5th etsi speech Quality Test Event Anonymous Test Report

Download 0.63 Mb.

Page	13/13
Date	06.08.2017
Size	0.63 Mb.
	#27239

1 ... 5 6 7 8 9 10 11 12 13

4.2G.711 Hands-free mode

4.2.1Phone No.2

The sensitivity expressed by the loudness ratings meet the tolerances of 12 ± 3.5 dB SLR and 3 ± 3.5 dB RLR. The listening speech quality in sending direction of 3.7 TMOS is high whereas the receiving quality is significantly disturbed by audible distortions (loudspeaker saturation). The measured frequency responses in both transmission directions are relatively balanced (fig. 7.22 and 7.23).
	LR	TMOS	Background noise is completely transmitted as analyzed in fig. 7.24.
SND	16.8 dB	3.7
RCV	1.9 dB	1.8

Fig. 7.22: Sending			Fig. 7.23: Receiving	Fig. 7.24: Café (VAD on)
				The ETSI EG 202 396-3 results are low due to a strong AGC. The send signal is completely muted (digital zero) during the application of a far end signal (type 3 double talk implementation). This muted send signal can also be analyzed from the Relative Approach representation in fig. 7.26 (single talk).
			Fig. 7.25: Near end speech (DT)
The analysis curve is equal to zero during speech activity in receiving direction. The application of far end signals also leads to a complete suppression of background noise signals in sending direction. Comfort noise is not inserted (fig. 7.27 and 7.28).

Fig. 7.26: Echo, Rel. Approach			Fig. 7.27: Noise transmission with far end signal	Fig. 7.28:  RA. Noise transm. with far end speech

4.2.2Phone No.3

Both transmission directions meet the requirements for the loudness ratings (12 ± 3.5 dB SLR, 3 ± 3.5 dB RLR). Listening speech quality in sending direction is low (2.3 TMOS) due to high idle noise level. The receiving direction provides a listening speech quality of 2.8 TMOS, which slightly violates the recommended limit of 3.1 TMOS. The frequency responses in both transmission directions are relatively balanced (fig. 7.29, 7.30) except an attenuation around 800 Hz in receiving direction.
	LR	TMOS	The background noise is transmitted but parts are substituted by comfort noise (see fig. 7.31).
SND	12.4 dB	2.3
RCV	0.4 dB	2.8

Fig. 7.29: Sending			Fig. 7.30: Receiving	Fig. 7.31: Café (VAD on)
				The quality pie indicates a very low speech quality in the presence of background noise. The implementation does not provide double talk capability (type 3). The measured echo attenuation (TCL_w 50 dB) is sufficiently high. Short term components are detected by the Relative Approach (fig. 7.33).
			Fig. 7.32: Near end speech (DT)
During the background noise transmission tests the send direction is muted coincident to the application of the far end signal (see fig. 7.34 and 7.35). The attenuation is not removed after the excitation signals ends.

Fig. 7.33: Echo, Rel. Approach			Fig. 7.34: Noise transmission with far end signal	Fig. 7.35:  RA. Noise transm. with far end speech

4.3G.722 Handset and Hands-free mode

4.3.1Phone No.1

SLR = 5.1 dB TMOS = 3.1	RLR = 4.7 dB TMOS = 2.3
Fig. 7.36: Sending	Fig. 7.37: Receiving
The sensitivity in both directions is sufficiently high in handset mode (SLR 3.2 dB, RLR 4.7 dB). Listening speech quality of 3.1 TMOS in sending direction is slightly lower than expected, although the frequency response is balanced (see fig. 7.36). The idle noise limits the result. The receiving frequency response shows a strong high pass characteristic violating the tolerance below approx. 1 kHz. This limits the listening speech quality to 2.3 TMOS only.
The echo attenuation of 53.7 dB slightly violates the recommended limit of ≥ 55 dB. The spectral echo attenuation tolerance assumes a “worst case“ round trip delay of 500 ms in fig. 7.38. The tolerance is violated for frequencies higher than 2 kHz. This is caused by low level echo components and the audible idle noise level in sending direction.
	Fig. 7.38: Spectral echo attenuation
The listening speech quality in the presence of background noise slightly violates the limits due to audible noise (see Quality Pie).
Wideband HFT not yet implemented

4.3.2Phone No.2

SLR = 3.2 dB TMOS = 3.8	RLR = 3.2 dB TMOS = 4.1
Fig. 7.39: Sending	Fig. 7.40: Receiving
The handset provides a high sending sensitivity (SLR 3.2 dB) and a high listening speech quality. The frequency response is balanced (see fig. 7.39). Listening speech quality in receiving direction is also high as indicated by the TMOS of 4.1. The frequency response demonstrates that the handset is wideband capable (fig. 7.40).
The echo attenuation of 57 dB is sufficiently high. The slight violation of the tolerance scheme in fig. 7.41 is caused by audible noise but not by echo components. The N-MOS for the car and pub noise are very high showing the strong influence of attenuation control. The overall performance is balanced.
	Fig. 7.41: Spectral echo attenuation

The phone also provides a high speech level in hands-free mode. The SLR and RLR are in the recommended range. The TMOS results are sufficiently high. Both frequency responses in sending and receiving direction (7.42, 7.43) indicate a wideband capable microphone and loudspeaker performance. The measured echo attenuation of 74 dB is high. Listening speech quality in the presence of background noise is low –especially for the café noise scenario- as indicated by the quality pie chart.	SLR =3B HFT not implemented)onährend eines ST Echos das sende-Signal 9.6 dB TMOS = 3.5	RLR = 4.2 dB3B TMOS = 3.2
	Fig. 7.42: Sending	Fig. 7.43: Receiving

