What do we hear through the Telephone?
G. J. Barnes, Chairman OFTEL WGHI
In recent years a number of new international standards relating to the coupling of telephone sets to hearing aids have been approved or are being finalised (CCITT P.37, ITU Draft Standard P.370, ETSI pr ETS 300 381, pr ETS 300 488, pr ETS 300 679). These standards cover inductive coupling, additional receiving amplification, which may also be associated with acoustic coupling, inductive coupling or used directly at the ear, and electrical coupling. With the recent move also to standardise the inductive input to hearing aids and relate this to the microphone sensitivity (revision planned to IEC 118-4), it becomes possible to estimate end to end transmission over a typical telephone connection for the three types of coupling.
The paper makes comparisons between the various types of coupling when used on a telephone connection and compares the end to end acoustic to acoustic path of each with a face to face one metre air path loss. Drawing on previous work carried out by Oftel WGHI (Working Group for the Hearing Impaired) typical end to end frequency characteristics are estimated and the extent to which these standards have met their objectives briefly considered.
For many years a number of standards relating to the measurement of the sensitivity of hearing aid inductive pickup coils have been in existence, e.g. IEC 118-1 1 . Additionally other standards relate to the inductive field to be provided by room loops and, more recently, by telephone earphones (IEC 118-4 2 , CCITT Rec. P.37 3 , ETSI pr ETS 300 381 #4 , EIA/TIA RS 504 5 ) in order to provide, by means of the inductive coupling feature, satisfactory communication to a hearing aid user. More recently standards have been developed covering additional receiving amplification and electrical coupling to hearing aids, e.g. Draft ITI-T Rec. P.370 6 , ETSI pr ITS 300 488 7 and pr ITS 300 679 8 . (A brief review of some of these standards is given in Annex 1).
Unfortunately, in the case of telephony, instances have been reported, albeit often anecdotally, of telecommunications users who are hard of hearing having difficulty using the telephone, even when using the inductive coupling feature. A number of reasons for poor communications can be suggested, but until recently it has not been possible to be objective and to suggest corrective action with any confidence. Part of this is due to the lack of an agreed international standard for the input sensitivity of a hearing aid inductive pickup coil, although this is now receiving attention in IEC. Another reason may be due to insufficient cross fertilisation of ideas and expertise between the hearing aid and telephone industries.
In a paper 9 last year the author assessed the likely effect of recently agreed standards for inductive coupling on hearing impaired telephone subscribers, using an assumed standard for the sensitivity of a hearing aid inductive pickup input. This was based on the standard in use in the UK 16, and uses the same sensitivity values. In this, the inductive input sensitivity is linked to the sound level occurring at a hearing aid microphone over a one metre air path, thus avoiding the need to be specific about particular volume control setting on the hearing aid, and leads to the possibility of calculating end to end levels over a typical telephone connection.
In the above paper 9 it was assumed that the loss of speech signals between the two ends of a telephone connection having characteristics similar to an IRS (Intermediate Reference System -ITU Rec. P.48 11 ), is approximately equal to the loss at 1 kHz. In this paper a more detailed assessment of the overall frequency characteristic of the telephone transmission path with a hearing aid coupled inductively, acoustically and electrically is compared with a hearing aid acoustic input over a one metre air path.
One of the objectives of developing standards for the coupling of telephones to hearing aids is to ensure that as far as possible conversations over a typical telephone connection result in sound levels at the ear of a hearing aid user that are similar to those experienced in face to face conversation using the hearing aid microphone. This paper looks briefly at how well this objective is met.
Telephone loudness ratings were established by the ITU in the 1980s as the recognised means for controlling the end-to-end losses of telephone connections. The calculation method is described in ITU-T Rec. P.79 12 and the actual values recommended given in Rec. G.111 10 and others.
Basically the end-to-end acoustic path of the IRS-to-IRS connection is taken as the standard and comparisons of the connection to be tested made against this. The loudness rating algorithm 12 assumes a speech spectrum (ITU-T Rec. P.50 15) at the MRP (Mouth Reference Point) of the sending end and using the sensitivities of the IRS, a computation is made of the loudness of the signal arriving at the ERP (Ear Reference Point), taking into account the nature of speech, the characteristics of the human ear, its loudness growth function and etc.
