Enhanced Voice Services – EVS IWAENC 2014 Presentation

September 2014

Benefits of EVS Standardization framework Specifications Algorithmic overview and special aspects Delay and complexity Embedding EVS in 3GPP system Quality characterization Deployment / commercialization aspects

2 EVS – Next Gen 3GPP Speech Coding for Improved User Experience in Telephony

AMR 4.75 kbps 12.2

AMR-WB 6.6 kbps 23.85

EVS 5.9 kbps 128

EVS

Quality AMR- WB AMR

1995 2002 2014 3 3GPP Voice Service Evolution

AMR AMR-WB EVS

Standards 2000 2002 2014

Commercial 2001 2010 2015

WCDMA+

4 Different Voice Solutions

Carrier Grade Voice Best Effort Voice

CS Voice IMS/RCS VoIP OTT-VOIP

OPERATOR OPERATOR 3RD PARTY PORTAL PROVIDED PROVIDED PROVIDED PROVIDED SERVICE SERVICE SERVICE SERVICE TAPI Operator 3rd party HLOS Dialer Client Client Client

CDMA UMTS VoMBB VoLTE VoWiFi VoWAN

Very reliable, fully interoperable, but Ensures Consistent & Seamless Interoperable but Limited to Lacks IP Mobility, Connectivity & Interop lacks personalization Rich Voice Experience VoLTE Coverage with other mVoIP

EVRC Family AMR Family Customized Family EVS Family Internet Family G.7xx Family

5 What is EVS?

3GPP Speech Conversation / Telephony Coder EVS – Enhanced Voice Services − Next generation 3GPP speech coding − Following the successful FR, HR, EFR, AMR, AMR-WB codecs − Designed for packet-switched networks / mobile VoIP − VoLTE is a key target application − Application in other networks − AMR-WB interoperable mode − Rel-12 Work Item in 3GPP preceded by a Study Item TR 22.813

Key features − Super-wideband speech (32 kHz sampling) – improved speech quality − Source-controlled variable bit-rate operation – improved capacity − Designed for VoIP – improved robustness − Improved music performance − Wide bit-rate range and all bandwidths for maximum flexibility − Backward interoperable mode to AMR-WB

Standardization process − Qualification phase − Selection phase − Characterization phase

6 3GPP EVS is the Next Generation Speech Coder Speech quality determines user experience − Ensuring voice quality on new VoLTE deployments − EVS addresses all networks – mobile VoIP with QoS, best effort VoIP, CS 3GPP goals of Enhanced Voice Services (EVS) standardization − Feature-rich coder − Designed for VoIP applications such as MTSI in TS 26.114 − It is further desirable that the benefits of EVS are available for users of other networks such as CS − NB, WB, SWB bandwidths, FB optional, high robustness mode − Bit rates: 7.2, 8, 9.6, 13.2, 16.4, and 24.4 kb/s gross rates that comply with LTE TBSs; 32, 48, 64, 96, 128 kb/s − Quality improvements – improving user experience − Better quality in VoLTE and UMTS (with no new RAB) − Evolution path: EVS provides SWB at around 13 kbps – lower rate and lower delay SWB than other industry coders without sacrificing quality − Better quality for music and mixed content in conversational applications − Capacity improvements – increasing system efficiency − VBR at 5.9 kbps provides high capacity mode − Robustness improvements – optimized behavior in VoIP applications − More robust NB/WB through significantly better error resilience − High robustness mode

7 EVS- Enhanced Voice Services The Ultimate Codec of Choice for Mobile Telephony

EVS More natural sounding speech SWB and improved music quality

@13.2 kbps Better VoiceBetter Quality

AMR-WB

WB Improves voice clarity and intelligibility @12.65 kbps

AMR NB Toll quality narrowband voice @ 12.2 kbps

EVS @13.2kbps provides Super Wideband Voice Quality at comparable bit-rate to AMR & AMR-WB 8 EVS Benefits Better Capacity Enhanced Error Resiliency Super Wideband: 13.2 – 128 kbps Optimized for VoLTE and Circuit Switched Networks Wideband: 5.9 – 128 kbps Improved Robustness to packet Narrowband: 5.9 – 13.2 kbps loss compared to AMR-WB Support of Source-Controlled Variable Bit Rate operation

