Avaya Audio Quality Tuning User Manual

FIELD APPLICATION NOTE

Audio Quality Tuning for IP Telephones

Issue 2.0

July 10, 2007

COMPAS ID – 120942

Abstract

IP Telephones can be fine-tuned to provide optimal audio quality performance by administering audio parameters in the file server settings file. This document provides a guide to administering those parameters as well as a description of each parameter and its range of settings. Examples of common customer audio quality problems are discussed with audio parameter settings identified to fix those problems. New to this document are the additional audio sidetone and noise reduction settings made available starting with IP Telephone patch release R2.810.

1 Introduction

This document describes the audio tuning parameters administrable via the 46xx/96xx settings

file

and adjustments that can be made to optimize the phone for specific customer environments and audio impairments. The available range of programmable system parameters will allow the engineer/technician to adjust the audio performance of the IP Telephone (IPT) for the various customer locations and their respective environments.

After being introduced in R2.4, the AUDIOENV, AUDIOSTHS, and AUDIOSTHD parameters have underwent value changes in order to meet the needs of our customers. With the release of R2.810, new sidetone settings have been added to AUDIOSTHS and AUDIOSTHD so that any level available in a previous release can now be obtained with one of the ten settings now available in R2.810. AUDIOENV has also added two new noise reduction settings, which are described in section 3.2.2.

Appendix A provides a “drill down” of the various audio parameter settings that are now available. Appendix B contains a Quick Start Guide that identifies specific audio quality symptoms and the tuning parameter setting(s) that can be used to fix the issue. Appendix C provides guidelines for performing noise measurements.

2 Document History

Release Number

1.0

2.0

Date Comments

June 29, 2006 Initial release

July 10, 2007 Updated to add new audio settings

3 Audio Quality Tuning Parameters

Audio quality tuning parameters were developed to address customer audio quality issues that arise from incoming audio signal impairments or those that originate from the customer’s environment. The engineer/technician can mitigate these audio quality issues using the customizable audio parameters were not designed to fix system component defects or network performance issues.

The following sections provide an explanation of each audio parameter so that the reader can gain an understanding of these parameters and the types of audio issues that they can address. This will help when deciding which adjustments should be made to mitigate the customer’s audio quality issue(s).

3.1 Tuning Parameter Changes from R2.4 to R2.810

AUDIOENV, AUDIOSTHS, and AUDIOSTHD parameter settings have changed in the interim releases (from R2.4 through R2.810) due to customer feedback. Because of this, an audio parameter setting in one release may not have the same performance as that same setting in a later release. This has all been remedied in release R2.810. New settings have been added to AUDIOSTHS and AUDIOSTHD so that any sidetone level available since R2.4 can be obtained with one of the ten settings now available in R2.810. See Table 1 of Appendix A.

available in the settings file. Note that the tuning parameters

Except the 4690 and 4625 (prior to R2.7)

Starting with IPT Release R2.4

COMPAS ID 120942

Page: 2 of 28

The ability to turn off the expander (now more appropriately called NR3) was added in R2.7 and a more aggressive noise reduction setting (Excessive noise) was added in R2.810. The reason for the name change from expander to noise reduction (NR) is to more accurately describe the actual function being used – noise reduction (NR) as opposed to expansion (Exp).

See Table 2 and 3 in

Appendix A.

Because of the two added states in AUDIOENV, the AUDIOENV matrix has grown from 108 to 299 possible settings. It is certainly not intended that trial and error be used when selecting an AUDIOENV setting to fix a customer audio quality issue. Rather, the intent is for the engineer/technician to isolate the audio impairment and select an AUDIOENV value that has a variable setting designed to address that audio impairment and improve the performance of the phone. See Table 4 in Appendix A.

3.2 Customizable Audio Tuning Parameters

The following audio parameters can be administered5 on a system-wide basis, or, when a customer has multiple operating environments, the Group Feature

or local DHCP/file server7 can section the customer environments into groups of users. Seek out the 4600 Series IP Telephone LAN Administrator Guide or escalate to SSE Endpoints for assistance in making these network adjustments.

For AUDIOENV changes to take effect, they must be made prior to boot-up of the phone.

3.2.1 AUDIOSTHS / AUDIOSTHD

The AUDIOSTHD and AUDIOSTHS variables provide sidetone customization for both headset and handset respectively. Typical use of these settings is to reduce the level of ambient noise (which has entered the microphone) from being heard in the handset/headset receiver. Appendix A provides a table of sidetone adjustments.

AUDIOSTHS – controls the sidetone for the handset. The range of values is from zero (default setting) to nine. The levels allow adjustment in ~3dB steps as defined in Table 1 of Appendix A.

AUDIOSTHD – controls the sidetone for the headset. The ranges of values are the same as those defined for AUDIOSTHS above.