An unknown telephone set receiving end can be “rated” against the IRS by substituting it in place of the IRS receiving end and comparing the resulting end-to-end loss against the IRS-to-IRS connection. The Receive Loudness Rating (RLR) is given by the attenuation that needs to be introduced into the IRS-to-IRS connection that makes its loudness equal to speech path of the connection that includes the unknown part. Rating an unknown send end is tackled the same way. Thus loudness ratings are equivalent to loss, and negative values of RLR mean that the unknown is louder than the IRS. As well as SLR and RLR, an Overall Loudness Rating (OLR) can be deduced and is used to assess the end-to-end loss of a complete connection. Thus a complete connection equivalent in loudness to the IRS-to-IRS connection with 10 dB introduced between the two ends would have an OLR of 10 dB. In fact any acoustic path that starts at the MRP and finishes at the ERP may have its OLR calculated and this has been done below for a 1 metre air path.
Similarly the combination of a hearing aid coupled to a telephone set may be treated as a telephone receiving end and may be rated as a RLR. This gives a very useful way to quantify the different methods of coupling and variations within those methods and is used in this paper.
The general equation for calculating loudness ratings is;
Where Sn is the sensitivity of the path to be rated at the 1/3rd octave frequencies, W n is the appropriate weighting to be applied at each frequency for the particular rating (RLR, SLR, OLR) N is the number of frequencies, usually 14 over the range 200 to 4000 Hz.
Speech path mouth to ear losses for a typical (optimum) telephone connection and a metre air path Telephony
The Figure 1 shows the end-to-end loss of a telephone connection meeting the ITU long term aim of 10 dB OLR given in ITU-T Rec. G.111. This is based on the use of telephones at the send and receive ends having the same sensitivity frequency characteristic as the IRS. At 1000 Hz the IRS send sensitivity is -3.7 dBV/Pa and the receive sensitivity is 12.6 dBPa/V.
|Sound level @ MRP||Loss||Sound level @ ERP|
|89.3 dBSPL (pure tone)||1.1 dB||88.2 dBSPL (1 kHz)|
|88.8 dBSPL (Speech)||5.5 dB||83.3 dBSPL (0.2 to 4 kHz)|
Metre air path
|89.3 dBSPL (Speech)
(100 - 8000 Hz)
|30.2 dB||59.1 dB SPL (Free field)|
|29.0 dB||63.3 dB SPL
(including room & head effects)
|88.8 dBSPL (Speech)
(200 - 4000 Hz)
|28.9 dB||62.9 dBSPL
(including room & head effects)
Figure 1 Speech path mouth to ear losses for a typical telephone connection, having an OLR of 10 dB, and a metre air path.
If the IRS send and receive ends were connected directly together, (giving by definition an OLR of 0 dB), at 1 kHz this connection would have an overall gain of
12.6 -3.7 = 8.9dB
Calculations over the telephone bandwidth (0.2 to 4 kHz) have suggested that for speech the overall acoustic gain is about 4.5 dB, i.e. 3.4 dB less. Thus given a typical overall loudness rating path of 10 dB, equivalent to connecting a 10 dB loss pad between the IRS send and receive parts, the end to end acoustic path would have a gain of
8.9 - 10 = - 1.1 dB @ 1000 Hz
8.9 - 10 -3.4 = - 4.5 dB for speech
The mouth to ear loss of a one metre air path, under free field conditions in a straight line horizontally from the MRP, is usually taken as 30.2 dB. [This can be deduced from the assumption for telephony purposes that the virtual source of signals from the mouth originates from a point 6 mm behind the lip plane and that the MRP is 25 mm in front of the lip plane. The inverse square law is invoked in the calculation].
Given the long term speech level at the mouth reference point this results in a figure of
89.3 - 30.2 = 59.1 dBSPL at a position 1 metre away.
If the obstacle effect of the head is taken into account a further 1.8 dB (1 kHz) may be added. In reverberant conditions, assuming a reverberation radius of 1 metre another 3 dB may be added giving a final figure of 63.9 dBSPL. Calculations using speech over the full and telephone bandwidths (0.1 to 8 kHz & 0.2 to 4 kHz resp.) including the obstacle effect of the head and an allowance of 3 dB for room reverberation give figures as follows;
89.3 dBSPL @ MRP gives 63.3 dBSPL @ ERP 1 metre distant (0.1 - 8 kHz)
88.8 dBSPL @ MRP gives 62.9 dBSPL @ ERP 1 metre distant (0.2 - 4 kHz)
(ERP, Ear Reference Point, is defined as a point located at the entrance to the ear canal of the listener’s ear).
