Compression Made Easier

Document Sample
Compression Made Easier Powered By Docstoc
					Compression Made Easier

       Terry S. Griffing
  Audina Hearing Instruments
 The only thing certain about compression is
 that it is somewhat difficult to explain in a
               simple sentence.

This is as it should be. If it were easy, I would
     not have a topic to discuss. I believe
compression circuitry, which is ever changing,
    is difficult to always fully understand.
 Our goal today will be an attempt to review
and understand the basics exceptionally well.

This achievement will serve as our foundation
  for understanding, and secondly, to better
     evaluate new, perhaps yet to appear
compression hearing aid circuitry technology.
  We need to re-orient our thinking about
compression. It is not today primarily used to
keep sounds from being too loud delivered to
  the ear through the hearing instrument.

 In today’s technology compression is more
       about selecting amplification.
   To make sure the life and speech sounds
function at a comfortable listening level with
  the boundaries being the upper and lower
 levels, which define an individual’s dynamic
             Compression Basics
   Input/Output Graphs           Input Compression

   Input/Output Compression      Attack Time

   Threshold Kneepoint Tk        Release Time

   Compression Control           Time Constants

   Clinical Applicants           Interaction Factors

   Output Compression
     Understanding Graphs

Output                                        80 SPL
                                            - 50 SPL
                                           30dB Gain


         Must be labeled as to function.

Graphs are used to illustrate function.
Input Versus Output Compression

 First basic difference is where the volume
              controls function.

   If it is output compression the volume
 controls function before sound goes to the
amplifier. In other words think of it as a door
 opening…how wide is it open controls how
     much sound goes into the amplifier.
Input Versus Output Compression

If it’s input compression the volume control is
          located just before the receiver.

In other words this door controls how much
sound is permitted to exit the hearing aid via
                the receiver.
Input Versus Output Compression

The volume control location creates dramatic
  differences in how hearing aids function.

Also, realize if there is a volume control, it is
    managed by the hearing aid wearer.
 Let’s look at two graphs that
 illustrate the effect of volume
control changes, output versus
Where compression starts.
       Output Compression
The volume control affects gain and the kneepoint but not the sspl.
          Input Compression
The volume control affects gain and sspl but not the kneepoint.
   The level-detecting device is located before the volume
    control. Should an input signal excel the threshold
    kneepoint, the gain at the pre-amplifier is reduced.

  The input signal alone determines if the gain is reduced.

The detecting device is located after the volume control. The
  kneepoint changes are dependent upon volume control

            Let’s look at our previous graphs.
    The fundamental characteristic of all
    compression circuits is threshold of
 compression referred to as the kneepoint.
The kneepoint is the level for incoming sound
  to the instrument that is loud enough to
      trigger the action of compression.

   It is seen by looking at the input/output
 function. In this case we are not examining
  input or output compression, merely the
  We have a kneepoint of 77 dB sspl, when
   input exceeds this level, it compresses,
 reducing the gain. In the next illustration
the kneepoint, where compression starts is
                60 dB sspl.
        Compression Levels (Tk)
   Sometimes hearing aids are described as
    having a low or high compression kneepoint.

   A low compression Tk would be 45-65 dB

   A high compression kneepoint would be 65
    dB or greater.
        Compression Ratio
Simply stated, the ratio describes how much
of the incoming signal beyond the kneepoint
         will be changed or reduced.
                For Example

  The compression ratio is 2:1. For each 10 dB SSPL
input increase, we read a 5 dB increase; hence 10:5 is
2:1. For energy 2 dB increase of input there is a 1 dB
     SSPL increase output, past the kneepoint.
              For Example

         A linear circuit functions as 1:1.
 A low compression ratio might be less than 3:1.
A high compression would be higher, such as 5:1.
       Attack and Release Time

                    Attack Time

  Should be as short as possible to prevent sudden
 loud sounds from being amplified and reaching the

 The human ear cannot perceive temporal changes of
less than 5 or 10 milliseconds. It is important that the
 attack time be very short of less than 5 milliseconds.
        Attack and Release Time

                       Release Time

It is the time lag or delay for the system to restore itself to the
                      normal gain function.

Generally speaking, the system stabilizes to within 2 db of its
 steady state value. Release times of 45 to 150 milliseconds so
 the user is not distracted by rapid or abrupt changes in gain.
Too short 50 mo or less creates a pumping noise and can affect
                     intelligently for speech.
     Some Compression Distinctions
1.    It is useful to look at compression limiting as acting to reduce gain
      above some level to reduce sounds that are too loud.

