Embed
Email

ITU-T Study Group 12

Document Sample
ITU-T Study Group 12
International Telecommunication Union







ITU-T Study Group 12

Speech Communication from Mouth to

Ear: Voice Quality Aspects in Modern

Telecommunication Networks



Dr.-Ing. H. W. Gierlich

HEAD acoustics





SG12 Workshop Dakar Oct. 2001









Parameters Influencing Speech

Quality

ITU-T

Study Group 12

Sound Quality &

Intelligibility

Naturalness

Speech Individual

Characteristics Listening

Listening & Talking Speech Conversational

Effort Effort

Quality

Environmental Network

Conditions Double Talk Conditions

Expectation

Performance

... Transmission of ...

Background Noise



SG12 Workshop Dakar Oct. 2001 2

01.10.2001

The Telephone Conversation:



ITU-T

Study Group 12









End to End Speech Quality

=

Mouth to Ear Speech Quality







SG12 Workshop Dakar Oct. 2001 3

01.10.2001









Network Configuration and

Transmission Parameter

ITU-T

Study Group 12









SLR, RLR, TELR ERL Ie, qdu ERL

LSTR, STMR T









ERL Ie, qdu ERL SLR, RLR, TELR

LSTR, STMR









SG12 Workshop Dakar Oct. 2001 4

01.10.2001

Interaction Example: DCME and

Echo Canceller

ITU-T DCME Echo Canceller

Study Group 12

s (t) T

2Dr s' (t)

Adaptive /

Filter Echo 4Dr g (t)







g’ (t) T NLP









o Cascaded comfort noise

o Cascaded switching

o Double talk detection







SG12 Workshop Dakar Oct. 2001 5

01.10.2001

TG 020









Interaction Example

Terminal/Network

ITU-T Terminal, Network Components

Study Group 12 Mobile









(Room) (Room)

Noise Noise





o Noise and addition of comfort noise

o Cascaded switching

o Speech detection and noise

o Cascaded speech coding & various types of noise







SG12 Workshop Dakar Oct. 2001 6

01.10.2001

Signal and Noise Transmission in

Handset Telephony

ITU-T 1 Signal (Voice)

2 Noise, Transmission Noise and

Study Group 12 Room Noise

3 Noise, Sidetone Coupling

4 Noise, Acoustical Sidetone

5 Noise, Structure Borne

6 Noise, Bone Conduction

7 Noise, Coupled to Ear

not Covered by the Handset

4

6 6

2

1 7

3



5









3









SG12 Workshop Dakar Oct. 2001 7

01.10.2001









The Hands-Free Situation



ITU-T

Study Group 12









e









SG12 Workshop Dakar Oct. 2001 8

01.10.2001

Parameters Influencing Speech

Quality During:

ITU-T Conversation

Study Group 12 Talking

linear distortions Listening

• transfer functions

• attenuation linear distortions

• delay • echo

• reverberance .... • reverberancel linear distortions

• delay ...

non linear distortions • transfer functions

• (in) harmonic distortions • attenuation

non linear distortions • reverberancel ....

• clipping • (in)harmonic distortions

• coding ... • clipping non linear distortions

• coding ...

additive noise components • (in)harmonic distortions

(of the echo signal)

• clipping

• system noise

• coding ...

• background noise additive noise components

• echo • system noise

• double talk ... additive noise components

• background noise • system noise

• channel noise...

during single- • background noise

• channel noise...

and double talk (during talking)









SG12 Workshop Dakar Oct. 2001 9

01.10.2001









Instrumental Measures for:



ITU-T Intelligibility Speech quality Quality during

Study Group 12 conversation





measurements during single talk:

- frequency responses

Measurements during single-

- loudness ratings

and double-talk:

- switching characteristics

- delay

- distortion measurements

- echo-loss

- measurements of S/N or N

- switching characteristics

measurement signals: - frequency responses

artificial, eventually speech like - loudness ratings

- variation of background noise

measurement during single talk

psychoacoust. motivated measures: measurement signals: speech

P.862 (PESQ) or speech-like

TOSQA

PACE

PAMS ...