These figures are not too different from the value of 65 dB often assumed in hearing aid design circles if one takes into account a probable raised voice level compared with telephony speech.
Another standard published by the ITU in recent years is Rec. P.58 covering a Head And Torso Simulator (HATS)13 . This covers a number of useful characteristics that are applicable to the acoustics of the average human head, including the obstacle effect of the head, the eardrum to ear reference point correction and acoustic impedance of the ear coupler, in this case the simulator specified in IEC 711 17 . Using the acoustic path loss from MRP to DRP (ear Drum Reference Point) under free field conditions, the obstacle effect of the head, the correction DRP to ERP, it is possible to calculate the acoustic path loss MRP to ERP. If one does this the effective OLR may be determined and results in;
OLR = 31.6dB (0.2 - 4 kHz)
This may be compared with the ITU long term aim for all telephony connections to be 10 dB OLR given in ITU-T Rec. G.111. In other words, a typical telephone connection will in the long term be at least 20 dB louder, (on a loudness rating basis), than a 1 metre air path. Currently many local calls will be louder than this and some long distance calls quieter.
In what follows a number of measurements and calculations have been made on telephones alone, also coupled to hearing aids. The measurements were carried out using speech weighted noise and 1/3rd octave analysis. The telephone used, unless otherwise stated, is the British Telecom Tribune set equipped with an inductive coupling coil. The hearing aids were set generally to Reference Test Gain (RTG) or at maximum. The test results below, drawn largely from the report 18 of work carried out by/for WGHI, are given by way of example only to show orders of magnitude - further study is clearly required to obtain firm figures for particular aids and coupling conditions.
Most telephone earphones are designed to be loaded in a sealed ear, consequently when loosely coupled to an ear the low frequencies are usually lost. Figure 2 illustrates this. When coupled to a hearing aid the acoustic coupling, for BTE aids at least, is far from being sealedand the deficiencies in the low frequencies is carried through to the response at the ear of the listener. Furthermore the way the handset is held relative to the hearing aid microphone is important. Figure 3 shows, for a BTE aid, the effect of positioning the handset over the ear, as if not wearing an aid, and over the microphone of the aid itself. This difference can be translated to a difference in loudness rating of about 7 dB.
For an in the ear (ITE) aid, see Figure 4, the situation is somewhat better as the natural inclination is to position the handset over the ear, resulting in a better frequency characteristic and a louder signal, as a result of a better approximation to sealed conditions. There is also a reduction in the room noise at the ear due to the shielding effect, although some noise still reaches the ear via the telephone sidetone path. With this aid (Classic) connected to a typical connection the OLR expected would be about -16 dB, i.e. about 47 dB louder than a 1 metre air path.
There is a class of telephone earphones that do not suffer so much when operating under poor seal conditions, i.e. those that have a low not possible to go into any great detail on this subject but an example such a telephone compared with a more typical type under the same The geometry and characteristics of the panna of the KEMAR ear if not impossible to achieve, and Figure 5 shows the responses of (low acoustic impedance) telephones. Note the good low frequency set. The RLR of the Beocom (0.5 dB) is 4.5 dB louder than the Tribune (5.0dB).
As might be expected when the Beocom is coupled acoustically to frequencies are stronger, as the coupling is more efficient. Figure coupling is to a BE51 aid in maximum volume setting, and borne the low acoustic impedance telephone resulting in coupling that is dB RLR).
On a typical telephone connection these examples would be of the order of 46 and 50 dB louder respectively than a 1 metre air path.
In order to meet the requirements of the prETS 300 488 and ITU-T Rec. P.370, telephone sets will need to provide additional receive amplification between 10 and 20 dB. A telephony connection involving a set meeting these standards would therefore typically provide an OLR of 0 to -10 dB, i.e. 30 to 40 dB louder than a 1 metre air path and having an acoustic gain that is equivalent to some hearing aids. If used with an aid however it is quite possible that some overloading could occur.
Inductive coupling offers a number of advantages over acoustic coupling, in particular an effective reduction of room noise and an improvement in the frequency characteristic. To get the best coupling however some experimentation on the part of the user is generally required. This can be explained by the fact that the magnetic field from the telephone earphone is usually provided axially, resulting in a field that is strongest horizontally, whereas the hearing aid pickup coil is usually orientated vertically in the interests of coupling to room loops. For in the ear aids difficulties can arise and there is probably a case for arranging the pick-up coil in the aid to be at 45 degrees to the vertical, to meet both requirements.