2.    WDRC acts to increase gain below some level for sounds that are
      too weak.

3.    WDRC acts to provide gain for weak sounds that the ear cannot
      provide for itself with impaired or reduced outer hair cell function.

4.    One valid criticism on output limiting is that it almost waits too
      long to do it’s job.

5.    A 40 db loss typically shows a 2:1 increase in slope over the normal
      loudness-growth curve. In other words, a compression amplifier
      with that ration will restore the normal loudness. A 60 db loss needs
      a 3:1 ration for loudness restoration.
      Types Of Compression

 Evaluating Compression Variables:

A Field Study of Patients Fitted With
          DSP Hearing Aids
   Gennum Paragon DSP Circuit

   Evaluate patient listening comfort and clarity
    utilizing various processing strategies and various
    compression dynamics.
   Processing strategies
      WDRC
      AGC-o Compression Limiting
      Linear

   Compression Dynamics/Attack and Release Times
      Single Slow
      Twin Average Detection
      Syllabic
      Phonemic
                 WDRC Rationale
   WDRC – Wide Dynamic Range Compression

   By definition CR<5:1, TK<65

   Strengths
       Widest Application.
       Attempts to mimic the loudness functions of the ear.

   Weakness
       Previous Linear user may reject due to lack of “loudness”.
    AGC-o Compression Limiting Rationale

   By definition CR >5:1, TK>65

   Strengths
        Provides Comfort in Noise
        Reduces Gain for high inputs to provide comfort in
         noisy/loud situations.
        Increased linearity

   Weakness
        Does not effectively “reach down” to amplify the soft sounds
         or restore the normal loudness growth Function.
                Case Study I

   NR Response BC        Used Super Power BTEs

   Previous User 30yrs   Rx: Linear
                 Case Study II

   AC=BC
                          Likes Power – Some reported
   New User               problems with loud sounds -
   Rx: AGC-o/Linear       Reduced Dynamic Range
               Case Study III

   AC=BC
                      Works in library - believes HL
   New User           due to noise (shooting high
   Rx: WDRC           power rifles)
 Now That We Have Covered
Stage 1, Fitting Strategies, Let’s
Take a Look at Time Constants.
         Twin Average Detection
   Utilizes two level detectors (one fast and one slow)
    to control both fast sudden loud noises (i.e. door
    slamming) and steady, consistent noise (i.e. crowd).
   Twin average detector employs one fast and one
    slow average detector in the compressor circuit.
    Resulting in a smoothed response.
   In the average detection compressor the two sets of
    attack and release times vary with the incoming
    loud signals.
            Syllabic Compression
   Fixed Short attack and release times.

   The time constants are specifically designed to be
    shorter than the typical speech syllable which is
    about 200msec.
          Phonemic Compression
   Fixed - Even shorter attack and release times than
    syllabic compression.

   Time Constants less than 100 milliseconds.

   Premise is to allow the HA to make the softer
    sounds of speech (consonants) more audible
    without simultaneously making the normally louder
    part of speech (vowels) from becoming too loud.
         Single Slow Compression
   Fixed slow attack and release times to avoid rapid
    fluctuations in gain.

   Attack and Release times >275 milliseconds.
Now That We Have Discussed
the Various Strategies We Will
 Move on to the Study Set up.
               Subject Selection
   Six subjects collected by word-of-mouth.

   Combination of new and previous HA users.

   Sensori-neural hearing loss – Mixed hearing losses
    and conductive losses were excluded.
            Stage One of the Study
   3 user adjusted memory set up

       WDRC
       AGC-o Compression Limiting
       Linear

   Subject given a two page questionnaire to
    evaluate/compare the 3 different memories for
    comfort and clarity in various situations.

   NO external Volume Control.
Wide Dynamic Range Compression

                     WDRC
                     Compression Ratio <5.1
                     Lower Tk <65

                  Study Set Up:
                   Low Freq > 1.5:1 CR

                   High Freq > 2.1:1 CR

                   Lower TK = 50 dB
AGC-o/ Compression Limiting

                    AGC-o
                    Compression Ratio > 5:1
                    TK > 65

                 Study Set Up:
                  Upper TK: >90

                  Overall AGC-o Control

                    to function as
                    Compression Limiting.
               Subject Selection
   Six subjects collected by word-of-mouth.

   Combination of new and previous HA users.

