Speech Coders – a VoIP perspective Roar Hagen CTO SIP/email: roar.hagen@globalipsound.com
Agenda <ul><li>Speech Coders – a VoIP perspective </li></ul><ul><li>Demo </li></ul><ul><li>Q&A </li></ul>
QoS – (endpoints) status ”A lot of talk, ... but not much work” <ul><li>Year after year the same story  </li></ul><ul><li>...
QoS – status Industry’s perspective quality concerns unproven technology PSTN works fine too busy to switch not compelling...
Background - Diverse Environment PSTN Managed network Wireless Public Internet Managed network Next generation codecs shou...
Packet Loss San Francisco – Hong Kong
Jitter San Francisco – Hong Kong
Homg Kong to China VoIP Call
Wireless VoIP – The Big Unknown ? LAN 2G/2.5G WLAN Bluetooth 3G Vehicle Walk Fixed Mbps 100 10 1 0.1 Mobility
Approach <ul><li>We need holistic view/approach for both </li></ul><ul><li>Horizontal (end-to-end) perspective </li></ul><...
Vertical (Top Down) Perspective Physical Link MLPPP/FR/ATM AAL1 Network IP/WFQ/IP-prec Transport RTP/UDP/RSVP Session SIP/...
VoIP Aspirations <ul><li>IP innovation rather than PSTN replication </li></ul><ul><li>New features and services through vo...
Current speech processing technology  not  designed for packet switched environments *  M EAN  O PINION  S CORE C ONGESTED...
… congestion related VoIP  QoS problems can be solved  without  over provisioning … *  M EAN  O PINION  S CORE C ONGESTED ...
Matching PSTN Quality Telephony bandwidth speech test result S OURCE  L OCKHEED  M ARTIN GLOBAL  T ELECOMMUNICATION (COMSA...
Jitter Buffer/PLC Enhancements Source: Lockheed Martin Global Telecommunications (COMSAT)
Delay gain with NetEQ™ approx. 30-60ms compared to traditional jitter buffers   Adaptive jitter buffer Jitter Fixed jitter...
The NextGen Speech Codec Ideal <ul><li>Need one concept that will work for a long time </li></ul><ul><ul><li>footprint imp...
iLBC (internet Low Bitrate Codec) <ul><li>Speech sampled at 8 kHZ, </li></ul><ul><li>using a block-independent linear-pred...
MOS Results Source: Dynastat Inc.
iLBC - IETF work <ul><li>IETF deliverables, submitted during February ‘02: </li></ul><ul><ul><li>iLBC codec specification ...
Why iLBC !? <ul><li>Current low bit rate codecs: ITU G.729, G.723.1, GSM-EFR, and 3GPP-AMR were developed for circuit swit...
More information <ul><li>Coming Soon - web site  www.ilbcfreeware.org  with: </li></ul><ul><ul><li>Info about initiative <...
Summary <ul><li>Current speech coding technology not suited for VoIP </li></ul><ul><li>VoIP opens possibilities </li></ul>...
Demo
Upcoming SlideShare
Loading in …5
×

Download

399
-1

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
399
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
5
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide
  • iLBC codec + enhancement Standardization activity (wireless 3GPP, Wideband AMR,..) ITU standard process (circuit switched). Bit exact fixed point standard – leads to IPR problems people today no possibility to differentiation, guarantees a certain quality Internet model – free core in the codec, open standard, the best implementation (companies take IPR on implementations but this more flexible) Interoperability/lower rate (differentiation on implementation and enhancement), minimum performance Codec need footprint (picture) (limited number of codecs will survive) Feature – what kind of codecs are required? End-to-end, PSTN included need to aim for higher than mobile quality. Must be able to handle public internet (robustness, latency) – picture (PSTN, Managed network, Internet) Summary slide: Need to do it right (cannot change all the time)  internet model Handles everything including (all kinds of networks)
  • Question of managing growth… Current voice technology very sensitive to congestion Why? Well let’s go back to the FedEx &amp; UPS analogy
  • Result is: Graceful degradation of sound quality when network is congested
  • We have done it! We have proven that t can be done. Allowing for superior/hifi real-time sound over the public internet!
  • iLBC codec + enhancement Standardization activity (wireless 3GPP, Wideband AMR,..) ITU standard process (circuit switched). Bit exact fixed point standard – leads to IPR problems people today no possibility to differentiation, guarantees a certain quality Internet model – free core in the codec, open standard, the best implementation (companies take IPR on implementations but this more flexible) Interoperability/lower rate (differentiation on implementation and enhancement), minimum performance Codec need footprint (picture) (limited number of codecs will survive) Feature – what kind of codecs are required? End-to-end, PSTN included need to aim for higher than mobile quality. Must be able to handle public internet (robustness, latency) – picture (PSTN, Managed network, Internet) Summary slide: Need to do it right (cannot change all the time)  internet model Handles everything including (all kinds of networks)
  • Changed the word saving with gain and added “ approx. 30-60ms compared to traditional jitter buffers”. If you would like to change it back please feel free to do so. iLBC codec + enhancement Standardization activity (wireless 3GPP, Wideband AMR,..) ITU standard process (circuit switched). Bit exact fixed point standard – leads to IPR problems people today no possibility to differentiation, guarantees a certain quality Internet model – free core in the codec, open standard, the best implementation (companies take IPR on implementations but this more flexible) Interoperability/lower rate (differentiation on implementation and enhancement), minimum performance Codec need footprint (picture) (limited number of codecs will survive) Feature – what kind of codecs are required? End-to-end, PSTN included need to aim for higher than mobile quality. Must be able to handle public internet (robustness, latency) – picture (PSTN, Managed network, Internet) Summary slide: Need to do it right (cannot change all the time)  internet model Handles everything including (all kinds of networks)
  • Download