Extended audio bandwidth: 50 Hz to 16 kHz Better quality NB and WB Voice compared to AMR & AMR-WB Superior Quality Entertainment quality music coding Super Wideband (Voice and Music) Wideband (Voice) Narrowband (Voice) ~7kHz 16kHz

Low frequencies increases naturalness, presence High frequencies improves voice clarity and Reproduces better and comfort intelligibility audio and music

9 Enhanced Error Resiliency Superior Voice Quality Error Resilience Improvement for 3GPP Speech coding performance delay loss profile (6% FER) (with background car noise) 3.5 4

3 EVS Performance 3 2.5

2 2 23.85 13.2 3.5 Music coding performance (3% FER) 12.65 23.85 13.2 12 16 AMR-WB EVS AMR-WB EVS Opus

2.5

2 19.85 23.85 13.2 AMR-WB EVS Enhanced In-Call Music Quality 10 EVS – Solution for Each Situation

Jitter Buffer Management NB WB SWB FB Stereo

New EVS New EVS AMR-WB New EVS New EVS Modes (CBR) New EVS Modes New EVS Modes New EVS Modes Modes Modes Interop Modes 7.2-128 kb/s 7.2-128 kb/s (optional) (optional) (CBR) (VBR) Modes (VBR) Speech Speech Speech Speech Music Speech Speech Music Speech Music Speech Music 7.2-13.2 kb/s 5.9kb/s 6.6-23.85 7.2-128 kb/s 7.2-128 5.9 kb/s 13.2-128 13.2-128 7.2-128 kb/s 7.2-128 kb/s 7.2-128 kb/s 7.2-128 kb/s (avg) kb/s kb/s (avg) kb/s kb/s

Better Capacity Better Music Better Quality Improved Error Resilience

Same NB/WB quality as legacy Near AAC Quality at much Same capacity as legacy Much better than AMR-WB, VoIP

EVS? lower delay NB/WB Optimizations WhyDeploy

11 3GPP EVS Standardization Process in Rel-12 Requirements phase – design constraints and performance requirements Candidate coders − 13 companies submitted a candidate by 16 November 2012 − Ericsson, Fraunhofer, Huawei, Motorola, Nokia, NTT, NTTDoCoMo, Orange, Panasonic, Qualcomm, Samsung, VoiceAge, ZTE − Standardization by competition Qualification phase − Aim is to keep the most promising 5 candidates for selection − Extensive testing − 12 experiments, each candidate is tested in-house and in another listening lab − Global Analysis Lab performs collection and analysis of test results − Qualification meeting in March 2013 agreed in 5 candidates All proponents announced a collaborative development of a joint candidate Selection phase – single joint candidate − Codec selection is based on extensive testing in neutral listening labs − Selection meeting in August 2014 agreed to adopt the joint candidate as EVS standard − Agreement on most EVS specifications Characterization phase − Aim is to test the coder performance for all conditions and special signals / conditions Approval of remaining EVS Specifications and Technical Report

12 EVS Rel-12 Standardization Timeline

3GPP SA4#80

Aug 4 3GPP SA4#80bis: codec selection and approval of specifications Submission of EVS Aug 30 Approval of EVS executable 3GPP SA4#81 Technical Report and for testing SA approval of floating-point spec EVS standard Nov 6 June 27 Dec 10 Sep 15

2014 2015 Jun Jul Aug Sep Oct Nov Dec

Selection Testing Characterization Testing

EVS Prototypes Available for Preliminary Lab/Field Testing EVS Engineering Build Available For IOT and Field Trials 2015 EVS over 3G UTRAN CS Work Item (Rel-13)

13 EVS Is A Global Collaboration

Broad Industry Support Across the Ecosystem

Keys For Successful Deployment

Qualcomm • Codec hw/sw support (i.e., chipset, IMS/RCS client, Fraunhofer Samsung voice pre/post-proc, etc.) VoiceAge Nokia • Super Wideband terminal acoustic designs

Ericsson EVS Panasonic • Infra support (IMS, gateways, etc) 12 Party Collaboration • Test Equipment Support (call box, IMS, SWB Huawei NTT acoustics, voice quality) ZTE NTT DoCoMo Orange • EVS support in voice services outside of mobile ecosystem (e.g., wireline VoIP, Enterprise VoIP & Video Telephony, etc.)