3.2.2 AUDIOENV

AUDIOENV serves as an index into a matrix containing four audio parameters. Each AUDIOENV value is unique in that at least one audio parameter changes value as the AUDIOENV value changes. The AUDIOENV variable can be set to values ranging from 0 – 299 possible settings with the default setting being 0. The four audio parameters are defined below.

Starting with R2.810, two new noise reduction states/settings have been added: one to turn off noise reduction and one to increase the loss added to noise.

AUDIOENV variable names Exp_thresh_Hs and Exp_thresh_Hs are now known as NR_Thresh_HS and

NR_Thresh_HD, respectively.

Expansion reduces background noise by inserting loss on lower level microphone output signals while noise reduction

inserts loss on any microphone output signal that is deemed noise.

Although the customer can make sidetone adjustments on their own and to their liking, it is highly recommended that AUDIOENV setting adjustments be made only with direction from Avaya Services or the Business Partner technical representative.

Requires the phones to be programmed manually – consult Tier IV.

Note that settings files can be local to the environment or building if it has its own DHCP server.

COMPAS ID 120942

Page: 3 of 28

AGC_Dyn_Range - Receive AGC Dynamic Range: Used to boost low-level speech from the far end party. Be aware that increasing the AGC dynamic range will not only boost speech, but will also boost any low-level environmental noise from the far end or circuit noise from network equipment in the call topology.

• AGC dynamic range has default value of 0 (+/-9dB). Other available settings are 1 (+/12dB), 2(+/-15dB), and 3 (+/-18dB).

NR_Thresh_HD - Headset Noise Reduction Threshold. This variable is used to reduce/eliminate environmental (ambient) noise

from being transmitted to the far end party. Adding more loss will

reduce the level of ambient noise from being transmitted to the far-end party. MARK

• The noise reduction threshold for the headset has a default value of zero for a typical office environment, 1 for noisy office environments, 2 for very noisy office environments, 3 to turn off noise reduction, and 4 for excessive noise in the environment. Setting 4 is recommended9 for call centers or environments where there is sensitive information audible in the background.

NR_Thresh_HS - Handset Noise reduction Threshold. Same as NR_Thresh_HD above (except this affects handset).

HD_TX_Gain - Headset Transmit Gain. Used either to limit environmental noise from being heard in the headset (or being transmitted to the far end party) or to boost the headset user’s speech so that the far end party can hear them better. Setting 1 used when far end complains they can’t hear the headset user while setting 2 is used to limit the noise heard in the headset and/or limits the noise transmitted to the far end party.

• Headset transmit gain has a default value of zero for normal transmit gain, 1 for +6dB of gain, and 2 for -6dB of gain.

Appendix A provides the AUDIOENV matrix table that contains all the possible AUDIOENV settings.

4 Optimizing IPT Audio Quality for Customer Environments

This section identifies customer environments in which specific audio parameters were changed to successfully mitigate customer audio issues. The recommendation is to use the settings provided as a starting point and then, if needed, fine-tune the audio parameters for the specific customer environment. See Appendix C for typical noise levels and environments.

4.1 Normal or Quiet Offices

While most customer issues arise out of noisy environments, customers have also raised audio quality issues when using the IP Telephone in quiet environments. A quiet office is defined as having a dBA reading of about 40dBA while a normal office is defined as having a noise level of about 50dBA.

Usually when an office environment is quiet artifacts on the call that are usually masked describe the artifact as static, or an intermittent noise that sounds like cracking on the line. If the 46xx/96xx telephone user continues to hear this noise after they have upgraded their phone to

Noise can be either mechanical or low level nearby speech.

This setting is only a recommendation and is not a guarantee that sensitive information will still not be transmitted to the far end party. This is because the phones were designed to transmit speech, and if the nearby speech is loud enough, it will be passed to the far end.

“Quiet” is subjective and is not bound by the defined noise levels provided.

A weaker sound is masked if it is made inaudible in the presence of a louder sound.

COMPAS ID 120942

, a user will tend to hear or take notice of audible

in a noisier environment. The user will sometimes

Page: 4 of 28

R2.4 or later (which has a default AGC dynamic range of +/- 9dB), they should disable AGC on the headset, handset, and/or speakerphone depending on which mode of communication they are using to listen to the call. Disabling AGC will prevent the phone from boosting low level receive signals (such as noise).

4.2 Noisy Offices

A noisy office is defined as having a dBA reading of about 60dBA.

When the ambient noise in a customer’s office environment is loud, it can adversely affect the audio quality of both the near-end and far-end party on the call. Knowing what end of the call the room noise is affecting will determine what type of audio adjustment to make. The following sections provide solutions for the effects environmental noise has on the near end phone user as well as the effect the near end noise has on the far end listener.