    1. 1. Speech Coders – a VoIP perspective Roar Hagen CTO SIP/email: roar.hagen@globalipsound.com
    2. 2. Agenda <ul><li>Speech Coders – a VoIP perspective </li></ul><ul><li>Demo </li></ul><ul><li>Q&A </li></ul>
    3. 3. QoS – (endpoints) status ”A lot of talk, ... but not much work” <ul><li>Year after year the same story </li></ul><ul><li>More then 3000 papers since 1984 </li></ul><ul><li>Limited ToS support at the end points </li></ul>
    4. 4. QoS – status Industry’s perspective quality concerns unproven technology PSTN works fine too busy to switch not compelling economics 0 10 20 30 40 50 60 percentage of respondents source: Forrester Research/AT&T (2000)
    5. 5. Background - Diverse Environment PSTN Managed network Wireless Public Internet Managed network Next generation codecs should address the needs of all applications [ ]
    6. 6. Packet Loss San Francisco – Hong Kong
    7. 7. Jitter San Francisco – Hong Kong
    8. 8. Homg Kong to China VoIP Call
    9. 9. Wireless VoIP – The Big Unknown ? LAN 2G/2.5G WLAN Bluetooth 3G Vehicle Walk Fixed Mbps 100 10 1 0.1 Mobility
    10. 10. Approach <ul><li>We need holistic view/approach for both </li></ul><ul><li>Horizontal (end-to-end) perspective </li></ul><ul><li>Vertical (top-down) perspective </li></ul>
    11. 11. Vertical (Top Down) Perspective Physical Link MLPPP/FR/ATM AAL1 Network IP/WFQ/IP-prec Transport RTP/UDP/RSVP Session SIP/H.323 Presentation Speech Codecs/…
    12. 12. VoIP Aspirations <ul><li>IP innovation rather than PSTN replication </li></ul><ul><li>New features and services through voice and data convergence </li></ul><ul><li>End-to-end IP </li></ul><ul><li>Better than PSTN sound quality </li></ul>
    13. 13. Current speech processing technology not designed for packet switched environments * M EAN O PINION S CORE C ONGESTED N ETWORK O VERPROVISIONED N ETWORK MOS = U SER E XPERIENCE “ F ALL O FF A C LIFF” s hape of curve forces over provisioning
    14. 14. … congestion related VoIP QoS problems can be solved without over provisioning … * M EAN O PINION S CORE C ONGESTED N ETWORK O VERPROVISIONED N ETWORK MOS = U SER E XPERIENCE O perate A T and A BOVE congestion point without customer knowing
    15. 15. Matching PSTN Quality Telephony bandwidth speech test result S OURCE L OCKHEED M ARTIN GLOBAL T ELECOMMUNICATION (COMSAT) 5 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 N ETWORK C ONDITION (% P ACKET L OSS) MOS GIPS iPCM™-wb+ GIPS NetEQ™-wb G.722+ GIPS NetEQ™-wb G.722.1 Source + no PLC Better Than PSTN Quality Wideband speech 0% 5% 10% 15% 20% 25% 5 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 MOS N ETWORK C ONDITION (% P ACKET L OSS) narrow band sound quality equal to PSTN wide band sound quality GIPS Ehanced G.711+ GIPS NetEQ™ G.711+GIPS NetEQ™ G.711+ITU PLC G.729A G.711+No PLC 0% 5% 30% 10% 15% 25% 20%
    16. 16. Jitter Buffer/PLC Enhancements Source: Lockheed Martin Global Telecommunications (COMSAT)
    17. 17. Delay gain with NetEQ™ approx. 30-60ms compared to traditional jitter buffers Adaptive jitter buffer Jitter Fixed jitter buffer NetEQ™ 140 120 20 80 100 40 60 0 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Packet number Delay (ms)
