmission quality ）

4 Relations with Other Organizations ITU-T Recommendation CIAJ Terminal Standards VoIP Promotion Conference Application-Filing Guidelines TTC Domestic Standardization ETSI TIPHON Implementation Rules Opinions and suggestions Reference Downstream Contribution

5 ◆ Approval in Sep ・ TS-1001:A Method for Speech Quality Assessment of IP Telephony ◆ Approval in Apr ・ JJ :A Method for Speech Quality Assessment of IP Telephony (presently TS-1001, extension of a scope ) ・ JT-G107:The E-model, a computational model for use in transmission planning ・ JT-G113:Transmission impairments due to speech processing ・ JT-G165:ECHO CANCELLERS Deliverables

6 Purpose and Basic Policies of JJ ◆ The purpose of this standard is, taking account of the standardization trends of ITU-T and other organizations, to define a method of assessment with the total quality index, the R value, specified in the ITU Recommendation G.107 as the total quality index for IP telephony services and a method of assessment of parameters to complement the R value*. ◆ For the purpose stated above, we made a study in keeping with the following policies: (1) In principle, make sure that the standard conforms to related international standards (ITU-T, ETSI, and TIA standards), and ensure that it is consistent with international standards as a TTC standard. (2) Use the total quality index, the R value, for IP telephony services. (3) Specify evaluation conditions and method of assessment for the calculation of the R value. (4) In defining conditions and method of assessment, ensure the ease of assessment and fairness (strictness). * The R value is the index calculated based on the E-Model, which has been developed as a network planning tool. Because of the facts that there are quality factors that cannot be assessed with the R value and that the verification of the correspondence between the R value and subjective qualities is not satisfactory, parameters to complement the evaluation based on the R value are necessary, so is a method of their assessment.

7 Scope of JJ ◆ The following scope has been defined to maintain consistency with international standards: ・ Target services: IP telephony services (including interconnection with existing telephone services) ・ Quality type: Speech quality. (Reliability, grade of service such as probability of loss and dial tone delay are not in the scope.) ・ Time of application to a service: Before a service is offered and while the service is operation. ・ Terminal type: "Handset" (voice signal frequency band: 300 to 3400 Hz) Methods of assessment for terminals other than handsets are under consideration by ITU-T. The scope includes services using IP telephony software in terminals such as PCs. The fairness in measurement conditions will, however, be maintained by conducting measurement with handsets. ・ Methods of assessment: Method of assessment of the R value in conformance with the ITU Recommendation G.107 and a method of assessment of parameters to complement the R value ・ Mapping between the R value and MOSs: Mapping among the R value, the Western MOS, and the Japanese MOS

8 Characteristics of JJ ◆ Specifications to which revisions and additions have been made uniquely by the TTC ・ Extensions of the packet size conditions assumed in G.113 Appendix I (extensions of the packet size conditions in G.729) ・ The relationship between the Western MOS and the Japanese MOS necessary for conversion from the R value to conversational MOS ◆ Provision of guidelines for a method of speech quality assessment ・ Definition of reference configurations ・ Identification of applicable network configurations ・ Delay, echo, and listening MOS or PESQ as indexes to complement the R value

9 Concept: The law of additivity of psychological factors on psychological scale Elements Making up the R Value R = Ro – Is – Id – Ie,eff + A Advantage factor Complements the effect of the convenience of mobile or other communication on a subjective quality (satisfaction). Equipment impairment factor Represents subjective quality impairments due to low bit rate CODEC, packet/cell loss, etc. Delay impairment factor Represents subjective quality impairments due to talker echo, listener echo, and absolute delay Simultaneous impairment factor Represents subjective quality impairments due to OLR (loudness), sidetone, and quantization distortion. Basic signal-to-noise ratio Represents subjective quality impairment due to circuit noise, room noise at sending and receiving sides, and subscriber line noise. Ie,eff = f4(Ie, Bpl, Ppl) A Noise Loudness Echo and delay Distortion/ discontinuity No. of input parameters: 20. The R value is calculated from these input parameters.

10 Classification of R Value Parameters and Items That Require Assessment ◆ The 20 R value parameters are classified into those whose defaults are used and those that require assessment to ensure the ease of assessment and fairness (strictness). Major classification Concept of classificationParameter type Parameter and default Parameter whose default is used A Interpreted as currently having little impact. For example, speech quality parameters in analog transmission Noise, listener echo Nc=-70dBm0p ， Nfor=-64dBmp ， WEPL=110dB B The policy for application is not clear; at this point, the application of the parameter is inappropriate. Complementary factor such as convenience A=0 C An environmental factor, uncontrollable; a specific environment is assumed. room noisePs=Pr=35dB(A) D A terminal design parameter; standard characteristics are assumed. Loudness, sidetone, sensitivity difference, terminal echo SLR=8dB, RLR=2dB, STMR=15dB, LSTR=18dB, Dr=Ds=3, TELR （ terminal echo ） =65dB Parameter that requires assessment E An IP telephony service design parameter; requires assessment when the R value is to be derived. Listening quality qdu, Ie, Bpl, Ppl Echo TELR(PSTN echo ) Delay T, Ta, Tr ◆ JJ clarifies the relationship between the packet size and “le,eff”, that are not mentioned in G.113 Appendix I, and expands the scope of that Appendix.

11 Parameters to Complement the R Value and a Method of Their Assessment ◆ Problems with the R value (1) There are quality factors that cannot be assessed with the R value (such as burst packet loss) (2) The E-Model has been developed as a planning tool, so that it cannot satisfactorily reflect the implementation characteristics of devices. (3) The validity of the law of additivity of speech quality impairment factors such as delay, echo, and listening quality has not been sufficiently verified. ◆ To complement the total quality index with the E-model, the R value, a method of assessment of individual quality factors (parameters) is provided. ・ Delay ・ Echo ・ Listening quality: Subjective assessment (listening MOS value) or objective assessment (PESQ value)Note Note: A method has been provided to suppress the measurement noise that affects the assessment value in PESQ measurement via an electric I/F with an terminal. The requirements on the noise level are also determined.