   Sensori-neural hearing loss – Mixed hearing losses
    and conductive losses were excluded.
            Stage Two of the Study
   4 user adjusted memory set up
   Phonemic
        Twin Average Detection
        Syllabic
        Single Slow
   Subject given a two page questionnaire to
    evaluate/compare the 4 different memories for
    comfort and clarity in various situations.
   NO external Volume Control.
Stage 2: WDRC with Time
               Memories: Same
               settings except for
               change in Time
               We studied Twin
               Average, Phonemic,
               Syllabic, and Single
   Compare memories to each other.
   Good, better, and best scale for both comfort and
   Choose best overall memory for that particular
   Questionnaire was filled out by subject and discussed
    upon submission.
          Questionnaire Situations

One on One     Telephone       Driving-        Television
Conversation                   conversation
Little or no                   while in car

Movie          Small group     Place of         Conversation
Theater        conversation-   Worship –        with
               multiple        speaker up front background
               speakers                         noise
The Subjects:
Subject 1 (C.E.N.)

                        77 Year Old
                        Male
                        Physician
                        AC=BC
                        New User
         Subject 1 (C.E.N.)

Overall this subject preferred WDRC with no
significant listening difference across the four
           different time constants.
Subject 2 (S.M.)

                      45 Year Old
                      Male
                      Postal Service
                      AC=BC
                      Previous User
         Subject 2 (S.M.)

Overall this subject preferred WDRC yet had
         some preferences for AGC-o.

  Linear was quickly rejected as too loud.

 Subject preferred Twin Average Detection
Subject 3 (D.S.)

                      73 Year Old
                      Female
                      AC=BC
                      Previous User
          Subject 3 (D.S.)

Overall this subject preferred WDRC, stating
AGC-o and linear settings were “too loud for

The subject selected the Single Slow mode as
             the best for clarity.
Subject 4 (R.E.S.)

                        71 Year Old
                        Female
                        AC=BC
                        Previous
                         Linear User
                         (5 Years)
         Subject 4 (R.E.S.)

   Overall this subject preferred WDRC.

The subject selected Twin Average Detection
  mode as the best for comfort and clarity.

Subject stated Phonemic compression memory
  created an undesirable “crackling sound”.
Subject 5 (R.S.)

                      49 Year Old
                      Male
                      Financial
                      AC=BC
                      Previous User
          Subject 5 (R.S.)

Subject clearly preferred AGC-o compression
 limiting except in noisy situations. Subject
              preferred WDRC.

Subject selected Phonemic Compression mode
      as the best for comfort and clarity.
Subject 6 (C.R.B.)

                        59 Year Old
                        Male
                        Retired
                        AC=BC
                        New User
       Subject 6 (C.R.B.)

       Overall this subject strongly
           preferred WDRC.

Subject selected the Phonemic compression
 mode as the best for comfort and clarity.
Overall Results
Stage 1: One On One Conversation
   (Little or No Background Noise)
    Compression Strategy: One-on-One Conversation

           AGC-o           0%


                   WDRC   AGC-o    Linear
Stage 1: Conversation In Car

    Compression Strategy: Conversation in Car


        AGC-o                            WDRC
         50%                              50%

                WDRC   AGC-o    Linear
Stage 1: Home Television
      Compression Strategy: Home TV


  Could Not
    50%                              AGC-o
                      Linear          33%

    WDRC      AGC-o   Linear   Could Not Decide
Stage 1: Place of Worship
  Compression Strategy: Place of Worship


     Linear                              50%


              WDRC   AGC-o   Linear   N/A
Stage 1: Telephone
Compression Strategy: Telephone Use
   Could Not

   AGC-o                            50%

 WDRC      AGC-o   Linear   Could Not Decide
Stage 1: Small Group
 Compression Strategy: Small Group

  Linear 17%                    WDRC
   0%                            33%


      WDRC     AGC-o   Linear   N/A
Stage 1: Conversation in Noise
   Compression Strategy: Conversation in Noise

            0%      N/A


                   WDRC   AGC-o   Linear   N/A
Stage 1: Overall Preferred Setting

      Stage 1: Preferred Compression Strategy

         AGC-o           0%


                 WDRC    AGC-o   Linear
Stage 1: Quotes From Our Subjects

   C.E.B.: “ It’s very comfortable, not overbearing with
              background noise.”
               Referring to memory A – WDRC.

   R.S.: “It didn’t amplify the background so much.”
               Referring to memory A –WDRC.