14 EVS Design Requirements

Superwideband (0- 16 kHz) Coding of Speech better than AMR-WB

Constraints on Frame Length, Improved Error Max. Resilience Improved Algorithmic for both Circuit Wideband Source Coding of Delay, Switched and (0-8 kHz) Coding of Controlled Music Complexity, Packet Switched Speech better than Variable Rate for In-call Music JBM, Rate Communication AMR-WB; inclusion of Coding (Music on hold Switching, and AMR-WB IO and Ringback) PLC, RTP VoIP Capability Payload Format, VAD/DTX/CNG

Narrowband (0-4 KHz) Coding of Speech better than AMR

15 EVS Requirements in SWB at Low Rates

Category Bitrate (kbit/s) FER DTX Requirements

Clean speech 13.2 0% On†/Off NWT G.722.1C @ 32

-26,-16,-36dBov 16.4 NWT G.722.1C @ 48 24.4 NWT G.718B @ 36

Clean speech 13.2 x=3%, Off NWT G.722.1C @ 48, x% FER

-26 dBov 16.4 6% On† for 13.2 NWT G.719 @ 48, x% FER 24.4 NWT G.719 @ 56, x% FER

Noisy Speech (Car, Office, 13.2 0% On‡/Off NWT G.722.1C @ 24 when EVS DTX off

Street) NWT AMR-WB @19.85 DTX on when EVS DTX on -26 dBov 16.4 NWT G.722.1C @ 32 when EVS DTX off

NWT AMR-WB @23.05 DTX on when EVS DTX on 24.4 NWT G.722.1C @ 48 when EVS DTX off

NWT AMR-WB @23.85 DTX on when EVS DTX on

Noisy Speech (Car, Office, 13.2 x=3%, Off NWT G.722.1C @ 24, x% FER and DTX off

Street) 6% On‡ for 13.2 NWT AMR-WB @19.85, x% FER and DTX on when EVS DTX on -26 dBov 16.4 NWT G.722.1C @ 32, x% FER

24.4 NWT G.722.1C @ 48, x% FER 16 3GPP EVS Specifications

Spec No. Title Status: agreed TS 26.441 EVS Codec General Overview For approval TS 26.442 EVS Codec ANSI C code (fixed-point) For approval TS 26.443 EVS Codec ANSI C code (floating point) Draft TS 26.444 EVS Codec Test Sequences For approval TS 26.445 EVS Codec Detailed Algorithmic Description For approval TS 26.446 EVS Codec AMR-WB Backward Compatible Functions For approval

TS 26.447 EVS Codec Error Concealment of Lost Packets For approval TS 26.448 EVS Codec Jitter Buffer Management For approval TS 26.449 EVS Codec Comfort Noise Generation (CNG) Aspects For approval

TS 26.450 EVS Codec Discontinuous Transmission (DTX) For approval TS 26.451 EVS Codec Voice Activity Detection (VAD) For approval TS 26.114 MMTel CR Part for approval; part draft TR 26.952 EVS Codec Performance Characterization Not existing yet

17 Transmit Side

18 Receive Side

19 EVS Bit Rates and Supported Bandwidths Source codec bit-rates for the AMR-WB Interoperable Source codec bit-rates for the EVS codec Modes of the EVS codec

Source codec bit- Supported audio DTX availability rate (kbit/s) bandwidths Source codec bit-rate (kbit/s) Yes (Always On; Bit 5,9 (SC-VBR) NB, WB Rates are 2.8, 7.2, 8.0 6,6 kbit/s) 8,85 7,2 NB, WB Yes 8.0 NB, WB Yes 12,65 9,6 NB, WB, SWB Yes 14,25