	Fig. 7.44: Spectral echo attenuation

4.3.3 4.3.4Phone No.3

SLR = 13.3 dB TMOS = 1.8	RLR = 4.9 dB TMOS = 2.0
Fig. 7.45: Sending	Fig. 7.46: Receiving
The IP phone handset provides narrowband instead of wideband transmission capability. The acoustics does not seem to be adapted on the need of wideband transmission. Both frequency responses show the limitations (7.45, 7.46). Listening speech quality is poor (TMOS 1.8 and 2.0).
The echo attenuation of 46 dB is too low. Echo components lead to a violation of the tolerance scheme in fig. 7.47 over the whole frequency range. Listening speech quality in the presence of background noise is still high, although the bandwidth is limited. Transmission is not impaired by artifacts (see Quality Pie).
	Fig. 7.47: Spectral echo attenuation

The limitations of the acoustic transducers also occur in hands-free mode. The SLR and RLR are in the recommended range but the TMOS results are low. Both frequency responses signifycantly violate the tolerances. The HFT provides an echo attenuation of 45.7 dB. The tolerance is significantly violated in fig. 7.50. A continuously audible idle noise with strong tonal components limits listening speech quality in sending direction also in the presence of background noise (see results given in the quality pie).	SLR = 7.1 dB TMOS = 1.5	RLR = 4.7 dB TMOS = 2.9
	Fig. 7.48: Sending	Fig. 7.49: Receiving

	Fig. 7.50: Spectral echo attenuation

References

[1] ETSI TS 101 329-5: Telecommunications and Internet Protocol Harmonization Over Networks (TIPHON), Part 5: Quality of Service (QoS) measurement methodologies

[2] ITU-T Recommendation P.58, Head and Torso Simulators for Telephonometry

[3] ITU-T Recommendation P.57, Artificial Ears

[4] ITU-T Recommendation P.64, Determination of sensitivity/frequency characteristics of local telephone systems

[5] ITU-T Recommendation P.340, Transmission Characteristics And Speech Quality Parameters of Hands-free Telephones

[6] ITU-T Recommendation P.501, Test Signals for Use in Telephonometry

[7] ITU-T Recommendation P.502, Objective analysis methods for speech communication
systems, using complex test signals

[8] Instrumentelle Verfahren zur Sprachqualitätsschätzung - Modelle auditiver Tests; PhD Thesis Jens Berger, Christian-Albrechts Universität Kiel, 1998

[9] EG 201 377-1: Speech Processing, Transmission and Quality Aspects (STQ); specification and measurement of speech transmission quality; part 1: Introduction to objective comparison measurement methods for one-way speech quality across networks.

[10] ITU-T Recommendation P.862, Perceptual evaluation of speech quality (PESQ), an objective method for end-to-end speech quality assessment of narrowband telephone networks and speech codecs

[11] ITU-T Recommendation P.862.1, Mapping function for transforming P.862 raw result scores to MOS-LQO

[12] Anonymized Test Report, 1^st ETSI Speech Quality Test Event, ETSI Bake-off Service, Deutsche Telekom Berkom, HEAD acoustics, Dec. 2000

[13] Anonymized Test Report, 2^nd ETSI Speech Quality Test Event, ETSI Plugtests, T-Systems Nova Berkom, HEAD acoustics, June 2002

[14] Anonymized Test Report, Gateways, 3^rd ETSI Speech Quality Test Event, ETSI, December 2004

[15] Anonymized Test Report, Phones, 3^rd ETSI Speech Quality Test Event, ETSI, December 2004

[16] ETSI 4^th Speech Quality Test Event 2006, Anonymized Test Report, IP Gateways and IP Phones; ETSI Plugtests^TM Service, HEAD acoustics

[17] ITU-T Recommendation P.800, Methods for Subjective Determination of Transmission Quality

[18] ITU-T Recommendation P.830, Subjective performance assessment of telephone-band and wideband digital codecs, Annex D, 1996

[19] ITU-T Recommendation P.56, Objective measurement of active speech level

[20] Genuit, K.: Objective Evaluation of Acoustic Quality Based on a Relative Approach, InterNoise '96, Liverpool, UK

[21] Application of the Relative Approach to Optimize Packet Loss Concealment Implementations, F. Kettler, H.W. Gierlich, F. Rosenberger, DAGA, March 2003, Aachen, Germany

[22] ETSI EG 202 396-1 V.1.1.1 (2005-11): Speech Processing, Transmission and Quality Aspects (STQ); Speech Quality performance in the presence of background noise; Part 1: Background noise simulation technique and background noise database

[23] ETSI EG 202 396-3 V.1.2.1 (2008-11): Speech Processing, Transmission and Quality Aspects (STQ); Speech Quality performance in the presence of background noise; Part 3: Background noise transmission - Objective test method

[24] Junqua, J.-C.: The Influence of Acoustics on Speech Production: A Noise-induced Stress Phenomenon known as the Lombard Reflex, Speech Communication 20 (1996), page 13-22

[25] ETSI TBR 10, Digital Enhanced Cordless Telecommunications (DECT); General terminal attachment requirements; Telephony applications

[26] ITU-T Recommendation G.122: Influence of national systems on stability and talker echo in international connections

[27] ITU-T Recommendation P.800, Methods for Subjective Determination of Transmission Quality

[28] ITU-T Recommendation P.800.1, Mean Opinion Score (MOS) Terminology

[29] ITU-T Recommendation P.505, One-view visualization of speech quality measurement results

Download 0.63 Mb.

Share with your friends:

1 ... 5 6 7 8 9 10 11 12 13