2. 2nd Workshop on Wideband Speech Quality - June 2005 1 Some Aspects of Wideband Speech in Enterprise Telephony Eric J. Diethorn (ejd@avaya.com)ejd@avaya.com with Gary W. Elko (gwe@avaya.com) andgwe@avaya.com Joseph L. Hall (jhall01@avaya.com)jhall01@avaya.com Avaya, Inc. Avaya Labs, Research 233 Mt. Airy Road, Basking Ridge, New Jersey 07920 USA

3. 2nd Workshop on Wideband Speech Quality - June 2005 2 Outline  Physical acoustics  Echo  Voice coders  Conferencing  Wideband speech & intelligibility  Hallway demonstration – Avaya SIP Softphone

4. 2nd Workshop on Wideband Speech Quality - June 2005 3 Some introductory thoughts  Wideband speech telephony will instantaneously raise the bar of end-user expectation, at least for some applications.  Skype  We have standards for the reproduction of wideband speech, but is wider-band good enough?  Maybe [150, 5000] is good enough?  With greater bandwidth comes a greater range of potential artifacts that the acoustical-signal-processing engineer must address.  Low-frequency acoustic echo, earpiece hiss, speech-coder distortion, arbitration of multiple sampling rates.  The preferences of end users are uncertain.  Speech bandwidths policies (buddy lists, profiles)?  Suppose I have a physiological speech impediment. Do I want it emphasized?

5. 2nd Workshop on Wideband Speech Quality - June 2005 4 Physical acoustics  The physical design of terminal acoustics must change to render wideband speech.  Acoustical signal processing changes, too.

6. 2nd Workshop on Wideband Speech Quality - June 2005 5 Loudspeakers & enclosures Frequency (Hz) Sound Pressure Level (dB) Frequency response, traditional narrowband speakerphone, 80 dB-SPL50 cm

7. 2nd Workshop on Wideband Speech Quality - June 2005 6 Loudspeakers & enclosures Frequency (Hz) THD at harmonics (%) Total harmonic distortion, traditional narrowband speakerphone, 80 dB-SPL50 cm High distortion at low frequency end of wideband-speech spectrum Acoustic echo control difficult if not impossible without acoustical modifications.

8. 2nd Workshop on Wideband Speech Quality - June 2005 7 Earpieces Frequency (Hz) Sound Pressure Level (dB) Frequency response, wideband handset In order to satisfy wideband standards, acoustical modifications are necessary to extend the low-frequency response of most earpiece designs. This is particularly challenging for physical arrangements in which the earpiece is held to the ear with little pressure.

9. 2nd Workshop on Wideband Speech Quality - June 2005 8 Microphones  Most low-cost electret microphones used today have a frequency response that is practically flat beyond the range of wideband speech – they are “wideband ready.”  Multiple microphone arrangements – arrays – can be exploited to reduce the level of ambient noise at frequencies not present in traditional narrowband telephony.  Low-frequency rumble.  High-frequency hiss.  Short-time spectral modification methods of noise reduction can help, but the perception of artifacts from such processing is enhanced by the wider speech band.

10. 2nd Workshop on Wideband Speech Quality - June 2005 9 Microphones  Omnidirectional microphone (traditional)  Good pick-up of talkers in all directions  But, picks-up ambient noise from all directions  Directional microphone  Reduces off-axis noise  Reduces reverberation of talker’s voice  Reduces coupling from speakerphone (helping AEC)  But, talkers off axis can’t be heard well. Front of phone Front of phone

11. 2nd Workshop on Wideband Speech Quality - June 2005 10 Echo  Requirements on echo control may change.  The art of echo control must evolve to meet the challenge of wideband speech.