    18. 18. The NextGen Speech Codec Ideal <ul><li>Need one concept that will work for a long time </li></ul><ul><ul><li>footprint importance </li></ul></ul><ul><li>Need to handle large diversity of transport network </li></ul><ul><ul><li>low rate </li></ul></ul><ul><ul><li>high quality, high rate </li></ul></ul><ul><ul><li>packet loss </li></ul></ul><ul><ul><li>jitter </li></ul></ul><ul><ul><li>low delay </li></ul></ul><ul><li>Manageable IPR situation </li></ul><ul><li>Signal Robustness </li></ul><ul><ul><li>speech </li></ul></ul><ul><ul><li>music </li></ul></ul><ul><li>Suitable for variety of applications, e.g. IP video-conferencing </li></ul>
    19. 19. iLBC (internet Low Bitrate Codec) <ul><li>Speech sampled at 8 kHZ, </li></ul><ul><li>using a block-independent linear-predictive coding (LPC) algorithm. </li></ul><ul><li>Bandwidth 13.867 kbps (52 bytes per 30 ms) </li></ul><ul><li>Frame size 30 ms (support for 20 ms in the next revision) </li></ul><ul><li>Complexity and memory requirements are similar to ITU G.729A </li></ul><ul><li>Basic Quality is equal to or better than G.729. Packet loss robustness is significantly better than G.729. </li></ul><ul><li>Packet loss concealment - Integrated example solution </li></ul>
    20. 20. MOS Results Source: Dynastat Inc.
    21. 21. iLBC - IETF work <ul><li>IETF deliverables, submitted during February ‘02: </li></ul><ul><ul><li>iLBC codec specification draft - experimental standards track </li></ul></ul><ul><ul><li>iLBC RTP Payload Profile - regular standards track (AVT) </li></ul></ul><ul><ul><li>Statement about IPRs in iLBC and its “freeware nature” </li></ul></ul><ul><li>MOS results submission to the AVT mailing list during March ‘02 </li></ul>
    22. 22. Why iLBC !? <ul><li>Current low bit rate codecs: ITU G.729, G.723.1, GSM-EFR, and 3GPP-AMR were developed for circuit switched & wireless telephony and are all based on the CELP (Code Excited Linear Prediction) paradigm. </li></ul><ul><li>CELP coders are stateful, they have memory, error propagation results from lost or delayed packets. </li></ul><ul><li>iLBC treats every packet individually, making it suitable for packet communications. </li></ul>
    23. 23. More information <ul><li>Coming Soon - web site www.ilbcfreeware.org with: </li></ul><ul><ul><li>Info about initiative </li></ul></ul><ul><ul><li>Info about codec </li></ul></ul><ul><ul><li>Latest iLBC IETF drafts (spec and payload format) </li></ul></ul><ul><ul><li>Latest iLBC float point Source code </li></ul></ul><ul><ul><li>FAQ list </li></ul></ul><ul><li>IETF drafts: </li></ul><ul><ul><li>draft-andersen-ilbc-00.txt - codec spec (exper. stds track) </li></ul></ul><ul><ul><li>draft-duric-rtp-ilbc-00.txt - RTP payload profile (AVT group) </li></ul></ul><ul><li>Web site www.globalipsound.com </li></ul><ul><li>Free demo SIP client available, please request at: </li></ul><ul><li>SIP/email: [email_address] </li></ul>
    24. 24. Summary <ul><li>Current speech coding technology not suited for VoIP </li></ul><ul><li>VoIP opens possibilities </li></ul><ul><ul><li>Move quality exprience to the next level with wideband coders </li></ul></ul><ul><li>NGN will not be NGN unless we move step forward on all of its fields </li></ul><ul><li>iLBC – internet Low Bit Rate Codec </li></ul><ul><ul><li>Provide an open standard ”the Internet way ” for coder </li></ul></ul>
    25. 25. Demo
    1. A particular slide catching your eye?

      Clipping is a handy way to collect important slides you want to go back to later.

    ×