13,2 NB, WB, SWB Yes 15,85 13,2 (channel WB, SWB Yes aware) 18,25 16,4 NB, WB, SWB, FB Yes 19,85 24,4 NB, WB, SWB, FB Yes 23,05 32 WB, SWB, FB Yes 23,85 48 WB, SWB, FB Yes 64 WB, SWB, FB Yes The SID bit rate in AMR-WB IO corresponds to AMR-WB (2.0 kbit/s). 96 WB, SWB, FB Yes 128 WB, SWB, FB Yes The SID bit rate for EVS primary modes is 2.4 kbit/s. 20 Encoding Modes Hybrid coding scheme combining linear predictive (LP) coding techniques based upon ACELP (Algebraic Code Excited Linear Prediction), predominantly for speech signals Transform coding method, for generic content, as well as inactive signal coding in conjunction with VAD/DTX/CNG (Voice Activity Detection/Discontinuous Transmission/ Comfort Noise Generation) operation EVS codec is capable of switching between these coding modes without artefacts

Common Processing Classifier Information Input Signal Signal Signal Resampling Analysis LP-based Coding

Classifier Decision Bitstream Frequency Multiplex Domain Command Line Coding Parameters

Inactive Signal Coding/CNG 21 ACELP The input signal is split into high frequency band and low frequency band paths The high-frequency portion of the signal is represented with several different parametric representations. The parameters vary as a function of the bit-rate and the residual quantization strategy. The transmitted parameters include some or all of spectral envelope, energy information and temporal evolution information. In the LP based core, the configuration of the LP coefficient estimation, parametric HF representation and the residual quantization is similar to those of AMR-WB

High Band Input Signal Signal High Band Bandwidth Parameterization Splitter

Low Band Signal

LP Coefficient Bitstream Estimation & Multiplex Interpolation

LP Filter Residual Analysis Quantization 22 Frequency Domain Operation Separation into a control layer and a signal processing layer The control layer performs signal analysis to derive several control and configuration parameters for the signal processing layer. The time-to-frequency transformation is based on the Modified Discrete Cosine Transform (MDCT) and provides adaptive time- frequency resolution. The control layer derives measures of the time distribution of the signal energy in a frame and controls the transform. The MDCT coefficients are quantized using a variety of direct and parametric representations depending upon bit rate signal type and operating mode.

23 Inactive Signal Coding When the codec is operated in DTX on mode the signal classifier depicted in Figure 1 selects the discontinuous transmission (DTX) mode for frames that are determined to consist of background noise. For these frames a low-rate parametric representation of the signal is transmitted no more frequently than every 8 frames (SID frame). The low-rate parametric representation is used in the decoder for comfort noise generation (CNG) and includes parameters describing the frequency envelope of the background signal, energy parameters describing the overall energy and its time evolution.

24 Source Controlled Variable Bit-Rate Coding VBR coding describes a method that assigns different number of bits to a speech frame in the coded domain depending on the characteristics of the input speech signal This method is often called source-controlled coding − Typically, a source-controlled coder encodes speech at different bit rates depending on how the current frame is classified, e.g., voiced, unvoiced, transient, or silence. Note that DTX operation can be combined with VBR coders in the same way as with Fixed Rate (FR) coders; the VBR operation is related to active speech segments. The VBR solution provides narrowband and wideband coding using the bit rates 2.8, 7.2 and 8.0 kbps and produces an average bit rate at 5.9 kbps. Due to the finer bit allocation, in comparison to Fixed Rate (FR) coding, VBR offers the advantage of a better speech quality at the same average active bit rate than FR coding The benefits of VBR can be exploited if the transmission network supports the transmission of speech frames (packets) of variable size, such as in LTE and UMTS networks.

25 EVS – Source Controlled Variable (SC-VBR) Bit-Rate Encoder

Inactive EVS-SID (Silence Descriptor) speech

Unvoiced- speech NELP mode

Pre-processing Pre-emphasis -> Voice-Activity- EVS speech type Voiced- classification speech Has there Detection -> LPC analysis -> been 2 EVS Voiced mode Open-Loop Pitch Estimation consecutive EVS-Voiced frames ? NO

YES PPP WI mode First Voiced- speech after voiced onset EVS Transient mode All other speech EVS Generic mode