Figure 7 shows one example of the sound pressures that can be expected at the ear from a telephone inductively coupled to a BE17 aid over a typical connection, and resulting in an OLR of the order of -16 dB. Note that the noise level, which arrives at the ear via the telephone sidetone path, is significantly lower than that for acoustic coupling into the same aid (Figure 3) although not so different from that for the Classic in the ear aid (Figure 4) when coupled acoustically. Figure 8 illustrates the same telephone connection coupled inductively to the Classic aid. In this case the OLR calculates out to 14.9 dB, i.e. about 5 dB quieter than placing the telephone on an unaided ear.
Figure 9 shows the sort of coupling that can be achieved inductively with a medium gain aid such as the BE31, in this case resulting in an OLR of the order of -27 dB and about 59 dB louder than a 1 metre air path. When inductive coupling is used in association with additional receiving amplification clearly a hard of hearing telephone subscriber would experience considerable assistance, providing that the inductive input circuit can handle the large signals.
Although the to make any measurements on hearing aids with very few in existence. Neither has the author had sight of any standards describing frequency characteristics for such an aid, apart for what is given in IEC 118-6, which has some rather wide tolerances.
However, it is possible to make a preliminary estimate using the sensitivity of a typical telephone set, the relevant standards and assuming that the frequency response of the aid approximates that for flat acoustic input. On a typical telephony connection, and assuming that an electrical input to the aid of -35 dBV is equivalent to 70 dB SPL at its microphone, it is estimated that for a BE34 aid, the OLR would be -34 dB, more than 20 dB louder than the same aid would provide over a 1 metre air path. These figures will need to be confirmed. Fortunately the ETSI standard prETS 300 679, provides for a volume control to reduce the output voltage. However the IEC standard 118-6 may need some amendment to narrow the tolerance on input voltage.
A brief overview of the possibilities for hard of hearing subscribers when using the telephone, has been given. Generally speaking, the current standards for inductive coupling and additional receive amplification, should, if implemented, enable all but the most profoundly deaf users to communicate over the telecommunication networks. The Table 1 below summarises most of the results given in this paper. From Table 1 it is evident that all three types of coupling can be expected to provide useful assistance to the majority of hard of hearing telephone users, although one type may be better than another for particular individuals.
As to whether particular users are able to change from communication over a 1 metre air path to a telephony connection without changing their volume control setting will depend to a large extent on the type of aid worn. It will also depend on their preferences concerning frequency characteristic and any gain changes required to compensate for the loss of low frequencies, particularly when coupling acoustically. The results do however suggest that for the BE51 aid and using inductive coupling there is certainly room for adjustment of the coupling position in order to achieve a level that is equivalent to the 1 metre airpath using the aid acoustically. Both acoustic and inductive coupling methods provide for a range of OLR that is wider than the OLR of the 1 metre acoustic path, with or without an aid, with the exception of the ITE aid coupled inductively. With the new electrical coupling standard it should be possible to extend the coupling of the telephone network to a wider range of people, even perhaps those with cochlea implants. This is a preliminary comparison of the various means of coupling hearing aids to telephone networks - it is hoped to build on it further and provide more accurate data at a later date.