4.2.1 Far-end Audio Quality Problems caused by Near-end Environmental Noise

Ambient room noise (mechanical noise or nearby speech) entering the microphone of a handset or headset can be the cause of both a complaint from the far-end party (they cannot understand the handset/headset user) or a complaint by our customer that the nearby speech (containing sensitive information) is being sent to (and heard by) the far-end party. Whether mechanical or nearby speech, the transmission of this noise is unwanted.

Depending on whether the complaint is noise being transmitted or sensitive information making it to the far-end will determine what settings to start with or use to address the issue. The following parameter settings identify which parameters to adjust while the table provides the AUDIOENV settings (to mitigate ambient noise from being transmitted to the far-end).

Noise Heard at Far-end

AGC_Dyn_Range = whatever is currently set. Since AGC is a receive path function, it does not influence the far-end audio quality.

NR_Thresh_HS = Noisy office. This handset increases the amount of loss applied to background noise. If Noisy office does not fix the problem, try Very noisy office setting. If there is still noise, use the Excessive noise setting. If not a handset user, set this to typical office.

NR_Thresh_HD = Noisy office. This headset increases the amount of loss applied to background noise. If Noisy office does not fix the problem, try Very noisy office setting. If there is still noise, use the Excessive noise setting. If not a headset user, set this to Typical office.

HD_Tx_Gain = 0dB. 0dB is the default nominal setting. If the far-end can still detect speech or background noise coming from the headset user’s environment after making the above NR_Thresh_HD adjustments, set HD_Tx_Gain to -6dB. Note that HD_Tx_Gain only applies to headset – not handset.

AUDIOENV Variable Settings for Noisy Office

Customer’s current

AGC setting

+/- 9dB 9 3 5

+/- 12dB 36 30 32

noise reduction threshold setting

If Handset

user

1st choice 2nd choice (*)

If Headset user

COMPAS ID 120942

Page: 5 of 28

AUDIOENV Variable Settings for Noisy Office

Customer’s current

AGC setting

If Handset

user

1st choice 2nd choice (*)

If Headset user

+/- 15dB 63 57 59

+/- 18dB 90 84 86

(*) Note that these settings add loss to the transmitted speech and can affect the ability of the far-end party to hear the headset user

AUDIOENV Variable Settings for Very Noisy Office

Customer’s current

AGC setting

If Handset

user

1st choice 2nd choice (*)

+/- 9dB 18 6 8

+/- 12dB 45 33 35

+/- 15dB 72 60 62

+/- 18dB 99 87 89

(*) Note that these settings add loss to the transmitted speech and can affect the ability of the far-end party to hear the headset user

See following for Excessive noise AUDIOENV settings table.

If Headset user

Nearby Speech Heard at Far-end

AGC_Dyn_Range = whatever is currently set. Since AGC is a receive path function, it does not influence the far-end audio quality.

NR_Thresh_HS = Excessive noise. This handset applies the maximum amount of loss to background noise. If not a handset user, set this to typical office.

NR_Thresh_HD = Noisy office. This headset the maximum amount of loss to background noise If not a headset user, set this to

Typical office.

HD_Tx_Gain = 0dB. 0dB is the default nominal setting. If the far-end can still detect

speech or background noise coming from the headset user’s environment after making the above NR_Thresh_HD adjustments, set HD_Tx_Gain to -6dB. Note that HD_Tx_Gain only applies to headset – not handset.

AUDIOENV Variable Settings for Excessive Noise

Customer’s current

AGC setting

+/- 9dB 240 288 290

+/- 12dB 243 291 293

+/- 15dB 246 294 296

If Handset

user

noise reduction threshold setting

noise reduction threshold setting applies

If Headset user

1st choice 2nd choice (*)

COMPAS ID 120942

Page: 6 of 28

AUDIOENV Variable Settings for Excessive Noise

Customer’s current

AGC setting

If Handset

user

1st choice 2nd choice (*)

If Headset user

+/- 18dB 249 297 299

(*) Note that these settings add loss to the transmitted speech and can affect the ability of the far-end party to hear the headset user

4.2.2 Near-end Audio Quality Problems caused by Near-end Environmental Noise

Two common complaints of handset/headset users (typically headset users) who work in high noise environments are that they cannot hear/understand the far-end talker or that they hear the ambient noise in their headset (as sidetone).

User cannot hear/understand the far-end talker. This can be caused by a poor handset/headset ear seal that allows room noise to leak in and mask the far-end talker’s speech. Increasing the dynamic range of the receive AGC may help by boosting low level received speech from the far end party and allow the agent to better hear the far end party over the local environmental noise. See Appendix A for an AUDIOENV value that has a higher AGC dynamic range setting. Make sure to incorporate any AUDIOENV parameter changes already made (if any).

User hears ambient noise in handset/headset. This complaint of hearing the ambient noise in their handset/headset can be reduces/eliminated by adjusting the AUDIOSTHS or AUDIOSTHD setting to a lower sidetone setting. See Section 4.3.1.