   D.S.: “B & C are too loud for comfort.”
               Referring to AGC-o and linear setting.
Stage 2: One on One Conversation
         Time Constants: One-on-One Conversation
                   Could Not

              Single Slow
                 33%                         Twin Average

   Phonemic    Twin Average    Single Slow   Syllabic   Could Not Decide
Stage 2: Small Group Conversation
           Time Constants: Small Group Conversation

                    Could Not
                     Decide                     Phonemic
                      16%                         17%

                                                    Twin Average
                    Single Slow                         33%

    Phonemic    Twin Average      Single Slow   Syllabic   Could Not Decide
Stage 2: Place of Worship
       Time Constants: Place of Worship

     Phonemic 17%                  17%

   Single Slow                            Twin Average
      33%                                     33%

  Phonemic   Twin Average   Single Slow     Phonemic
Stage 2: Conversation in Car
           Time Constants: Conversation in the Car
                Could Not
                  16%                      Phonemic
                                                 Twin Average
               Single Slow

Phonemic    Twin Average     Single Slow   Syllabic   Could Not Decide
           Stage 2: Telephone
               Time Constants: Telephone

           Could Not
            Decide                          Phonemic
             33%                              34%

            Sylabbic                       Twin Average
              0%          Single Slow           0%

Phonemic   Twin Average   Single Slow   Sylabbic   Could Not Decide
Stage 2: Conversation in Noise
            Time Constants: Conversation in Noise
                 Could Not
             Syllabic                            33%

            Single Slow
               33%                         Twin Average

 Phonemic    Twin Average    Single Slow   Syllabic   Could Not Decide
Stage 2: Overall Time Constants
      Time Constants: Preferred Time Constants

                          Sylabbic   Phonemic
                            0%         17%

     Single Slow
        50%                                Twin Average

      Phonemic     Twin Average   Single Slow   Sylabbic
Stage 2: Quotes From Our Subjects

   R.S.: “Little reason to change channels.”

   R.S.: “Memory A would “crackle.”
           Referring to phonemic memory.

   C.N.: “Could not discern much difference between
   Preference for WDRC and/or AGC-o over linear
    processing strategies.

   No significant overall preference for the various
    time constants.

   Younger (<60 yrs old) subjects preferred multi-
    memory options compared to the older subjects.
                   In Conclusion
  Processing strategies and time constants are key
     elements of maximizing patient satisfaction.

  The flexibility of these options gives the hearing
  healthcare professional the tools to optimize every


Audina Hearing Instruments
(800) 223-7700
               Circuit Selection
Initially the dispenser goes through a brief two-step process:

                  1. Style of Instruments
                 2. Linear or Compression
              Circuit Selection
  Remember new users will usually select a style heavily
           influenced by cosmetic appearances.

A repeat replacement buyer usually places less emphasis on
    cosmetic appearance, showing much greater interest in

It is important for replacement purchasers to be evaluated
   with speech in noise. Remember the primary complaint of
       hearing impaired persons is difficulty in background
               The Quicksin Test

1.   Measures speech in noise performance.

2.   Takes only a minute or two.

3.   Can demonstrate if extended high frequency emphasis
     improves or degrades understanding of speech in noise.

4.   Demonstrate relative benefit of Frontwave dual
     microphones to improve speech intelligibility in noise.

5.   Provides valuable & useful information in counseling.
         Other Selection Factors
Severity of Loss
   a)   Mild or early to moderate hearing losses can be fitted with
        compression technology, especially WDRC.

   b)   Moderately severe losses usually require more gain than
        compression (WDRC) provides. Either linear or output
        compression may be preferred.
             Configuration Of Loss
   Flat – Circuitry that can be programmed to provide till/bill

   Sloping – Compensate for reduced dynamic range.

   Abrupt High Frequency Losses – Probably best first choice
    is AGC-o.
        Previous Hearing Aid Users
   Previous users fall into two groups: those reasonably
    satisfied and those who are not. In the second case, the
    dispenser must attempt to determine the root cause and
    attempt to correct with a new fitting.

   Clients who did not wear their aids should be considered
    new users WDRC digital usually does well.

   Previous semi satisfied users would accept new hearing aids
    if they perform similar to previous aids.
    Importance of MCL & UCL Measures

   The key for getting good measurements are directly related
    to the adequacy of the examiners instructions to the client.

   The second key is to be fully aware of dynamic range for the
    “normal” ear versus the impaired ear.

   Third to assume almost every sloping sensori-neural loss has
    a reduced dynamic range that is frequency specific for the
    higher frequencies; 2000, 3000, 4000 Hz.

Our goals are as follows when fitting hearing
          instruments in rank order:
1.   Need to demonstrate improved sound/speech audibility

2.   Have the patients cosmetic and psychological concerns
     been satisfied?

3.   Are the hearing instruments both physically and
     acoustically comfortable to the ears?

4.   Do the patients like the sound quality of the amplified
     sound in different listening environments?