26 SC-VBR Salient Features Enabled use of low-rate modes for certain speech types . PPP WI (Prototype Pitch Period Waveform Interpolation) mode for coding voiced speech (2.5 kbps) . Noise-Excited linear prediction (NELP) mode for unvoiced speech (2.3 kbps for narrow-band and 2.4 kbps for wide-band) . PPPWI and NELP are low-rate LP residual (excitation) signal coding schemes Illustration – improved system efficiency through lower average bit-rate

Speech Quality Active ADR

5 14

13.04 12 12.65 12.81 12.59 4 10

8 3 7.36 7.36 6

Quality

4 2 2

Average Data Rate (kbps) Rate Data Average

1 0 EVRC-WB AM R-WB at VM R-WB, EVRC-WB AM R-WB at VM R-WB, 12.65k bps M ode 0 12.65k bps M ode 0 CT1 - Input Level and FER Conditions CT2 - Noise Conditions 27 Advanced Error Resilience for VoLTE – Partial Redundancy Available in a special mode of the EVS coder (channel aware mode) offers improved performance under packet loss conditions in a VoIP system EVS offers partial redundancy based error robust channel aware mode at 13.2 kbps for both wideband and super-wideband audio bandwidths. Packets arrive at the decoder with random jitters in their arrival time; packets may also arrive out of order at the decoder. Since the decoder expects to be fed a speech packet every 20 ms to output speech samples in periodic blocks, a de-jitter buffer is required. Partial copies of the current speech frame are piggybacked on future speech frames, without increase in the total bit rate. Partial copy of a lost frame can be retrieved by polling the de-jitter buffer.

Frame n+3 n Frame n+2 Frame n+1 Frame n

• Adding partial copy of previous critical frame for better error resilience EVS-SWB-13.2 kbps with Advanced Error Resilience at 10% packet loss is equivalent in

4 quality to AMR-WB-23.85 kbps @ 3% packet loss 3.5 3 2.5 2 1.5 1 0.5 0 AMR-WB 12.65 3% FER AMR-WB 23.85 3% FER EVS-SWB-AdvErrRes-on -13.2 kbps EVS-SWB-AdvErrRes-off -13.2 kbps 10% FER 10% FER 28 Advanced Error Resilience in EVS fits well for VoLTE

AMR WB VoLTE network AMR WB Encoder Decoder

AMR-WB Received DeJitter Packets Packets Buffer

EVS VoLTE network EVS Encoder Decoder

DeJitter EVS Packets Received 29 Packets Buffer Frame Loss Concealment at Decoder Side The EVS codec includes frame loss concealment algorithms An extrapolation algorithm estimates the signal in a lost frame − For the LP based core this estimation operates on the last received residual and LP coefficients. − For the frequency domain core in some cases the last received MDCT coefficients are extrapolated and in addition the resulting time domain signal is guaranteed to give a smooth time evolution from the last received frame into the missing frames. Once the frame loss is recovered, i.e., the first good frame is received the codec memory is updated and frame boundary mismatches towards the last lost frame are minimized. For situations of sustained frame loss the signal is either faded to background noise or its energy is reduced and finally muted when no reasonable extrapolation can be assumed.

30 EVS Delay and Complexity Sampling frequencies − 48, 32, 16, 8 kHz at input and output EVS algorithmic delay − The coder operates on 20 msec frames − The algorithmic delay is less or equal to 32 msec − For 48, 32, 16 kHz sampled output, the delay consists of one 20 msec frame, 0.9375 msec delay of input resampling filters on the encoder-side, 8.75 msec for the encoder look-ahead, and 2.3125 msec delay of time-domain bandwidth extension on the decoder-side resulting in 32 msec − For 8 kHz sampled output, the decoder delay is reduced to 1.25 msec needed for resampling using a complex low-delay filterbank, resulting in 30.9375 msec overall algorithmic delay. EVS complexity − With all features supported and measured according to EVS-8b, the worst case complexity of the coder is 85.9 WMOPS which splits up to 56.7 WMOPS for encoder (24.4 kbit/s SWB with DTX on) and 29.2 WMOPS for decoder (48 kbit/s SWB with DTX off, FER=6%). − The coder uses 175KW of RAM (with no JBM included), 157 KW of ROM, and 116 KW of Program ROM. − The JBM solution was measured to consume 18 WMOPS and 49 KW RAM.