|At MRP (0.2 to 4 kHz)||88.8 dB SPL||incl. head effect, no room effect|
|At ERP (0.2 to 4 kHz)||59.9 dB SPL||(as above)|
|At ERP (0.2 to 4 kHz), typical telephone connection||83.3 dB SPL|
|At DRP (0.2 to 4 kHz)||62.2 dB SPL||incl. head effect, no room effect|
|At DRP (0.2 to 4 kHz), BE 34 aid set at RTG||113 dB SPL||(as above)|
|At DRP (0.2 to 4 kHz), BE 51 aid set at RTG||123 dB SPL||(as above)|
|Loudness Rating of the acoustic path (OLR)|
|1 metre air path MRP to ERP||31.6 dB|
|Typical telephone connection||10 dB||ITU-T long aim|
|1 metre air path plus BE34 aid set at RTG||-13 dB|
|1 metre air path plus BE51 aid set at RTG||-26 dB|
|Acoustic coupling (OLR)|
|Typical connection , Tribune set, BTE aid (BE17)||-13 dB||Best coupling|
|Typical connection , Tribune set, ITE aid (Classic)||-16 dB||Best coupling|
|Typical connection , Tribune set, BTE aid (BE51) at Max.||-25 dB||Best coupling|
|Typical connection , Beocom set, BTE aid (BE51) at Max.||-29 dB||Best coupling|
|Inductive coupling (OLR)|
|Typical connection , Tribune set, ITE aid (Classic)||14.9 dB||Best coupling|
|Typical connection, Tribune set, BTE aid (BE17)||-16 dB||Best coupling|
|Typical connection , Tribune set, BTE aid (BE31) at RTG||-27 dB||Best coupling|
|Typical connection , Tribune set, BTE aid (BE51) at RTG||-39 dB||Estimated|
|Electrical coupling (OLR)|
|Typical connection , Tribune set, BTE aid (BE34).||-34 dB||Estimated|
EIA RS-504 (1983) Addendum No1 (1994) extends to digital sets)
For a given input signal voltage (-10 dBV) at a local telephone exchange this standard sets the acceptable limits of both axial (>-22 dB rel. to 1 A/m) and radial (>-27 dB rel. to 1 A/m) magnetic fields emanating from a telephone earphone in order for the set to be able to claim "hearing aid compatibility". Measurements are made in a plane 10 mm away from the plane of the earcap and there are somewhat involved rules relating to the frequency response requirements. Recently there have been suggestions towards a change in this standard to make it similar to the ITU and ETSI standards by relating the magnetic field strength to the sound level at the earphone. This improves international consistency and avoids the need for digital telephones to be treated differently.
ETSI pr ETS 300 381 (1994)
Inductive field requirements are specified for an acoustic output of 80 dB into an artificial ear. The field is measured in a plane 10 mm away from the earcap plane and the maximum field strength, which can have any orientation, is 'Acceptable' if it lies between -30 and -17 dB rel. to 1 A/m at 1000 Hz. A 'Preferred' range of -17 to -25 dB rel 1 A/m is suggested as likely to be required for good coupling with hearing aids designed to couple to induction room loop systems. Frequency characteristics are also specified. An Annex points out that the radial field is the most important.
ETSI pr ETS 300 488 (1995)
The addition receive amplification, measured as RLR, is specified as 15 ± 5dB louder than the relevant terminal standard for the network into which it is to be installed. The standard allows for a volume control which must not be possible to set more than 15 dB quieter than the relevant terminal standard. Test methods are included for both analogue and digital telephones.
ETSI pr ETS 300 679 (1996) Draft
This standard sets the electrical output from a telephone set for received signals as -35 ± 5 dBV when the sound level from the earphone, when loaded in an artificial ear, is 80 dB SPL. It also specifies the coupling method (2.5 or 3.5 mm coaxial miniature audio jack), test methods. Attention is drawn to the necessary safety issues but these are not specified in this document but referred out to the EC Low Voltage Directive.
IEC 118-1 (1975)
Concerned entirely with measurement method for setting up and calibration of a known inductive field into which the hearing aid is placed for sensitivity and frequency characteristic measurements. The sensitivity is expressed as the output sound pressure into an artificial ear coupler for magnetic field strength of 1mA/m.
IEC 118-4 (1981)
Specifies the measurement methods and levels (100 mA/m) of the inductive field associated with room loop systems. (Expected to be amended to link the required inductive input sensitivity to the microphone input).
ITU-T Recommendation P.37 (1993)
Recommends measurement methods and field strength requirements for an acoustic output of 80 dB into an artificial ear. The field is measured in a plane 10 mm away from the earcap plane and the maximum field strength, which can have any orientation must lie in the range -30 to -17 dB rel to 1 A/m at 1000 Hz. The frequency characteristics are also recommended.
ITU-T Recommendation P.370 (1996) Draft
This is an update of the Rec. P.37 and now includes inductive coupling, additional receive amplification and electrical coupling. The 3 parts of the Recommendation are the same, technically, as the 3 ETSI standards ETS 300 381, 300 488 and 300 679. Approval of the current draft is expected in May 1996.
UK Dept of Health Specification for hearing aid type BE 50 Series.
Specifies the inductive coupling input sensitivity for hearing aids for issue in the UK National Health Service. The inductive coupling input sensitivity (10 mA/m) is specified at a known air to air gain setting (40 dB) and a known output sound level (95 dB). In this way it is possible to link the microphone and inductive inputs irrespective of the volume control setting and thus give equivalence with 1 metre air path conversational levels.