Handset/headset users in environments with noise levels in the mid to high 60’s (dBA) can find some improvement by increasing the receiver volume, but if the receiver volume is already at the maximum, then acoustic treatment of the room is typically needed. There are no AUDIOENV settings which increase the receive volume levels range.

4.3 Call Centers

Room noise in a call center is usually loud12, and depending on the amount of isolation between call center agents, conversations from other agents may be picked up by the headset microphone of another agent and heard by the far-end. If a call center agent has been told by customers that they can clearly hear another agent’s conversation, then an AUDIOENV of 8 should prevent that from occurring. As indicated in section 4.2.1, only use setting 8 if setting 6 does not fix the issue because an AUDIOENV of 8 will lower the headset transmit gain and as a result, cause complaints from the far end party that they can’t hear the agent.

4.3.1 Background Noise in Agent/User Headset

If the agent is hearing other background chatter in their own headset receiver, then we recommend lowering their AUDIOSTHD setting by a step or two (see Table 1 in Appendix A). Lowering the sidetone along with using an AUDIOENV of 6 or 8 can be very effective in preventing the far end party from hearing any call center background noise/speech.

See Appendix C for measuring noise and noise level table.

COMPAS ID 120942

Page: 7 of 28

4.3.2 Agent to Agent Calls across Networks – with background noise issue

Under some circumstances, calls between agents across the customer’s network may have a different complaint. The complaint may be that the agent has trouble hearing the other agent even after increasing their receive volume to maximum. In this case, an AUDIOENV value of 35 will help boost this level by setting the AGC dynamic range of +/-12dB. If the agent still cannot hear the far-end, then an AUDIOENV value of 62 will set AGC dynamic range to +-15dB (which should correct the problem). Note that the AUDIOENV settings of 35 and 62 have the same noise reduction threshold and transmit gain settings. See Table 4 of Appendix A.

5 Other System Component Settings

5.1 DS1 Echo Canceller

DS1 echo canceller settings can affect incoming audio quality by producing echo canceller artifacts during echo cancellation. Because of this, it is best to identify DS1 echo canceller artifacts first and make the appropriate e-can setting change prior to making any AUDIOENV changes.

5.2 Network Region Codec

Voice quality can be affected by the type of speech coder utilized in the telephone network. A low-rate coder such as a G.729 codec reduces speech to a low transmission rate (8000 bits-persecond) but sacrifices voice quality in the process. In contrast, the G.711 codec has a higher transmission rate of 64,000 bits-per-second that uses more bandwidth, but allows it to better represent the talker’s voice (less impairments). For more information on distortions caused by low-rate coders, see Troubleshooting Audio Quality Problems in the Field

5.3 Comfort Noise Generator on Media Processor and Crossfire

When comfort noise is enabled either through the Media Processor or Crossfire boards, customers may complain of a hiss or noise generated when they stop talking. What the user is most likely hearing is the change in the idle-channel noise from the far-end as the echo canceller suppresses the echo and adds comfort noise to fill the “gap” left by the echo canceller. The comfort noise is “white noise”, and if the customer is complaining, then the noise coming from the far-end environment is not likely “white” in character. Therefore, the user is hearing that transition from white noise created by the e-can to some non-white noise generated at the far-end talker’s environment.

The best AUDIOENV setting (if comfort noise is used) is the default setting of zero that sets the AGC at +/- 9dB; thereby limiting the gain applied to low-level signals/noise. Typically, it is best to turn off the comfort noise generator in the MedPro/Crossfire board in order to eliminate this customer complaint.

5.4 Terminal Parameter Settings

Terminal Parameter gain settings are not downloaded to, or used by, the IP Telephones. A common misconception is that Primary TX Gain is used by the 46xx phone. Not true. A change in Primary TX Gain on page 2 of the Terminal Parameters form will only affect the gains used by the Media Processor when a DCP phone is on a call with an IP phone. This is necessary because the DCP phones and IP phones have different transmit/send levels and by modifying the gains in the

On the Tier IV website: http://support.avaya.com/japple/css/japple?PAGE=ProductArea&temp.productID=275624

COMPAS ID 120942

Page: 8 of 28

MedPro, an IP phone user will maintain the same talker level experience they have with an IP phone user as with a DCP phone user.

The Terminal Parameter settings on page 2 that do affect the IP Telephones are the following:

The Volume for IP Types setting can be set to either retain the receive volume between calls or leave the levels at default between calls.

5.5 Loss Plan

The loss plan in CM provides for the ability to change transmit and receive levels for any loss group (such as IP Telephones). It is not recommended to change or customize these levels under any circumstance in troubleshooting 46xx/96xx IP Telephone audio quality problems.

COMPAS ID 120942

Page: 9 of 28

+ 19 hidden pages