12. 2nd Workshop on Wideband Speech Quality - June 2005 11 Requirements on Talker Echo Source: Transmission Systems for Communications, Bell Telephone Laboratories, Inc., 5 th Edition, 1982.  Roundtrip, mouth-to-ear, echo loss requirements were measured on populations for narrowband speech. How well do these data apply to wideband speech echo paths? Echo annoyance as a function of roundtrip, mouth-to-ear loss and delay, for narrowband speech. Acoustic-to-acoustic echo-path loss (dB) Percent Good-or-Better

13. 2nd Workshop on Wideband Speech Quality - June 2005 12 Talker Echo, Continued Source: Transmission Systems for Communications, Bell Telephone Laboratories, Inc., 5 th Edition, 1982.  Being strictly digital, wideband- speech network paths do not suffer from analog circuit noises, however, analog and environmental noises enter calls at the endpoint. Should requirements on talker echo incorporate such (wideband) noise phenomena? Echo annoyance as a function of roundtrip, mouth-to-ear echo-and-noise loss. Long-haul (~1000 mi.) PSTN connection, circa 1980.

14. 2nd Workshop on Wideband Speech Quality - June 2005 13 Wideband speech coding  G.722, G.722.1 and G.722.2  G.722 is cheap.  G.722.1 often comes with video-on-the-enterprise (Polycom).  Proprietary codecs  Silicon solution providers have their favorites. Some are pretty good.  Linear 16-bit encoding?  Speech-transmission bandwidth (bits-per-second) is becoming a non-issue in the enterprise, at least for wired LANs.  Architecturally appealing within the enterprise. Let boundary gateways worry about transcoding.

15. 2nd Workshop on Wideband Speech Quality - June 2005 14 Multirate audio conferencing IP-1 Conference bridge server narrowband speech Wide- and narrow-band speech Leased WAN (compressed speech, e.g., G.729, G.726) PSTN IP-2......  Rate arbitration  Transcoding  Multirate mixing  (Artificial) bandwidth extension

16. 2nd Workshop on Wideband Speech Quality - June 2005 15 Stereo audio conferencing Hands-free, wideband-speech communications with stereo echo cancellation ROOM 1 ROOM 2 h1h1 h1h1 ~ talker + - echo h2h2 h2h2 ~ g2g2 g1g1 NL

17. 2nd Workshop on Wideband Speech Quality - June 2005 16 Stereo Conferencing (Placeholder, video demonstration)

18. 2nd Workshop on Wideband Speech Quality - June 2005 17 Wideband speech & intelligibility  Siemens – “…wideband transmissions can reduce speech ambiguities by as much as 90 percent, increasing conversational intelligibility and reducing listener fatigue.” (2003 press release)  Polycom – “For single syllables, 3.3 kHz bandwidth yields an accuracy of only 75 percent, as opposed to over 95 percent with 7 kHz bandwidth.” (2003 white paper)  Marketing vs. science – both required

19. 2nd Workshop on Wideband Speech Quality - June 2005 18 Experimental study *  Similar to Diagnostic Rhyme Test and Diagnostic Alliteration Test, except we generated our own word pairs  e. g., “tie” & “pie” (“hot” & “hop”)  Subject hears one of the two, is shown both, is asked “Which of these two did you hear?”  Clean anechoic speech filtered to 3 bandwidths  [50,3300], [50,5000] and [50,7000] Hz.  Investigate all nine combinations of three bandwidths and three additive-noise levels (0 dB, +12 dB, +24 dB SNR).  Reference: G.A. Miller and P.E. Nicely, “An analysis of perceptual confusions among some English consonants” Lincoln Laboratory, MIT, 1955 (J. Acoust. Soc. Amer. Vol. 27, pp. 338-352) * For questions concerning aspects of this study, contact Joseph L. Hall, Avaya Research, jhall01@avaya.comjhall01@avaya.com

20. 2nd Workshop on Wideband Speech Quality - June 2005 19 What do they sound like? “Seed, feed, seed” at different bandwidths and additive noise levels. 3.3 kHz LP5 kHZ LP7 kHZ LP CLEAN 24 dB SNR 12 dB SNR 0 dB SNR

21. 2nd Workshop on Wideband Speech Quality - June 2005 20 Representative results

22. 2nd Workshop on Wideband Speech Quality - June 2005 21 Summary of results

23. 2nd Workshop on Wideband Speech Quality - June 2005 22 Hallway Demonstration -- Avaya wideband SIP softphone  Wideband speech (16 kHz sampling, bandwidth limited by PC sound architecture).  Voice codecs  G.711, G.729, G.726  G.722