31 Embedding EVS in 3GPP System

Multimedia Telephony (MMTel) Servíce over IMS uses EVS − Support of AMR and AMR-WB for VoIP in 3GPP networks prior to Rel-12 − Inclusion of EVS support for VoLTE in Rel-12

EVS for 3G UTRAN Circuit-Switched networks − Agreed Work Item − Rel-13 time frame − Enable users of UMTS to make benefit of the enhanced quality due to EVS

32 EVS Quality – Clean Speech Inputs (Proprietary Tests) 5

4.5

4 AMR-WB EVS-SWB 3.5 EVS-WB Opus 3 G718B

Subjective DMOSSubjective Test G719 G722.1c 2.5

2 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 Bit Rate (kbps)

Note: OPUS bit rates are target average bit rates. The true average and standard deviations of the frame bit rates vary across bit rates

33 Official EVS Selection Tests in 3GPP – Experiments Exp. Content Methodology # of Exp. N1 NB clean speech under clean channel condition including input level dependency ACR 1 N2 NB clean speech under impaired channel conditions including delay/jitter profiles ACR 1 N3 NB noisy speech under clean channel condition and impaired channel conditions DCR 1 N4 NB mixed content and music under clean channel condition and impaired channel ACR 1 conditions including delay/jitter profiles W1 WB clean speech under clean channel condition including input level dependency ACR 1 W2 WB clean speech under impaired channel conditions including delay/jitter profiles ACR 1 W3 WB noisy speech under clean channel condition DCR 1 W4 WB noisy speech under impaired channel conditions including delay/jitter profiles DCR 1 W5 WB mixed contents and music under clean channel condition DCR 1 W6 WB mixed contents and music under impaired channel conditions DCR 1 W7 WB mixed contents and music under impaired channel conditions including delay/jitter DCR 1 profiles I1 AMR-WB IO clean speech under clean channel condition including input level ACR 1 dependency I2 AMR-WB IO clean speech under impaired channel conditions ACR 1 I3 AMR-WB IO noisy speech under clean channel condition DCR 1 I4 AMR-WB IO noisy speech under impaired channel conditions DCR 1 I5 AMR-WB IO mixed contents and music under clean channel condition DCR 1 I6 AMR-WB IO mixed contents and music under impaired channel conditions DCR 1 S1 SWB clean speech under clean channel condition including input level dependency DCR 1 S2 SWB clean speech under impaired channel conditions including delay/jitter profiles DCR 1 S3 SWB noisy speech under clean channel condition DCR 1 S4 SWB noisy speech under clean channel condition DCR 1 S5 SWB noisy speech under impaired channel conditions DCR 1 S6 SWB mixed contents and music under clean channel condition DCR 1 S7 SWB mixed contents and music under impaired channel conditions including delay/jitter DCR 1 profiles Total 24 34 Experiment N1 -- NB Clean Speech, Clean Channel