measurement signals: speech

or speech-like









SG12 Workshop Dakar Oct. 2001 10

01.10.2001 tg_e056

Instrumental Measures



ITU-T “Traditional” Instrumental Measures





Study Group 12 Single Talk Situation: Double Talk Situation:



- Transfer function during double talk

- Transfer function - Loudness Ratings

- Loudness Ratings - Distortion

- Distortion - System-dynamic

- System-dynamic - System-linearity

- System-linearity

- Sensitivity against room noise

- Sensitivity against room noise

- Background noise transmission

- Background noise transm.



- Switching characteristics - Switching and echo characteristics







- Measurement applied to complete end to end configuration

- Terminals must be included

- Noise conditions (at the subscriber) need to be included

“Listen and Analyze"







SG12 Workshop Dakar Oct. 2001 11

01.10.2001









The Most Important Parameters

for LTI-Terminals (e.g. Handset

ITU-T Terminals)

Study Group 12



• Frequency response in sending & SLR



• Frequency response in receiving & RLR



• Background noise sensitivity in sending ( SM)

• Background noise sensitivity in receiving (LSTR)

• Echo performance (TCL, ERLt)









SG12 Workshop Dakar Oct. 2001 12

01.10.2001

Performance Limits:



ITU-T

Study Group 12 o Echo: G.131(single talk and double talk

o G.168 (echo canceller)

o Loudness Ratings: P.79

o Switching characteristics: P.340

o E-model: G.107, G.108, G.108.1, G.113









SG12 Workshop Dakar Oct. 2001 13

01.10.2001









Typical Single Values



ITU-T SLR

7dB

CLR

0dB

CLR

0dB

RLR

3dB

Study Group 12

TELR a(Echo) a(Echo) TELR





RLR CLR CLR SLR

3dB 0dB 0dB 7dB

OLR = 10dB



planning values:

Fig. 1 G.131

- loudness ratings

- Impairment factors Ie

- delay -> echo cancellation/suppression





problems:

• values frequency dependant

• non linear, time variant transmission properties

• proper consideration of codecs, cascading of codecs, echo cancellers..

• interaction of different transmission systems (e.g. PSTN, IP...)





SG12 Workshop Dakar Oct. 2001 14

01.10.2001

Double Talk Requirements



ITU-T echo during double talk:

Echo level offset single talk/double talk

16

12

Study Group 12

the effect of echo level 8

dB

increase during double 4



talk on subjects judgement 0

1 1,5 2 2,5 3 3,5 4 4,5

-> MOS









MOS

5

4 level variation between

3

single- and double talk:

2

the effect on subjects

1

0 3 6 9 12 16 20 judgement

-> loudness variation single/double talk









SG12 Workshop Dakar Oct. 2001 15

01.10.2001









Conclusions



ITU-T

Study Group 12 o Typical network conditions and user

conditions have to be taken into account

o Overall quality description not complete

yet for non LTI-systems

o Various quality parameters for single and

double talk situations available in ITU









SG12 Workshop Dakar Oct. 2001 16

01.10.2001


Related docs
Other docs by garrickWilliam...
Associate in Arts(10100ELEM)
Views: 5  |  Downloads: 0
DRAFT SYLLABUS!!!
Views: 30  |  Downloads: 0
pages.stern.nyu.edu~mjohnsondbmsLecture14.ppt
Views: 6  |  Downloads: 0
POSITRON EMISSION TOMOGRAPHY (Diploma)
Views: 17  |  Downloads: 1
Tunis Agenda
Views: 6  |  Downloads: 1
Demande d'admission pour Associés
Views: 7  |  Downloads: 0
By registering with docstoc.com you agree to our
privacy policy

You are almost ready to download!

You are almost ready to download!