5.00

4.43

4.42

4.38 4.38 4.38

4.50 4.38

4.37

4.34

4.31

4.28

4.27

4.26

4.25

4.24 4.24

4.23

4.21 4.21

4.19

4.18

4.15

4.12

4.10

4.07

4.04

4.03

3.98 3.98

3.96 3.92

4.00 3.90

3.79

3.76

3.71 3.71

3.61 3.51

3.50 3.45 3.33

3.00 2.90

2.50

2.00 1.94

1.50 1.29

1.00

0.50

0.00

35 Experiment W1 -- WB Clean Speech, Clean Channel

5.00

4.48

4.46

4.45

4.45 4.44

4.50 4.41

4.35

4.32

4.31 4.31 4.31

4.30

4.24

4.22 4.22

4.21

4.19

4.17

4.11

4.09

4.08

4.07

4.06

4.05

4.04

4.03

4.02

3.99 3.98

4.00 3.93

3.84

3.72 3.51

3.50

3.31

3.29

3.28

3.25 3.17

3.00

2.87 2.79

2.50 2.02 2.00

1.50 1.37 1.14

1.00

0.50

0.00

36 Experiment W7 -- WB Music & Mixed Content at FER

6.00

5.00 4.85

4.39

4.22

4.13

4.11 4.06

4.00 3.91

3.78

3.71

3.68

3.66

3.64

3.59

3.49

3.45

3.24

3.19 2.96

3.00 2.92

2.57 2.40

2.00 1.92

1.27 1.03 1.00

0.00

37 Experiment S1 -- SWB Clean Speech, Clean Channel 6.00

5.00

4.79

4.76

4.75

4.73 4.73

4.73

4.69

4.68

4.66

4.65

4.63

4.62 4.62 4.62

4.61

4.60 4.60

4.59

4.58

4.57

4.56

4.55

4.53 4.53

4.50

4.48

4.40

4.24

4.16 3.98

4.00 3.94

3.78

3.62 3.59

3.00 2.07

2.00 1.28

1.00

0.00

38 Experiment S3 -- SWB Speech / Background Noise (Street, SNR=20dB)

5.00

4.42 4.39

4.50 4.38

4.36

4.35 4.35

4.35

4.29

4.28

4.18

4.15

4.13

4.06

4.05

3.97 3.90

4.00 3.89

3.77

3.64

3.63 3.59 3.50

3.00

2.50 2.38

2.00

1.50 1.38

1.00

0.50

0.00

39 Experiment S5 -- SWB Speech / Background Noise, FER (Car, SNR=15dB)

6.00

5.00

4.74

4.72

4.70

4.28

4.23

4.14

3.95 3.95 3.92

4.00 3.85

3.81

3.78

3.76

3.72

3.65

3.56

3.53

3.52

3.49

3.46

3.33

3.28

3.24

3.23

3.21

3.09

3.08 3.05

3.00 2.88

2.82

2.79 2.55

2.00

1.79 1.10 1.00

0.00

40 Experiment S7 -- SWB Music & Mixed Content, FER

5.00

4.67

4.63

4.61 4.52

4.50 4.41

4.37

4.32

4.19

4.12

4.11

4.06

3.98 3.93

4.00 3.89

3.79

3.77

3.68 3.68

3.67 3.67

3.63 3.57

3.50 3.39

3.32

3.29

3.26

3.22

3.18

3.05 3.03

3.00 2.59

2.50

2.36 2.17

2.00 1.94

1.50 1.21

1.00

0.50

0.00

41 EVS Rel-12 Standardization Timeline

3GPP SA4#80

2014 2015 Jun Jul Aug Sep Oct Nov Dec

Selection Testing Characterization Testing

EVS Prototypes Available for Preliminary Lab/Field Testing EVS Engineering Build Available For IOT and Field Trials 2015 EVS over 3G UTRAN CS Work Item (Rel-13)

42 EVS Deployment

EVS targets – VoLTE and other networks − VoLTE mass deployment is on-going − EVS in 3G UTRAN CS networks

VoLTE trials − Goal is to make EVS available for VoLTE trials and deployment − Pre-commercial phase during 2015

Qualcomm is key partner in EVS deployment − Serve ecosystem by making EVS codec available in mobile chipsets − Pre-standard version available for VoLTE trials and branded voice services

Deployment of standardized version can begin − IOT, field testing is first step

Qualcomm supports best-in-class voice quality for IMS based voice service deployments on LTE with a complete suite of tools and features, including EVS, IMS client, and voice enhancement.

43 Thank you

All data and information contained in or disclosed by this document is confidential and proprietary information of Qualcomm Technologies, Incorporated and all rights therein are expressly reserved. By accepting this material the recipient agrees that this material and the information contained therein is to be held in confidence and in trust and will not be used, copied, reproduced in whole or in part, nor its contents revealed in any manner to others without the express written permission of Qualcomm Technologies, Incorporated. © 2013 Qualcomm Technologies, Inc. All rights reserved. Qualcomm, Snapdragon, and Gobi are trademarks of Qualcomm Incorporated, registered in the United States and other countries. Trademarks of Qualcomm Incorporated are used with permission. Other products and brand names may be trademarks or registered trademarks of their respective owners.

QUALCOMM Technologies, Incorporated, 5775 Morehouse Drive, San Diego, CA 92121-1714